Authority, Cooperation, and Accountability 0192862413, 9780192862419

How should we decide a single employee's accountability in a corporation that commits egregious wrongs? What about

233 39 2MB

English Pages 272 Year 2022

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Dedication
Contents
Preface and Acknowledgments
Introduction
I. THEORY
1. Divisions of Agential Labor
2. Authority-Based Accountability
3. Establishing a Division of Agential Labor
4. Imperfect Divisions of Agential Labor
II. APPLICATIONS
5. War Ethics
6. Accomplice Liability
7. Respondeat Superior and Enterprise Liability
8. Institutional Racism
Conclusion
Appendix
Bibliography
Index
Recommend Papers

Authority, Cooperation, and Accountability
 0192862413, 9780192862419

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

Authority, Cooperation, and Accountability

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

Authority, Cooperation, and Accountability SABA BAZARGAN-FORWARD

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom 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 is a registered trade mark of Oxford University Press in the UK and in certain other countries © Saba Bazargan-Forward 2022 The moral rights of the author have been asserted First Edition published in 2022 Impression: 1 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 licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2022934330 ISBN 978–0–19–286241–9 DOI: 10.1093/oso/9780192862419.001.0001 Printed and bound in the UK by TJ Books Limited Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

This book is dedicated to my parents, Shahrzad and Mohsen Bazargan

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

Contents Preface and Acknowledgments

xi

Introduction

1 I. THEORY

1. Divisions of Agential Labor 1.1 Function and Authority in a Division of Agential Labor 1.1.1 On Agentive Functions 1.1.2 Authority in an Interpersonal Division of Agential Labor

23 24 25 26

1.2 Autonomy in Divisions of Agential Labor 1.3 Conclusion

28 33

2. Authority-Based Accountability 2.1 Argument for Authority-Based Accountability

35 37

2.1.1 Whose Motivating Reasons? 2.1.2 Are Protected Reasons Necessary for Authority-Based Accountability? 2.1.3 Implications for Executor Accountability 2.1.4 Authority-Based Accountability in Groups 2.1.5 Authority-Based Accountability Versus Ordinary Accountability

2.2 Contrast with Other Views 2.2.1 Abraham Roth 2.2.2 Christopher Kutz 2.3 Counterarguments and Arguments from Casuistry 2.3.1 Deontic Relevance of Motivating Reasons 2.3.2 Causally Inert Promises 2.3.3 Consequentialist Reasons to Cooperate in Wrongdoing 2.3.4 Group-Size and Authority-Based Accountability 2.4 Conclusion

3. Establishing a Division of Agential Labor 3.1 Promises 3.1.1 3.1.2 3.1.3 3.1.4

Normative Power Views Conventionalism Expectationalism Immoral Promises

38 44 45 47 53 54 54 58 60 60 66 67 68 70

73 76 76 78 80 81

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

viii

 3.2 Requests and Agreements 3.2.1 Requests 3.2.2 Agreements 3.3 Commitments in Shared Action 3.3.1 Margaret Gilbert 3.3.2 Michael Bratman 3.4 Conclusion

84 84 87 88 89 91 95

4. Imperfect Divisions of Agential Labor 4.1 Alienation

97 98

4.1.1 Alienated Executors 4.1.2 Alienated Deliberators

4.2 Coercion 4.2.1 Coercion and Accountability in General 4.2.2 Coerced Promises 4.2.3 Coercion in Cooperatively Committed Harm 4.3 Ignorance 4.3.1 Ignorant Executors and Ignorant Deliberators 4.3.2 Ignorance in Cooperatively Committed Harms 4.4 Conclusion

99 102

105 107 108 114 116 116 121 126

II. APPLICATIONS 5. War Ethics 5.1 Revisionism and Authority-Based Accountability in War 5.1.1 5.1.2 5.1.3 5.1.4

Revisionist Accounts of War Ethics Cooperatively Committed Harms in War Example of a Cooperatively Committed Harm in War Challenges

5.2 Mitigating Conditions and Authority-Based Accountability in War 5.2.1 Ignorance in War 5.2.2 Alienation in War 5.2.3 Coercion in War 5.3 Conclusion

6. Accomplice Liability 6.1 Standard Account of Accomplice Liability 6.2 Superfluity of Complicity 6.2.1 ‘Complicity’ as Non-Derivative Liability 6.2.2 Non-Derivative Liability and Reckless ‘Complicity’

6.3 Complicity and Authority-Based Accountability 6.3.1 The Re-Emergence of Complicity 6.3.2 Complicity as Vicarious Liability 6.4 Conclusion

139 140 141 143 146 149

151 151 154 158 160

162 163 165 165 169

172 172 175 179

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi



7. Respondeat Superior and Enterprise Liability 7.1 Enterprise Liability as Grounds for Respondeat Superior 7.1.1 The Scope of Employment 7.1.2 Occasioning vs. Aiding a Tort

7.2 Inadequacy of Attempts to Ground Respondeat Superior in Enterprise Liability 7.2.1 Assumption of Risk and Enterprise Liability 7.2.2 Causation and Enterprise Liability 7.3 Authority-Based Accountability as a Solution 7.3.1 Duties of Care and Asymmetric Vulnerability 7.3.2 Reparative Duties and Conferring Duties of Care 7.4 Conclusion

8. Institutional Racism 8.1 Metaphysics of Racism 8.1.1 Non-Reductionist Accounts of Racism 8.1.2 Reductionist Accounts of Racism

ix

182 183 184 187

191 191 195 197 197 202 205

207 210 211 215

8.2 Accountability for Institutional Racism 8.2.1 Implicated Employees 8.2.2 Relevance of Others’ Intentions 8.2.3 Modal Tracking and Authority-Based Accountability 8.2.4 Alternative Explanations of Employee’s Wrongdoing 8.3 Conclusion

218 218 220 222 224 226

Conclusion

229

Appendix Bibliography Index

232 237 247

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

Preface and Acknowledgments How do we make sense of moral accountability in cases where multiple individuals are cooperating in a way that results in a wrongful harm? In this book I make the case for thinking that distinct aspects of human agency, normally wrapped up in a single individual, can be practically distributed across different individuals. The resulting division of agential labor makes possible a distinctive way in which one person might be accountable for the actions of another. Over the first half of this book, I develop this form of interpersonal accountability—which I call “authority-based accountability”—into an approach for handling problems of collective responsibility in the context of cooperative activity. In the second half, I apply this account to war ethics, accomplice liability, business ethics, and institutional racism. I hope to show over the course of this book that what matters morally is not just our causal contributions to what we together do. In addition, the purposes we confer upon one another matter as well. Though the subject matter and methodology adopted in this book is patently philosophical, the implications are personal. It is my hope that the considerations raised here will prompt us to scrutinize the ways in which we are complicit in wrongdoing committed by the corporations and institutions of which we are a part. This book is an offshoot of my graduate thesis at Rutgers University; as such, I am deeply indebted to my doctoral committee members for guiding me in the development of my dissertation. I thank Jules Coleman, Doug Husak, Christopher Kutz, and Jeff McMahan. Were it not for Christopher Kutz’s book Complicity: Ethics and Law for a Collective Age, I would not have been inspired to write my book. I am also indebted to Jeff McMahan. I benefited incalculably from his patience and magnanimity. Were it not for him, I would not be where I am today. The individual who had the most influence on the overall shape of this book, and who probably devoted the most time to providing comments, is Seana Shiffrin. She generously labored through a largely incomprehensible early draft. Those who kindly read a slightly more developed early draft of the book include Lucy Allais, Richard Arneson, Reuven Brandt, David Brink, Eddy Chen, Cory Davia, Kathryn Joyce, Andy Lamey, Dana Nelkin, Samuel Rickless, Clinton Tolley, and Manuel Vargas. I benefited greatly from discussions with them and others at a workshop in UC San Diego. Others who provided helpful written commentary in response to reading particular chapters include Dafna Mark BenShabat, Kimberly Ferzan, Helen Frowe, Michael Hardimon, Ryan Hayes, John Kinney, Kirk Ludwig, Avia Pasternak, and Deborah Tollefsen. I also benefited greatly from discussions of cases with Caroline Arruda, Oisin Deery, Lilian

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

xii

  

O’Brien, Carol Rovane, Marion Smiley, Michael Smith, and Sergio Tenenbaum, all of whom I thank. In addition, I am indebted to the two anonymous referees from Oxford University Press. There are many conferences at which I presented whose audiences provided invaluable feedback. These include the 2012 International Law and Ethics Conference Series at the University of Belgrade, the 2014 Collectivity Workshop at the University of Bristol, a 2014 invited colloquium at the University of British Columbia Philosophy Department, the 2015 meeting of the Society for Applied Philosophy at Dublin University College, the 2016 Central Division of the APA, the 2016 Bled Philosophy Conference in Slovenia, the 2016 Gothenburg Biennial Responsibility Conference in Sweden, an invited symposium at the 2017 Eastern Division of the APA, an invited colloquium at the Arizona State University Philosophy Department, the 2018 International Social Ontology Society Conference at Tufts in Boston, a 2018 invited colloquium at the University of Memphis Department of Philosophy, the 13th Annual Northwestern Society for the Theory of Ethics and Politics in 2019 at Northwestern University, the 2019 Society for Psychology and Philosophy Conference at UC San Diego, the 2019 European Network for Social Ontology Conference in Tampere, the 2019 Workshop on New Horizons in Action and Agency at the University of Helsinki, and the 2021 Small Acts, Big Harms Conference also at the University of Helsinki. Funding for work and research on this book came from the Hellman Fellowship, the UC San Diego Center for the Humanities Faculty Fellowship, and the UC San Diego Institute of Arts and Humanities Manuscript Forum. Finally, I want to thank my wife, Kristina Bazargan-Forward for her patience and unwavering support for me on all the days in which I was immersed in my work on this book.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

“Stand round me, men. Ye see an old man cut down to the stump; leaning on a shivered lance; propped up on a lonely foot. ‘Tis Ahab— his body’s part; but Ahab’s soul’s a centipede, that moves upon a hundred legs.” —Herman Melville, Moby Dick, Ch. 134.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

Introduction “If I were asked to put forward an ethical principle which I considered to be especially certain, it would be that no one can be responsible, in the properly ethical sense, for the conduct of another.” H. D. Lewis, “Collective Responsibility,” 1948¹ On a common-sense picture, an individual’s accountability for a wrongful harm is limited by her causal reach. Did she causally contribute to the harm in question? If so, how much? Could she have prevented the harm in question? If so, by how much? The problem with this common-sense view, though, is that it seems to under-impute accountability in important cases. Here are three examples, drawn from real life. In October of 2008, Aisha Ibrahim Duhulow was traveling on foot to Mogadishu to visit her grandmother when three armed men accosted and raped her.² After reporting the crime to Al-Shabaab Islamist militants who controlled the southern city of Kismay, they arrested Duhulow on the charge that she had “chatted up” the men and then committed adultery with them. A tribunal sentenced her to death by stoning. On October 27, several militants transported Duhulow to a public stadium in Kismayo where four militants forced Duhulow into a hole in the ground, burying her to her neck. About fifty militants participated in the ensuing lapidation. After ten minutes, she was dug out of the hole; upon discovery that she was still alive, she was buried again. The stoning resumed until she was killed. How do we determine who is accountable for what, among the fifty militants who stoned Duhulow to death? Are we supposed to adjudicate each militant’s accountability by tracing his causal contribution to Duhulow’s demise? Putting aside the epistemic difficulties, such a strategy is morally fraught in that some of the militants, despite their best efforts, might not end up accountable for murder. This conclusion is hard to accept. Surely each of the militants is accountable for the murder that they together committed. By the end of the second world war, over 7,000 SS personnel had been stationed at Auschwitz,³ of which 75 percent performed guard duty. External camp security was under the authority of an SS unit known as the “ ‘Wachbataillon’ ”. These guards manned watchtowers and patrolled the perimeter fences of the camp.

¹ (Lewis H. D., 1948).

² See (Howden, 2008).

³ See (Weale, 2010).

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0001

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

2

, ,  

Members of the Wachbataillon seldom, if ever, had direct contact with prisoners. (Exceptions occurred due to prisoner escapes or uprisings, such as the 1944 Crematorium Revolt.) Auschwitz would likely not have operated absent the guards stationed there. Yet, the typical SS member of the Wachbataillon did not cause or substantially contribute to even a single death among the 1.1 million who were murdered at Auschwitz. If an individual’s accountability for a wrongful harm is limited to what is within her own causal reach, then the typical member of the Wachbataillon bears little to no accountability for the genocide committed at Auschwitz. This conclusion is, again, hard to accept. Surely the SS guards at Auschwitz bear substantial individual accountability for the genocide that occurred there, even if their individual causal contributions were so small as to be unnoticeable.⁴ Both of these examples involve individuals contributing to murder. Perhaps we are inclined to collectively inculpate them—despite that their individual contributions are negligible—because the individuals involved seem “hell-bent on doing wrong.”⁵ The common-sense picture of accountability, though, also has difficulty with more mundane cases. Consider those who work for Philip Morris International (PMI)—an American transnational cigarette and tobacco manufacturing company, with products sold in over 180 countries. The company produces and peddles one of the most addictive and harmful products in the world— tobacco cigarettes—the use of which is the single greatest cause of preventable human death globally. Of course, not everyone who works for PMI is involved in the sale or manufacture of tobacco products. In particular, some are euphemistically called “outreach workers”—their job is to micro-target citizens on social networking sites via “front groups” with the aim of cajoling or cozening them into signing petitions, voting on referenda, engaging with elected officials, and testifying at state or local hearings, all to further PMI’s goal of fighting anti-tobacco legislation. Any given outreach worker, individually, might make virtually no difference to PMI’s chances of affecting such legislation. But in combination, their work is quite effective. If an individual is accountable only for what is within her own causal reach, then no single outreach worker is accountable for what they together do. To some, this consequence is difficult to accept. These cases (and others)⁶ help evince a recalcitrant difficulty in our attempts to attribute individual accountability for the wrongful harms committed cooperatively. In these three cases the contribution that any given ‘cooperant’ makes is just a drop in the bucket when it comes to what they together do. How ought we to make sense of individual accountability in such contexts? Is an individual cooperant accountable only for what is within his or her own causal reach? Or can he or ⁴ This is assuming we can even make sense of the claim that causation comes in degrees. See (Sartorio, 2019). ⁵ See (Lepora & Goodin, 2013, p. 8). ⁶ See (Kutz, 2000, pp. 115–123) for examples.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



3

she be accountable for more than that? And if so, how? These are the questions I address in the first half of this book. In responding to the kind of challenge that questions like these raise, some appeal to Kantianism or, alternatively, to Rule-Consequentialism in order to show that an individual cooperant can be accountable for more than what is within his or her own causal reach.⁷ But these solutions are largely unavailable to those who are skeptical of these normative theories. We might instead argue that the individual cooperants are together, as a group, accountable.⁸ This approach is not contingent upon any particular theory of normative ethics. But it risks ontological extravagance insofar as it suggests that accountability can be properly attributed to entities other than individual persons. The novel account I develop in this book avoids these problems; it is analytically flexible in that it does not require adopting any particular theory of normative ethics, and it is ontologically modest in that it does not attribute accountability or agency to groups qua groups. I will argue that individual cooperants furnish a particular wrongful purpose for every other cooperant’s conduct. Though each Al-Shabaab militant, each SS member of the Wachbataillon, and each PMI outreach worker, has a particular purpose in virtue of having accepted the role designated by superior officers and administrators, these cooperants also confer purposes upon one another. This is because they agree, if only implicitly, to work together as a team in furtherance of achieving a particular goal. The accountability that any given cooperant bears for what they together achieve exceeds the accountability she bears for her own negligible contribution to that outcome because she is among those who furnished for the other cooperants the purpose of achieving such an outcome—or so I will argue. To show this, I make the case for thinking that distinct aspects of human agency, normally wrapped up in a single individual, can be practically distributed across different individuals. The resulting division of agential labor makes possible a distinctive way in which one person might be accountable for the actions of another. I develop this form of interpersonal accountability into an approach for handling problems of collective responsibility. Begin with a single-person case. I note that, in general, a person’s purpose can determine whether her conduct is wrongful. Suppose a politician under criminal investigation offers an expensive gift to the spouse of the district attorney. Or suppose the White owner of a business fires a Black employee. Or suppose an airforce pilot drops a bomb on a munitions factory knowing that the ensuing explosion will kill nearby villagers. The purpose these persons assigned to their actions will determine whether the politician committed a bribe, whether the business-owner committed an act of racism, and whether the pilot committed a war-crime. ⁷ See (Kutz, 2000, pp. 124–145) for helpful discussion and criticism of these possibilities. ⁸ For an overview of this strategy, see (Collins, 2019).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

4

, ,  

So, when evaluating a putative wrongdoing, we must evaluate the purpose for which it was done.⁹ It is generally presumed that the wrongdoer is the one that fixes such a purpose. But a thesis of this book is that such a presumption is mistaken. To evaluate a wrongdoer’s conduct, we need to evaluate its purpose— and to do that, we will often need to repair to the reasons that someone else took there to be in favor of that conduct. This happens when the wrongdoer’s role is to execute the decisions of someone whom the wrongdoer is tasked with helping. A dyadic case of this sort, where one individual helps another achieve a goal, will serves as a ‘building block’ for more complex cases. Consider the following dyadic example. Olympic Sabotage Contender is vying with Victim for first place in an Olympic competition. Contender consequently hires Goon to maim Victim in a way that will put her out of the running. Goon promises to do so and Contender accepts that promise. Goon subsequently maims Victim. Imagine that after the assault, Victim demands that Goon explain why he attacked her. Goon might give his reasons for attacking her, of which there might be a variety: he wanted the money, he enjoyed the work, he liked Contender, he disliked Victim, and so on. But at some point, he might say “Look, I attacked you because it was my job. If you really want to know why you were attacked, you’re going to have to ask Contender.” If Victim wants a full account of the purpose for which Goon acted, there is a sense in which she has to ‘look through’ Goon’s reasons to Contender’s. Put less metaphorically, Contender constitutively determines the purpose of what Goon does. Suppose Goon does not know that Contender and Victim are competing for first place. Rather, he is just given a face and a name, and instructed to maim the target. Nonetheless, Contender has made it so that Goon’s action has the purpose of facilitating Contender’s victory in the upcoming competition. If Contender had some other reason for contracting the assault on Victim, the purpose of Goon’s action in committing the assault would ipso facto change accordingly. To better see this, suppose, contrary to what was assumed, Contender did not know that the person she asked Goon to attack was Victim. Instead, Contender hired Goon to discover and maim the person who has been pseudonymously posting disparaging remarks about her on social media. It is only after the assault that Contender realizes Victim was the target. Though the assault forces Victim

⁹ Some theorists deny this. I address this issue in section 2.3.1, chapter 2, p. 60.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



5

out of the competition, that is not why she was assaulted in this version of the example. In this version, the purpose that Contender furnishes for Goon’s conduct is different from the purpose she furnished in the original version. We can even imagine a case where Contender contracts the assault for Victim’s benefit. Perhaps there is a credible threat on Victim’s life should she compete; Victim refuses to take the threat seriously despite evidence of its credibility. Contender consequently contracts the assault in order to save Victim’s life. The point here, again, is that Contender can affect, from afar, the purpose of what Goon does. Contender is thereby accountable for a morally relevant feature of what Goon does: the purpose for which the assault is committed. By furnishing from afar the purpose for which Goon acts, Contender is thereby accountable for that difference she makes in the purpose that Goon’s action has. Of course, Contender is also accountable for what Goon does in a more familiar sense: by contracting the assault, Contender causally influences what Goon subsequently does. Contender presumably bears accountability for doing so.¹⁰ But the point here is that that Contender is also accountable for constitutively determining the purpose of Goon’s actions. This brief analysis of a rudimentary example demonstrates, if only roughly, that morally evaluating a wrongdoer’s conduct will sometime requiring adverting to someone else’s reasons, which constitutively determines the purpose of the wrongdoer’s conduct. We can now generalize. I will call an individual who has the function of constitutively determining a purpose for which another person acts a ‘deliberator’. And I will call an individual who has the function of acting for a purpose which someone else constitutively determined an ‘executor’. Where the purpose that the deliberator furnishes for the executor is morally bad, the deliberator can be accountable for that wrong-making feature of what the executor does. An upshot is that, in morally evaluating conduct, the agency that matters belongs not only to those we metaphysically attribute the conduct, but also to those whose ends the conduct has the function of enacting. But how do you come to be my deliberator? And how do I come to be your executor? Put differently, how do you come to have the function of constitutively determining a purpose for which I act? And how do I come to have the function of acting for the purpose you furnish? The answer is that you qualify as a deliberator and I qualify as an executor if you possess practical authority over me. The word ‘authority’ naturally brings to mind formal and regimented relationships, such as the relationship a commanding office has to her subordinate, or the relationship a bishop has to a priest, or the relationship an employer has to an

¹⁰ Some argue that intervening agency diminishes accountability. For more on the moral relevance of intervening agency see (Bazargan-Forward, 2016). For canonical discussion of this issue in the context of legal theory, see for example (Hart & Honoré, 1958), (Kadish S., 1985). See also (Zimmerman, 1985), (Hurd, 2001), and (Moore, 2009).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

6

, ,  

employee. But the sense of ‘authority’ I am concerned with here is perfectly prosaic. We place others under our authority by soliciting promises, by forming agreements, by making requests, by issuing demands, and by undertaking shared action. Practical authority permeates our social lives. More specifically, you have practical authority over me just in case you confer upon me what Joseph Raz called a “protected reason” to act in the way that you wish.¹¹ A protected reason comprises a first-order reason and a second-order reason. The first-order reason is to do as you instruct because you instructed it. The second-order reason is to ignore a certain class of first-order reasons against doing as you instruct. Take, for example, a simple promise. Suppose you’re thinking about hand-delivering a package to a co-worker for his birthday tomorrow. But you’re unable to make the drive. I promise that I will make the delivery should you end up wanting me to do so. You accept this promise. When the time comes, you ask me to deliver the package. There might be a host of reasons for me to do so. (Maybe I want to make you happy, maybe I want to make your co-worker happy, maybe I just want to go for a drive, etc.) But among the reasons is the very fact that I promised to deliver the package. It is in this respect that the promise yields a first-order reason in its favor. In addition, my reason to deliver the package excludes certain competing reasons; for example, the fact that I do not feel like delivering the package does not count among the reasons against doing so. Rather than weigh the pros and cons of delivering the package, the promise I made is, in and of itself, supposed to ‘settle the matter’ for me as a practical issue (absent a relevant change in circumstances). The combination of the first- and second-order reasons yields for me a protected reason to deliver the package should you ask me to do so. It is in this respect that you have practical authority over me when it comes to deciding whether I should deliver the package. And it is in virtue of this authority you have over me that you constitutively determine the purpose of my conduct. Put differently: it is in virtue of this authority that you count as a deliberator and I count as an executor. You are supposed to weigh the reasons for and against sending the package, whereas I am supposed to simply act on the decision you reach; you thereby furnish the purpose of my conduct. The example helps show that the sort of authority a deliberator has over an executor is not limited to regimented, formal relationships. When I make a promise that you accept, you thereby take on the role of a deliberator and I thereby take on the role of executor—within the scope of conduct falling under the aegis of that promise. In this case, you are in a privileged position with respect to what it is I should do, practically speaking, because I volunteered to grant you that authority by making the promise in the first place.

¹¹ (Raz, 1979, p. 18).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



7

If in such cases you furnish my conduct with a wrongful purpose, you are thereby accountable for a wrong-making feature of my conduct. The purpose you furnish depends on the reasons you take there to be—that is, your motivating reasons (as I will use the term). Suppose your reason in gifting the package to your coworker is not to express felicitations on her birthday but rather to commit a bribe. You are then accountable for furnishing my conduct with the wrongful purpose of committing a bribe. To be clear, your accountability in no way indemnifies me against accountability for what I do under the aegis of your authority. I can bear accountability too if I suspect that you are intent on committing a bribe. The point, though, is that because of you, my conduct has a particular wrong-making feature: a wrongful purpose. In virtue of your status as a deliberator, and my status as an executor, you are accountable not only for the difference you make in whether the package is sent, but also for the difference you make in the purpose for which it is sent. I call this “authority-based” accountability. Over the course of this book, I will develop, defend, and apply this concept. Authority-based accountability might seem otiose in dyadic cases. Consider again Olympic Sabotage. A common-sense picture of accountability is enough to get us what we want: both Contender and Goon are accountable for the harm done to Victim by virtue of intentionally causing that harm. But authority-based accountability in dyadic cases helps us locate accountability in more complex examples of cooperation. Return now to cases of more complex cooperative activity, such as the examples involving the Al-Shabaab militants, the SS Wachbataillon, and the PMI outreach workers. Recall that by hypothesis the cooperants in each case implicitly agree with one another to help achieve a wrongful goal. The result is tantamount to a case in which each cooperant makes pairwise agreements with each and every other cooperant. As a result of these agreements, each cooperant has authority over every other cooperant, in that each confers upon every other a protected reason, and thus a purpose, to do her part. Should any cooperant fail to ‘pull her weight’, the other cooperants would have a basis for complaint. In virtue of the agreement they made, each cooperant qualifies as both a deliberator and executor in n-1 pairwise agreements, where ‘n’ is the total number of cooperants. By conferring protected reasons upon each other, the cooperants thereby constitutively determine for one another the purposes for which each acts. The point can be put this way: in Olympic Sabotage, Goon acted at Contender’s behest, whereas the Al-Shabaab militants, the SS Wachbataillon, and the PMI outreach workers are acting at one another’s behest. What is the purpose that the cooperants furnish for one another? That depends on the practical reasons each cooperant takes there to be. Suppose I am one of fifty members on the PMI outreach team. I thus serve as both a deliberator and an executor. I want everyone else on the team to help forestall anti-tobacco legislation because success of that sort will increase my chances of procuring a raise. I thereby

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

8

, ,  

furnish that end for every other teammate, in my capacity as a deliberator. The result is that each teammate’s conduct has—wittingly or not—the following purpose, among others: to help me procure a raise by successfully forestalling anti-tobacco legislation. Likewise, the other teammates severally furnish for my conduct, in my capacity as executor, a concomitant purpose derived from each of their motivating reasons.¹² Such a description of the relationship the teammates bear to one another might seem inapt; it seems to suggest that each teammate functions as a kind of “miniboss” insofar as each counts, in her own right, as a deliberator capable of furnishing a purpose for each of her teammates’ actions. Note, however, that unlike a boss, the teammates also serve as executors for one another. Indeed, what makes for a boss on the picture I have outlined is precisely that the individuals over whom the boss has authority do not possess symmetrical authority over the boss. Each teammate, then, qua deliberator, furnishes a purpose for her other teammates’ actions. Recall, though, that when an individual qua deliberator furnishes a wrongful purpose for someone else qua executor, the deliberator is accountable for a wrong-making feature of what the executor does: the purpose for which she acts. The same goes for the PMI outreach team. Each teammate, qua deliberator, is accountable for the purpose she furnishes for her teammate’s conduct. If it is wrongful to forestall anti-tobacco legislation as a means to procuring a raise, then I am accountable for conferring that wrongful purpose upon my forty-nine teammates’ conduct. So, even if I have little or no causal influence over what the others do, and even if I fail to contribute much to what we together do, I still have normative influence—that is, I have influence over what they are supposed to do and why they are supposed to do it. This is not to say that the other teammates would have acted differently absent me. Neither is this to say that they are aware of the specific purpose that I furnish for them. Rather, it is to say that regardless of what actually motivates any other member of the team to contribute, every other teammate’s action will have, because of me, a purpose that it would not have otherwise had; mutatis mutandis for every other teammate. We can tell a similar story about the Al-Shabaab militants and about the soldiers in the SS Wachbataillon.

¹² On the picture I have outlined so far, a purpose derives from a motivating reason, which determines the content of that purpose. The deliberator confers a purpose upon the executor, in that the deliberator’s motivating reasons determine a purpose that the executor has. The executor comes to have such a purpose by adopting a protected reason to act accordingly. Purposes and reasons are both teleological in nature in that they both putatively speak in favor of a specified end. What, then, is the difference between the two? One difference between purposes and reasons, as I understand them, is that reasons are subject to an internalist constraint making them suitable as premises in practical reasoning, whereas purposes might remain introspectively opaque to the executor. I discuss this in greater depth in sections 2.1.2 and 2.2.1, chapter 2.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



9

How bad is it, though, to furnish for others a wrongful purpose to act in a way that they would have acted anyway? After all, absent my participation in the PMI outreach team, the other cooperants would still have had the purpose of forestalling anti-tobacco legislation, given that we’re each conferring upon one another a purpose roughly of that sort. It seems, then, that I make little difference not only to the harm that my team causes, but also to the purpose my teammates have. In both cases, my contribution is just a drop in the bucket. So, it might seem that furnishing a wrongful purpose does not do much to inculpate me after all. Suppose you are also a PMI outreach worker. Thus, you are among the cooperants for whom I furnish a purpose. It’s true that I am not a but-for cause of the fact that what you do is wrongful. But the purpose I furnish is by itself sufficient in that it alone is enough to make what you do wrongful. The result is that we ought to severally rather than jointly evaluate what each cooperant does when she furnishes a purpose, in which case it turns out that the purpose I furnish is wrong-making after all. In evaluating what I do when I furnish a purpose for you, the purpose I furnish is not a drop in the normative bucket. It instead can be construed as the whole bucket; mutatis mutandis for each of the other teammates who furnishes a purpose for you. The claim that a deliberator furnishes a purpose for an executor might be misread. The idea is not that the executor now takes himself to have the purpose that the deliberator furnished for him. Indeed, as I mentioned, the executor might not even be aware of what his purpose is. It might be more felicitous, then, to say that the deliberator furnishes a purpose for the executor’s conduct (rather than the executor himself). This distinction is especially important since, on the account I develop, the deliberator is accountable for features of the executor’s conduct, rather than for features of the executor as an agent. But for the purpose of brevity, I will sometime describe the deliberator as furnishing a purpose for the executor, rather than for the executor’s actions. I will briefly recap. The purpose a deliberator furnishes for an executor serves as a basis for evaluating what the executor subsequently does qua executor. Where that purpose is problematic, the deliberator is thereby accountable for a wrongmaking feature of what the executor does. Moreover, in evaluating the deliberator’s accountability for that wrong-making feature, we treat the deliberator as if she is the only one furnishing a purpose for the executor. To be clear, the argument I develop is meant to establish that the deliberator can be accountable for features of the executor’s conduct—not for features of the executor as an agent. That is, a culpable deliberator who instructs her executor to perform a wrongful act, doesn’t necessarily make the executor a bad person. Rather, the culpable deliberator makes the executor’s conduct bad. And the culpable deliberator will typically be accountable for that difference she makes, constitutively, to the executor’s conduct. Meanwhile, the executor might remain none-the-wiser that her conduct has been made bad in this way. If the executor

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

10

, ,  

knows or ought to know that her conduct is made bad in this way, then she might indeed be on the hook for it—as I argue in Chapter 8 where I discuss structural racism. So, in our PMI outreach group, when evaluating what I do when I, qua deliberator, confer upon you, qua executor, the purpose of helping forestall antitobacco legislation, it turns out that your conduct counts as wrongful because of me. Ditto for every other member of my team; in evaluating any given teammate when she, qua deliberator, confers upon you, qua executor, the purpose of helping forestall anti-tobacco legislation, it turns out that your conduct counts as wrongful because of her. Likewise, when evaluating what you do when you, qua deliberator, confer upon me, qua executor, the purpose of helping forestall anti-tobacco legislation, it turns out that my conduct counts as wrongful because of you. Proceeding in this way, every individual teammate is accountable for making it so that every other teammate’s conduct counts as wrongful. The cases I’ve discussed so far raise many questions. Here are a few, focused on the PMI outreach team. 1) The members of PMI’s outreach team individually agree with one another to do something morally wrong; how can an immoral agreement yield normative reasons in general, and protected reasons specifically? 2) Each member of PMI’s outreach team is party to an agreement with every other member as a result of which she servers as both a deliberator and an executor. What about groups the members of which cooperate without making or accepting any such agreement? Do the cooperants bear authority-based accountability? 3) Is the project manager’s unilateral authority over the outreach workers compatible with their authority over one another? And does her authoritybased accountability differ from those that the outreach workers bear? 4) The example presumes that each teammate is committed to forestalling anti-tobacco legislation as a means to achieving her own goals, which, in turn, informs the purpose she furnishes for her other teammates. But what about those who aren’t committed to that wrongful end, even instrumentally? Do they bear authority-based accountability for what the others do? 5) All the members of PMI’s outreach team know what they are doing. But what happens when some or all of the cooperants are ignorant of or deceived about what they’re doing together? Do they bear authority-based accountability? 6) The members of PMI’s outreach team are cooperating voluntarily. How does authority-based accountability apply in cases where the individuals are coerced into cooperating?

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



11

I address all these issues in the first half of the book. The upshot is this: the argument for authority-based accountability grounds the view that cooperating individuals can each be accountable for more than what is within her own causal reach. More specifically, any given cooperant can be accountable for the fact that what the other cooperants do counts as wrongful. What exactly does it mean, though, to say that individuals cooperating can each be accountable “for more than what is within her own causal reach”? After all, one might point out that the process of establishing a division of agential labor is itself a causal process. To be clear, though, the basis for a deliberator’s authority-based accountability is that she constitutively (rather than causally) determines the purpose of the executor’s actions. It’s true that this happens only once they establish the right sort of relationship, which is, in and of itself, a causal affair. But that relationship doesn’t necessarily serve as a cause of the executor’s subsequent actions.¹³ And even if the relationship does indeed serve as a cause of what the executor subsequently does, authority-based accountability inculpates the deliberator for much more than that.¹⁴ Thus, the moral significance of authority-based accountability cannot be “ ‘reduced’ ” to the moral significance of causation. The account I develop helps us make sense of individual accountability for harms that cooperants commit together. We want an individual cooperant to end up accountable for potentially more than what is within her own causal reach. Yet we need to provide a theoretical framework explaining why. We cannot just stipulate this. To do so is to imply that we have hit bedrock in our moral analysis. To allege that our spade is turned is problematic since the intuition at hand seems to conflict with the equally plausible pronouncement quoted at the outset of this introduction: that “no one can be responsible, in the properly ethical sense, for the conduct of another.” The recalcitrance of this problem in ethics and in legal theory is hard to overstate. But I believe that the account of authority-based accountability, developed in the first half of this book, helps solve this problem. Though the machinery I develop is complex, the intuitions undergirding authority-based accountability are simple and, I believe, compelling: when I commit a wrong at your behest, my victim can licitly direct her resentment and indignation not just to me, but to you, since there is a sense in which you are acting vicariously through me. When cooperants together cause a wrongful harm, there is likewise a sense in which they are acting vicariously through one another, as a result of which each cooperant can be accountable for more than what is within her own causal reach. As part of its intuitive appeal, the argument for authority-based accountability makes sense of accountability for what others do in the context of cooperation ¹³ I discuss this point in section 2.3.2, chapter 2, p. 66. ¹⁴ I discuss this point in section 2.1.5, chapter 2, p. 53.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

12

, ,  

using only our ordinary notions of intentions, with ordinary events and states of affairs as their objects, and ordinary individual human being as their subjects. That is, I do not invoke group-agents,¹⁵ novel propositional attitudes,¹⁶ propositional attitudes with alternative “modes”,¹⁷ or propositional attitudes with irreducibly collective content.¹⁸ This is not to say that accounts invoking these concepts are mistaken. Neither is it to say that such concepts are morally irrelevant to accountability for what others do—far from it. But I believe that such concepts are not necessary in understanding, at the most fundamental level, why individuals can be accountable for what others do in the context of cooperative action.¹⁹ This project, though, is limited in two ways. First, the argument for authoritybased accountability applies solely to cases in which one or more individuals serve as executors for one or more deliberators. I will say that the harms that they together cause qua deliberators and executors are “cooperatively committed”.²⁰ The argument for authority-based accountability does not apply to the aggregate of the harms that individuals commit when acting on their own. For example, individuals around the world use excessive amounts of disposable plastic; though the typical individual makes only a negligible difference, the aggregate of all their individual actions result in substantial environmental harms. The argument for authority-based accountability does not apply in this sort of case, because the aggregate harm is not cooperatively committed.²¹ Second, authority-based accountability is not meant to serve as the sole basis by which individuals can come to be accountable for what others do. There might be a host of other ways. For example, perhaps willingly accepting a benefit from a wrong committed by others inculpates the beneficiary in that wrong.²² Or perhaps outcome luck is morally irrelevant to accountability. Thus, a mere attempt to act in a way that foreseeably risks facilitating the wrong I commit might inculpate you, even if the attempt fails due to luck. Or perhaps there is indeed a basis for thinking that agents who gratuitously contribute to collectively—but not cooperatively—caused harms, such as climate change, are accountable for more

¹⁵ See for example (List & Pettit, 2013). ¹⁶ See for example (Sellars, 1980) and (Tuomela, 2010). ¹⁷ See (Tuomela, 2006) and (Schmitz, 2017). ¹⁸ See (Kutz, 2000). ¹⁹ For an account that adopts a similar methodology, see (Miller, 2006) and (2020). ²⁰ How are cooperatively committed harms related to harms resulting from shared action? We share an action when we coordinate and combine our individual actions in furtherance of doing something together, as when we walk together or paint a house together. In such a case, the shared action is a result of a shared intention, which is itself comprised of individual intentions interdepending in the right way. As I will indicate in chapter 3, harms resulting from shared action are almost always cooperatively committed. But cooperatively committed harms aren’t necessarily the result of shared action. ²¹ For recent discussion of these sorts of cases, see (Schwenkenbecher, 2018), (Collins, 2019), and (Björnsson, 2020). ²² See (Goodin & Pasternak, 2016) and (Bazargan-Forward, 2021).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



13

than what is within their causal reach.²³ Whatever that account looks like, it might also apply to cooperatively committed harms. But this does not mean that the account I develop here is mistaken or otherwise redundant. Rather, it means that there is more than one basis for inculpating individuals who commit a harm cooperatively, which implies that, all things being equal, they are more accountable. Authority-based accountability has several practical implications, which I explore in the second half of the book. I begin with war ethics. On what has come to be known as “revisionist” accounts of war ethics, a combatant’s moral liability in a war depends not on her status as a combatant, but on the morality of the aims for which she is fighting and her own individual contributions to those aims. This view, though, has difficulty accommodating the moral liability of combatants who play highly subsidiary or marginal roles in war. I will explain how the argument for authority-based accountability can help address this challenge. In doing so, I address the worry that the account wrongly inculpates civilians as well. Warfare is a useful and challenging application of the argument for authoritybased accountability not only because warfare is replete with cooperatively committed harms, but because accountability-mitigating circumstances, such as ignorance, alienation, and coercion, are endemic to war. I thus address whether— and if so, how—these conditions affect authority-based accountability. In doing so, I hope to show the various ways that authority-based accountability sheds light on the ethics of war. After applying the argument for authority-based accountability to war ethics, I turn to the legal doctrine of accomplice liability. I am legally “complicit” in a wrongdoing which you committed and for which you are liable when I am also liable for that wrongdoing; I become complicit by culpably contributing to your wrongful conduct. I argue against the orthodox view which construes complicity as a separate basis of criminal liability in the law. We can analyze most cases of complicity, I argue, by adverting to standard causation-based accounts of criminal liability instead. But unlike others who take this heterodox view, I argue that law and morality still ought to make room for complicity as such. This is because on both the orthodox and heterodox views, I am complicit in your wrongdoing only if I causally contribute to that wrongdoing or otherwise causally influence you. But such accounts have difficulty accommodating complicity in cases where multiple individuals work together in furtherance of an overdetermined harm. ²³ See (Nefsky, 2019) for a helpful taxonomy of such views, and for a preliminary defense of the claim that individual contributors in such cases can be accountable when her act plays a “nonsuperfluous part” in changing the outcome even if the act is not itself sufficient to make a difference. See also (Wieland & van Oeveren, 2020) who argue in response that even “superfluous” contributions can be inculpatory.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

14

, ,  

I argue that complicity re-emerges here as a species of vicarious liability, where I am vicariously liable for the wrong you commit if I bear the right kind of formal relationship to you. The relationship that a deliberator has with her executor fits this bill. The upshot is that authority-based accountability helps us make sense of complicity by accommodating cases that “contributory” analyses of complicity fail to accommodate. There are other legal applications for authority-based accountability. According to the legal doctrine of respondeat superior, a person in a position of authority can be held legally liable for the torts that her subordinate commits. Contemporary accounts analyze respondeat superior by grounding it in the doctrine of enterprise liability, according to which those who own a business ought to bear the costs of its negative externalities—which include foreseeable employee misconduct. I argue that we can indeed often chalk up the torts that employees commit to the risks of running a business. But sometimes an owner does not serve as a cause for the tort her employee commits. In such cases, we cannot apply respondeat superior if we ground it in enterprise liability. Moreover, it is unclear why the owners should bear the costs in cases where the enterprise imposes risks on patrons who willingly, knowingly, and reasonably accepted the risk of harm by choosing to accept services from the enterprise. I respond to these two worries univocally. I argue that service-providing businesses have special duties of care toward their patrons, grounded in an asymmetry in the vulnerability of the two parties. An owner implements these duties of care by conferring them upon her employees—and they do this by establishing a relationship with them in which she serves as a deliberator and they serve as executors. Hence, should her employees violate the duties of care toward the patrons, the owner thereby does so as well. This, in turn, serves as grounds for compensatory liability. An upshot is that properly analyzing the doctrine of respondeat superior requires appealing to authority-based accountability. Turning from the law back to applied ethics, I argue that authority-based accountability can help us make sense of individual accountability for institutional racism. It has long been acknowledged that individuals are not the only sort of entity to which racism can be attributed. It can also be attributed to institutions as such. I focus on the accountability of individuals in such institutions. Suppose you are a low-level manager at an institution whose leaders wrongly adopt policies perpetuating racial injustice. Absent any excuse, these leaders are accordingly guilty of racist conduct. But what about you? Suppose that, unlike your superiors, you are not indifferent toward racial injustice. Fortunately, your conduct as an employee does not contribute to racial injustice. Does this mean that, despite working in a racist institution, you are free of accountability? It seems you must be; after all, your actions, your attitudes, and their effects are morally unproblematic.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



15

But I argue that the racist conduct of your superiors inculpates you at the bar of authority-based accountability. This is because the moral effects of racism can “transmit” across individuals—specifically, between deliberators and executors. Thus, what matters morally when we work for racist institutions is not just the harms we commit, but the roles we play in enacting the unjust plans of our employers. Even if, as employees, we carefully avoid harming members of the racialized group in question, we nonetheless commit a pro tanto wrong against them by serving as a means by which our employers enact their racist intentions vicariously through us. In what follows, I briefly summarize each chapter. Chapter 1. What does it mean for you to be my deliberator and for me to be your executor? I argue that we can make sense of such a relationship by starting with a single-person case in which an individual exercises diachronic agency by conferring authority on her past self when it comes to deciding how to act now. In such a case, her past self serves as a deliberator and her present self serves as an executor. This single-person case helps model the normative power that you have to “outsource” the executory function of your agency to me. In this way, I come to serve as your executor. I call this relationship a “division of agential labor”. Chapter 2. Having developed the concept of a division of agential labor, I argue that the deliberator in such an arrangement can end up accountable for what the executor does, quite apart from whether the deliberator causally contributes to the executor’s conduct. But what exactly do I mean by “accountability”? I answer this question by invoking Gary Watson’s concept of responsibility-as-accountability. I then argue that, in general, morally evaluating conduct requires adverting to the actor’s motivating reasons in favor of that conduct. But when you serve as a deliberator and I serve as an executor, your motivating reasons determine my purpose, via (what Joseph Raz called) “protected reasons” to do as you say. In explicating this view, I lean on the analysis of agentive functions that John Searle presents and Kirk Ludwig later develops. The functional relationship between the deliberator and the executor enables the deliberator to furnish a purpose for the executor; by furnishing an inimical purpose, the deliberator is thereby accountable for the fact that what the executor subsequently does is wrongful. This is the argument for authority-based accountability. After contrasting this account with those that Abraham Roth and Christopher Kutz have developed, I defend it against various criticisms and ostensible counterexamples. In doing so, I address the worry that my account depends on the view that intentions are deontically relevant. Chapter 3. How do we establish a division of agential labor? Here I argue that we do so via promises, requests, agreements, and, finally, commitments in shared

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

16

, ,  

action. Though there are a variety of accounts on offer for these social phenomena, the argument for authority-based accountability applies to all of them. With respect to promises, I consider normative-power accounts, conventionalist accounts and expectationalist accounts; each is compatible with the view that promises confer protected reasons, and thus, divisions of agential labor. Requests do so as well, even though, unlike promises and agreements, requests are not binding. In discussing shared action, I consider Margaret Gilbert’s and Michael Bratman’s accounts. I argue that Gilbert’s account necessitates the view that shared action confers protected reasons, whereas on Bratman’s account shared action will typically though not necessarily confer protected reasons. Either way, the result is that those engaged in shared action will typically (if not necessarily) bear authority-based accountability for what they together do. But what about immoral promises, requests, agreements, and shared actions? Do they confer protected reasons? I argue that they do, though the reasons in question are practical rather than moral. After considering some examples of authority-based accountability, I end the chapter by remarking that the account I have developed can help make sense of accountability in the context of group agency. Chapter 4. At this point, I consider less-than-perfect divisions of agential labor. In particular, I explore how alienation, coercion, and ignorance can affect a deliberator’s authority-based accountability for what the executor does. With respect to alienation, what happens when the deliberator does not care whether the executor acts accordingly? Imagine a disgruntled mid-level corporate manager indifferent to whether her subordinates abide by her instructions. How does this affect the manager’s authority-based accountability? Or suppose the executor herself does not care about her role as such. Imagine that I make a promise to ignore your partner’s phone call should he contact us while we’re out on the town. Suppose I end up fulfilling this promise, but by accident rather than because I recognize a protected reason to do so. Do you bear authority-based accountability for what I have done? And what happens when the deliberator—or the executor—denies that the former has any authority over the latter? Imagine a nihilist about promises who ostensibly engages in the practice but denies that it confers any reasons. How does this affect authority-based accountability? After addressing these issues, I turn to cases of coercion. Sometimes participation in a division of agential labor is not fully voluntary. If a deliberator is coerced into accepting a promise, or an executor is coerced into making a promise, then the putative promise is not binding qua promise. I consider the effects that this has on the deliberator’s authority-based accountability for what the executor does. Finally, I address the role of ignorance in authority-based accountability. Executors in a division of agential labor are often in the dark about the nature

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



17

of their tasks. Imagine a soldier ignorant about what exactly she is doing and why she is doing it. Alternatively, a deliberator might be ignorant of what it is she has instructed the executor to do, as is the case should you ask me to surprise you with a gift tomorrow. Even when the deliberator knows exactly what she has instructed the executor to do, she might not know how the executor will choose to fulfill those instructions. If I promise to get us tickets for an exclusive concert, and you accept that promise, do you bear authority-based accountability if I fulfill that promise by stealing the tickets? The end of chapter 4 marks the transition to the second part of the book, in which I apply the account I have develop to a host of issues in ethics and law: war ethics, accomplice liability, respondeat superior, and institutional racism. But before doing so, I cap the discussion so far by applying the concept of authoritybased accountability to a real-life event: the 1947 lynching of Willie Earle in South Carolina. Chapter 5. Warfare is a paradigm cooperative activity. How, then, do we make sense of an individual combatant’s accountability in an unjust war, given that the typical combatant sees no combat and contributes little to the war-effort? I argue that authority-based accountability can help answer this question. I also explain how authority-based accountability copes with the accountability-mitigating circumstances common in warfare, including ignorance, alienation, and coercion. Combatants, then, can bear authority-based accountability for what their comrades do. But what about civilians? Authority-based accountability might seem to illicitly overgeneralize by inculpating civilians as well—but I argue otherwise. Chapter 6. The argument for authority-based accountability has a natural home in legal theory—specifically when it comes to grounding the doctrine of accomplice liability. According to this doctrine, I am legally “complicit” in a wrongdoing which you committed and for which you are accountable when I am also liable for that wrongdoing by culpably contributing to your wrongful conduct. All the accounts on offer state that an individual counts as complicit in a wrongdoing only if she contributes to that wrongdoing or otherwise causally influences the wrongdoer. Such accounts have difficulty accommodating complicity in cases where several cooperants together achieve an overdetermined harm. I argue that we can accommodate the complicity of the cooperants by interpreting complicity as a species of vicarious liability. On this picture, I am vicariously liable for what you wrongfully do provided we are in the right kind of formal relationship with respect to each other—such as the relationship between a deliberator and an executor. The upshot is that authority-based accountability succeeds to accommodate complicity where “contributory” accounts fail.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

18

, ,  

Chapter 7. Authority-based accountability also finds a natural home in an analysis of the legal doctrine known as “respondeat superior”, according to which a person in a position of authority can be held legally liable for the torts that her subordinate commits. There has been, over the past few decades, a move to analyze respondeat superior in terms of enterprise liability, which states we ought to shift the foreseeable negative externalities of running a business to those who own, and thereby derive benefits from operating, the business. Insofar as foreseeable employee misconduct counts as a foreseeable negative externality, we have a basis for holding the owner accountable for the employee’s misconduct. But I argue that if the owner does not serve as a cause for the tort her employee commits, then we cannot make sense of the owner’s liability by appealing to enterprise liability. The result is that we cannot ground all instances of respondeat superior in enterprise liability. And even if we could, it’s unclear why we should shift the costs of addressing the negative externalities from the patrons to the owner, when the patrons willingly, knowingly, and reasonably accepted the risk of harm by choosing to accept services from the enterprise. Given these challenges, we cannot ground respondeat superior solely in enterprise liability. Instead, I take a radically different approach. I argue that there is an inherent asymmetry in vulnerability between service-providing businesses and their patrons. As a result, the owner of the business has a special duty of care toward their patrons. The owner doesn’t enact these duties in her own person. Rather she does so vicariously: she serves as a deliberator and her employees serve as executors upon whom she confers the function of enacting the special duties of care. Consequently, the owner is accountable—in that she bears compensatory liability—should her employees violate those duties of care. The argument for authority-based accountability, then, plays an indispensable role in analyzing the doctrine of respondeat superior in the law. Chapter 8. How do we make sense of individual accountability in the context of institutional racism? Those who adopt policies perpetuating racial injustice are presumptively accountable for doing so. But what about those who work in such an institution, but who are not tasked with adopting or implementing racist policies? If an individual’s “hands are clean” in this way, is she therefore free of accountability for the racial injustices her institution perpetuates? Not necessarily. Suppose a superior instructs a subordinate to do something that what would be racist if the superior were to do it in her own person, but would be non-racist if the subordinate did it in her own person. I argue that in such a case, the act counts as racist even though the subordinate performs it. This is because, according to authority-based accountability, we repair to the superior’s motivating reasons in evaluating the act in question. So, when we work for racist institutions, it is not just the harms we commit that matter, but the roles we play in fulfilling the racist plans of our employers. Even if,

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi



19

as employees, we carefully avoid harming members of the racialized group in question, we nonetheless commit a pro tanto wrong against them by serving as a means by which our employers vicariously enact their intentions. Conclusion. I end the book by considering the implications of authority-based accountability for Bernard Williams’s notion of integrity. In doing so, I argue that the account I have developed elucidates his notion of integrity. In discussing his famous example of the chemist he names “George”, I argue that the attack on his integrity lies not simply in that he causally contributes to an end he finds morally abhorrent. In addition, the attack on George’s integrity lies in accepting the purpose of contributing to such an end. By taking on a job in which he serves as an executor and his employer servers as a deliberator, George has relegated himself to the role of “a conduit” in furtherance of “initiatives, purposes, or concerns” antithetical to his own. The upshot is that the divisions of agential labor in which we participate are crucial not just in assessing accountability for cooperatively committed harms, but in reckoning with the demands of integrity. Appendix. Here I develop in a pseudo-formal system for schematizing kinds of collectives according to the ways in which authority structures the organizational relationships within these collectives.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

PART I

THEORY

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

1 Divisions of Agential Labor At its most basic level practical agency can be divided broadly into deliberative and executory functions. The former is the process by which candidate options are evaluated and selected, and the latter the process by which the selected candidate options are implemented. When an agent makes a practical decision, the agent transitions from the deliberative process to the executory process, in that the function of a decision is to enact via conduct (which might consist of an action or an omission) the practical reasons the agent takes there to be in favor of acting in a particular way.¹ A well-functioning agent will thereby implement the decision she reaches. I make these points in section 1.1. What does it mean to decide to do something? Decisions are types of intentions. An intention to do something is, in turn, a commitment to doing it. For Michael Bratman, intention is characterized by its role in planning for the future, in that it functions as a defeasible commitment, made in advance, to act in a particular way in the future.² Our ability to self-commit in this way yields several advantages. It enables us to act rationally in circumstances where there is not enough time to deliberate or where our deliberative capacities are compromised. And it enables us to manage complex projects requiring coordination with our future selves—or

¹ By ‘reasons’ I mean awareness-independent considerations which count in favor of an action. These are what some call “objective” or “fact-relative” rather than “subjective” or “belief-relative” reasons. For example, see (Schroeder, 2004). I agree with those who take fact-relative reasons to be primitive; see, e.g., (Raz, 1990, pp. 11–12), (Scanlon, 1998, p. 17), (Dancy, 2004, pp. 15, 29). Though this is a controversial view, nothing I say will ultimately depend on this stance. I will use ‘rationality’ as an evaluative term; I am rational to the extent that I act in accordance with my belief-relative reasons. Suppose that all the evidence suggests that taking a particular drug will cure a disease from which I’m suffering when in fact it will kill me. It is therefore rational for me to take this drug. See (Parfit, 1997, p. 99), (Scanlon, 1998, pp. 22–32), (Raz, 1999, pp. 67–89), (Kolodny, 2005). On this account, reasons are explanatorily prior to rationality. Other accounts reverse this order of explanation: reasons are considerations that motivate properly functioning, well-informed agents. For examples of this view, see (Foot, 2001), (Williams, 1981, pp. 101–113), (Korsgaard, 1986). ² (Bratman, 1987). There are two roles intentions as commitments play in our planning activity. On the ‘planning-toward’ sense of intention, we intend to ф when we plan on doing what promotes ф and on avoiding what conflicts with ф. On the ‘planning-from’ sense of intention, we intend to ф when we assume that we will achieve ф, in planning on what to do subsequently. (I borrow these terms from (Cullity, 2008).) For example, a job applicant might prudently apply to two faculty positions in a single department hoping to get one or the other but knowing that she cannot get both. She can rationally intend both outcomes in the planning-toward sense, but cannot rationally intend both outcomes in the planning-from sense. The sense of intentions relevant to the sort of commitments characteristic of decisions is the ‘planning-from’ sense of intention. For more discussion, see (Bratman, 2009) and (Broome, 2009). See also (McCann, 1991).

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0002

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

24

, ,  

with others—over time. This facilitates, on Bratman’s view, a form of selfgovernance characteristic of practical agency. Normally, a single agent embodies both the deliberative and executory functions of agency. That is, normally, you deliberate by evaluating and selecting among candidate options, and you consequently enact the option that you have selected. It is possible, though, for you to “outsource” the executory functions to me. In such a case, you attribute to me the role of enacting the practical reasons you take there to be; should I accept that role, you thereby change the object at which your practical reasoning is teleologically directed. Its object is no longer your conduct, but rather mine. That is to say, the practical reasons you take there to be, have the function of guiding my conduct; likewise, my conduct has the function of enacting the practical reasons you take there to be. (To enact a reason is just to act in the way that the reason prescribes.) Put differently, your ends determine the purpose of my conduct. (In the next chapter, I will focus on the role that this purpose plays in grounding the deliberator’s accountability for what the executor does.) In this way, we establish an interpersonal division of agential labor in that you count as the “deliberator” and I count as the “executor”. The point of establishing an interpersonal division of agential labor, then, is to separate out and assign to multiple agents the deliberative and executory functions of rational agency, so that it is the role of one agent to evaluate and select among candidate options, and the role of another agent to implement that option via conduct. I develop this account in sections 1.1.1 and 1.1.2. There is much that needs to be unpacked here. In particular: how do I come to have the “function” of enacting the reasons you take there to be? What does it mean to say that you “assign” or “attribute” to me the function of enacting those practical reasons? In short, how do we “outsource” the executory functions of rational agency? I argue that we do so by conferring practical authority on others.

1.1 Function and Authority in a Division of Agential Labor How do you and I establish an interpersonal division of agential labor in which I outsource to you some of the deliberative functions of my rational agency, and you outsource to me some of the executory functions of your rational agency? We do so when you attribute to me and I accept that executory functional role, and I attribute to you and you accept that deliberative functional role. This claim, though, needs to be unpacked. In particular, we need to know a) what a functional role is, b) what it means to attribute a functional role to someone, and c) what it means to accept a functional role. In what follows I explain these in turn.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

   

25

1.1.1 On Agentive Functions There are, at the broadest level, two kinds of functions—natural and agentive. We create agentive functions. As John Searle puts it, these are “functions that we do not discover, and that do not occur naturally, but that are assigned relative to the practical interests of conscious agents.”³ For example, a chair, a car, and a wrench have the functions that they do in virtue of our intentions—specifically, our intentions to put these objects to use in furtherance of achieving particular tasks. An agentive-function is design-based if the object possessing that function was created with the intention to use it in furtherance of some purpose. Alternatively, an agentive-function is use-based given an actual or dispositional use of that object in furtherance of some purpose. I will focus here on use-based functions.⁴ It’s possible for your decisions to serve as a use-based agentive function for me. For example, suppose that in certain situations I flip a coin to determine which of two equally choiceworthy options to select. I thereby assign to that coin-flip a oneoff, use-based, agentive function of determining for me which of the two options to select. Lacking a coin, however, suppose I decide to use your actions as my coin. Unbeknownst to you, I decide what to do based on whether you use a contraction in your next uttered sentence. In this sort of case, your agency plays a wholly passive role in fulfilling the agentive function I’ve assigned to you in that the agentive function plays no normative role in your deliberations. I will accordingly call these “passive, use-based, agentive functions”—or just “passive functions” for short. In some cases, though, a function operates by enjoining you, as its bearer, to follow it by acting in accordance with the conduct the function specifies. In this case, your agency plays an active role in fulfilling the agentive function, in that fulfilling it will typically require that you intend to do so (even if the intention is not occurrent at the time of action). I will call this an “active, use-based, agentive function”—or just “active function” for short.⁵ For me to assign an active function ³ (Searle, 1997, p. 20). ⁴ For much more detail on agentive functions see (Ludwig, 2018, p. 105) to whom I am indebted here. As he noted, use-based functions come in two varieties. A use-based function can be grounded in 1) a one-off use of that object in furtherance of some purpose, or 2) a stable disposition to use that object in furtherance of some function, independently of whether it is occurrently being used, or ever has been used, for that purpose. So, for example, I might use a brick to prevent a door from closing. In this case, I use the brick as a doorstop. It accordingly has on that occasion an agentive function—a function in relation to a task for which I put the object to use. If I come to regularly use that brick as a doorstop, then it becomes a doorstop, given the standing disposition to intentionally use it in that way. ⁵ Importantly, this should not be confused with what Kirk Ludwig calls a “status role”. See (Ludwig, 2018, pp. 140–141). Like a normative agentive function, a status role is a function assigned to a person which she accepts as her function. But on Ludwig’s view, status roles are functions that agents have due to collective acceptance of that function in a community. In virtue of that collective acceptance, the agent comes to have a role defined in terms of a constitutive rule for an “essentially intentional action type.” So, for example, an officiate at a wedding cannot perform her function absent collective acceptance of her role as such. The utterance “I pronounce you man and wife” has no purchase absent

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

26

, ,  

to you requires that I, by an act of will, treat you as a means (though not necessarily merely as a means) in furtherance of achieving an end (specified in the function’s telos—i.e., the end at which the function is normatively directed). So, for example, if I’ve hired you to paint a house, then you have the active function of doing so. Unlike a passive function, an active function requires uptake on your part, as its bearer. This is because a passive function operates by effectively exploiting your conduct as is in furtherance of that function’s end, whereas an active function operates by guiding your conduct in furtherance of that function’s end. The active function itself, though, does not physically force or otherwise compel you to comply (though of course the functions might be paired with coercive threats to motivate compliance). Instead, it normatively guides your conduct by providing instructions which in principle you have the power to heed or ignore. Inasmuch, I typically cannot cogently assign to you an active function unless you at least ostensibly agree to act accordingly. When you adopt a function, it provides you with instructions meant to promote the function’s specified end. You might, of course, have a variety of practical reasons for complying with those instructions. Doing so might be necessary to get paid, to keep your job, to earn a favor, or even just to have fun. But independent of those reasons, the function is supposed to be a source of practical reasons in and of itself. This fact is sometimes expressed colloquially when we hear it said that you’re supposed to do thus-and-so “because it is your job.” I will suggest in what follows that we can recharacterize active functions in terms of protected reasons. That is, I assign to you and you accept an active function just in case I solicit from you, and you confer upon me, practical authority imposing upon you a particular end and (possibly) a particular means by which to achieve that end, while simultaneously delimiting deliberative freedom by excluding some competing considerations from the balance of reasons. (As we’ll see in the next chapter, this protected reason will ultimately license attributing a practical purpose to the executor, where the content of that purpose is determined by the deliberator’s motivating reasons.)

1.1.2 Authority in an Interpersonal Division of Agential Labor We now have some grip on what it means for a deliberator and an executor to establish an interpersonal division of agential labor in which the deliberator outsources to the executor some of the executory functions of the deliberator’s

that collective acceptance. Unlike a status role, however, a normative agentive function does not require collective acceptance of a status function; more importantly, normative agentive functions lack the sort of transformative power in social transactions definitive of status roles.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

   

27

rational agency as they pertain to the deliberator’s end ϕ. They do so when the deliberator assigns to the executor—and the executor accepts—the active function of enacting the deliberator’s instructions regarding ϕ. And they do this when the deliberator solicits from the executor, and the executor confers upon the deliberator, the practical authority to demand that the executor heed the deliberator’s instructions regarding ϕ. What does it mean, though, to say that I have practical authority over you? Often, you decide whether to do some action ϕ by considering the reasons for and against it. Things are different, though, when you believe that someone else is authorized to tell you what to do with respect to ϕ. The reason you take there to be to do ϕ in such a case derives not from the merits of ϕ itself, but from the fact that you take the authority to have a practical claim against you that you do ϕ. H. L. A. Hart points out that regardless of how widely authorities differ in what they require of us, they all provide us with the same practical reason for compliance: the very fact that the authority has a practical claim against us that we so act.⁶ In addition, if you take someone to have authority over you with respect to ϕ, their practical claim against you that you do ϕ provides you with what you take to be a reason to exclude certain competing first-order considerations from deliberation pertaining to ϕ. (The authoritative claim does not exclude all competing considerations, however; the zone of exclusion varies with the command and the context in which it is given.⁷) As already discussed, the combination of a firstorder reason to act in accordance with the authoritative claim against you, and a second-order reason to exclude from deliberation certain competing first-order reasons, yields a protected reason.⁸ So, when you take me to have practical authority over you with respect to ϕ, you treat my practical claim against you pertaining to ϕ as a protected reason to act accordingly. As far as you’re concerned, my decision pertaining to ϕ settles the matter for you, insofar as it is “your job” to do as I say when it comes to ϕ. Deference to authority, then, requires doing what the authority commands precisely because the authority commands it. This account might make it seem as if you cannot even think about whether or why you should do ϕ. But this is not so. The protected status of the reason to do ϕ does not prohibit you from thinking about ϕ; rather, it prohibits you from deliberating about ϕ. The difference is that deliberating about reasons for action, as opposed to merely thinking about the reasons for action, has necessarily a practical orientation in that its point is to

⁶ (Hart, 1990, p. 101). For helpful discussion see (Shapiro, 2002), (Owens, 2008), (Westlund, 2011). ⁷ For helpful discussion, see (Shapiro, 2002, pp. 406–407), (Owens, 2008). ⁸ See (Raz, 1977), (Raz, 1990, pp. 35–84). Raz develops the concept of protected reasons to account for the nature of authority in general, rather than for the authority of future-directed decisions. (Hinchman, 2003) defends the view that decisions yield exclusionary (what he calls “pre-emptive”) reasons. See also (Sciaraffa, 2009, p. 248) and (Ferrero, 2010, p. 8). See (Raz, 1977), (Raz, 1990, pp. 35–84).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

28

, ,  

determine how we are going to act. Thus, a decision prohibits further deliberation (absent new relevant evidence). Scott Shapiro puts the point plainly: “To deliberate is not simply to engage in the thought process of weighing pros and cons. Rather, deliberation is essentially action-guiding. One who deliberates does so with the aim of forming an intention to act on the results of that deliberation.”⁹ To be clear, though, someone you believe to have authority over you needn’t actually issue a command that you do ϕ in order for you to have what you take to be a protected reason to do ϕ. It is enough if you believe that, in virtue of the person’s authority over you, that she has a practical claim against you that you do ϕ. There is, of course, a vast literature on what constitutes legitimate authority— which includes whether such authority is possible, especially in the context of whether governments are justified in compelling its citizens to obey laws. But here I limit myself to practical authority between individuals. For example, one way for an agent to gain practical authority over you is by accepting a promise you made; the reasons to fulfill such a promise are in general protected.¹⁰ So, if you promise to me that you will do ϕ, then I have practical authority over you with respect to ϕ by deciding whether or not to hold you to that promise. If I decide to hold you to that promise, then I have a practical claim against you that you do ϕ, where that claim provides you with a protected practical reason to do ϕ.¹¹ The claim that I have practical authority over you when you, for example, promise to take me to the airport, might seem aggrandizing, insofar as authority might seem to imply a kind of hierarchical command structure. But my use of ‘authority’ as a concept is more capacious in that it characterizes perfectly mundane relationships, such as those established by way of ordinary promises. The claim that I have authority over you that you take me to the airport just means that you have a protected reason to take me to the airport and only I have the standing to release you from that protected reason. What happens if you promise to do something immoral, and I accept that promise, and we both mistakenly think that this promise is binding? I will argue that even though in such a case you lack moral authority over me that I commit the wrongful act, you nonetheless retain practical authority over me that I do so. I will address this issue in section 3.1.4 (chapter 3).

1.2 Autonomy in Divisions of Agential Labor Where does this leave us? We establish a division of agential labor, in which I outsource to you the executory function of my rational agency as it relates to my deliberations pertaining to ϕ, when I assign to you and you accept the active ⁹ (Shapiro, 2002). See also (Moran, 2001, pp. 94–95). ¹⁰ See (Raz, 1986, pp. 35–37). ¹¹ See (Owens, 2008), (2012) for an extended discussion of promise-making as conferring authority.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

   

29

function of enacting my instructions regarding ϕ. And we do this when I solicit from you, and you confer upon me, practical authority to demand that you heed my instructions pertaining to ϕ, which, in turn, yields a protected reason for you to act accordingly. In this respect, the protected reasons characteristic of authority ground a division of agential labor. One might worry that no matter what sort of arrangement we make, my decisions about what you are to do cannot “settle the matter” for you, since you ineluctably retain control over what you are going to do. After all, establishing a division of agential labor in no way implies action at a distance, the literal sharing of mental states, or transferring physical control over conduct. This means it is psychologically up to you whether you are going to do as I instruct. The fact that you are capable of reviewing the decisions I make “for you”—and that you are capable of acting contrary to those decisions—might seem to suggest that I cannot literally function as your deliberator in a division of agential labor. But this worry is misplaced. To see why, it is helpful to repair to intrapersonal diachronic decision making. Suppose that yesterday I decided that, today, I will go for a run after work. That decision has the function of “settling the matter” for me as to what I should do after work today. This is despite the fact that I retain control over what I am going to do today. Diachronic decision-making typically does not function by causally limiting future options (Odysseus and the Sirens aside). Yet my decision yesterday still has the function of “settling the matter” for me today, in that the decision normatively limits my options. The fact that my future self retains control over what he is going to do does not mean that I cannot make decisions for my future self; likewise, the fact that I retain control over what I am going to do does not mean that you cannot make decisions for me. The issue can be put differently. It might be thought that in order to make decisions for someone, you need to have control over what they’re going to do. But if this is correct, then you cannot make decisions for your future self, since at the time you decide, you typically have no direct control over what your future self will do. This suggests that direct control is not necessary for decision making in general. The absence of interpersonal control, then, doesn’t undermine the possibility of interpersonal decision-making. But why believe that intrapersonal diachronic decision-making operates in such a way—that is, without causally compelling a future deliberator? Luca Ferrero answers this question. Suppose I settle a matter by deciding right now to do ϕ immediately. In his words, “[b]arring the interference of akrasia or paralysis, the decision marks the closing of the deliberation and the immediate transition to action.”¹² But things are different when we make future-directed decisions—a decision to do ϕ at some point other than just after the moment the decision is

¹² (Ferrero, 2010, p. 12). He cites Rundle (Rundle, 1997, p. 202).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

30

, ,  

made. Such decisions lack the executive force of present-directed decisions, in that there is no immediate transition to action following the decision. On Ferrero’s view, making decisions about the future is not a means of exerting distal physical control over your future self. Rather, when you decide now that you will do ϕ one month from now, which you subsequently do in compliance with the decision you made, you are at the time of conduct in control of your actions in that the decision from one month ago does not bypass your ability to exercise choice at the time of action.¹³ Neither is your future-directed decision an attempt to manipulate your own future conduct by coercing your future-self into adopting a particular course of action. That manipulative strategy is characteristic of pre-commitments which function by distally altering the decision-relevant circumstances relative to the status quo ante with the aim of compelling a particular course of action.¹⁴ Rather, your future-directed decision to do ϕ itself exerts a practical authority over your future conduct, in that once the time comes, you normally needn’t be guided by any other considerations pertaining to ϕ other than your past decision to do ϕ. As Richard Moran puts it, “When I deliberate about something, the conclusion of my deliberation settles the question for me [ . . . ] the aim and conclusion is the binding of oneself to a certain course of action (or proposition), not the production of a state of mind that I might then treat as (further) empirical evidence about how I should proceed.”¹⁵ Conferring authority upon our past self in this way not only secures a host of practical advantages, but also exemplifies the kind of diachronic self-governance characteristic of agency. In virtue of what, then, do future-directed decisions have practical authority over our conduct? One possibility is that the decisions themselves generate reasons.¹⁶ Consequently, at the time of action, you have a reason to do as you decided. But if simply adopting an end yields a reason to pursue it, then that means we can create reasons ex nihilo to do the unreasonable. It is implausible to think that we can “normatively bootstrap” our ends in this way.¹⁷ But, on Ferrero’s view, my future-directed decision to do ϕ, which I made at t₀, does not generate a new reason to do ϕ at t₁ over and above the reasons that there otherwise are to do ϕ. Instead, the earlier decision serves as a stand-in, or a surrogate, by which I determine whether to do ϕ now. Put differently, I defer to my earlier self in determining whether I should do ϕ. By doing so, I decline to deliberate anew whether to do ϕ. ¹³ See (Velleman, 1997, pp. 45–46), (Anderson, 1996, p. 542), (Hinchman, 2003, p. 40), (Ferrero, 2010). ¹⁴ On non-manipulation, see (Pink, 1996, pp. 6, 114, 269). ¹⁵ (Moran, 2001, pp. 94–95). ¹⁶ For an alternative strategy, see (Broome, 2001). But see Ferrero’s critical review of that strategy in (Ferrero, 2010). ¹⁷ See (Bratman, 1987, pp. 24–27) for the initial formulation of the bootstrapping objection. See also: (Broome, 2001), (Holton, 2004), (Broome, 2004), (Broome, 2005), (Setiya, 2007), (Cullity, 2008), (Setiya, 2014).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

   

31

But what does it mean, precisely, to “delegate” deliberation to my earlier self, or to “defer” to my earlier decision? Suppose at t₀ I decide to do ϕ at t₁. For me, at t₁, to defer to the decision I made at t₀, means treating that decision as authoritative in the sense that the decision at t₀ yields a “maximally protected” reason at t₁ to do ϕ. The zone of exclusion can vary; a maximally protected reason excludes all firstorder reasons pertaining to ϕ except for those specified. This leaves me only with my protected reason to do ϕ: I ought to do ϕ because that’s what I decided previously. Thus, my earlier decision is wholly decisive with respect to what I should do now—it authoritatively settles the matter for me. But why believe that my earlier decision serves as a protected reason for me to do ϕ? Absent an appreciable improvement in my deliberative circumstances, my earlier decision serves as a substitute for a decision now, in the following sense. I can reasonably expect a convergence in the conclusions I reached at t₀ with the conclusions I would reach if I were to deliberate anew at t₁. Given such convergence, complying with my original decision will mean doing what I would have decided to do if I had considered the matter anew. I thereby have a reason to act in accordance with what I decided earlier, since that just is what I would have decided to do now.¹⁸ The protected status of this reason effectively puts me in the position I would be in if I had just decided to do φ at that very moment. I will as a result just automatically do ϕ, without re-considering what to do. Deferring to my earlier self—to the decision I made—promotes epistemically virtuous deliberation. Such authority is warranted in cases where my deliberative circumstances at t₀ meet some minimal epistemic standard (e.g., I am not hallucinating, manic, suffering from a panic attack, etc.) and those deliberative circumstances do not appreciably improve between t₀ and t₁. Provided these conditions are met, deferring to my decision at t₀ enables me to avoid decision-making when I know that at t₁ I will be unable to deliberate efficiently or reliably. This explains why an improvement in my deliberative circumstances defeats the authority of my decision at t₀; given such an improvement, deference would worse-achieve the aim of deliberating virtuously. Notice that on this account decisions are not bootstrapping. This is because my decision at t₀ to do ϕ at t₁ does not add to my stock of reasons; instead, the same reasons that justified my earlier decision to do ϕ at t₁, are precisely what justify doing ϕ at t₁. Delegating authority to my previous self doesn’t yield a new reason to do ϕ; it instead functions, in Ferrero’s words, as an anaphoric device with respect to justification.¹⁹ And like any protected reasons, the reasons I have to do what I decided earlier are defeasible. The excluded first-order reasons cannot ¹⁸ In addition, I must regard the decision as my own insofar as it reflects my practical standpoint at t₁. For more on this condition, see (Morton, 2013). ¹⁹ (Ferrero, 2010, p. 13).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

32

, ,  

defeat a protected reason I have; but if I come to doubt the justification for excluding those reasons, then that can serve as a defeater. I might come to harbor such doubts if I suspect that I am misremembering the decision I reached, or if I come to believe that my deliberations were flawed. Suppose now that at t₀ after due deliberation I settle on exercising at the gym tomorrow at t₁. When the time comes, I needn’t deliberate again absent new evidence relevant to my decision. Nonetheless, my decision is still reversible, from the time I made it to the time it comes to implement it. That is, though it would be irrational for me to do so, I could just ignore the decision I made at t₀ by instead deciding on a whim, at t₁, to ignore the protected reasons I have by deliberating anew. (Assuming I took myself to have decisive reason for doing φ—which is why I decided to do φ—I would be acting contrary to the reasons that I take there to be should I subsequently choose not to do φ, absent any relevant change in circumstances. And this seems like a paradigmatic case of irrationality.) The fact that the option of acting irrationally in this way remains open to me subsequent to deciding on a course of action at t₀ does not vitiate the peremptory force of that decision. In Ferrero’s words, “future-directed decisions are not executive in the way of present-directed ones (if they were, they would either exercise action-atdistance or trigger mere causal, time-delaying mechanisms).”²⁰ Accordingly, “[i]n deciding to φ, rather than setting up some other mechanism of manipulative distal self-control, the agent exposes herself to the risk of a future repudiation of her decision. But this is the price that must be paid to make diachronic autonomy possible.”²¹ If I defer to the decision I made yesterday about what to do today, I count as having acted on the decision I made yesterday, even though at any point I might have chosen to “disobey” my past self. Return now to cases of interpersonal decision making. The practical authority I have over you might have nothing to do with epistemic efficiency.²² There might instead be any number of reasons we establish an interpersonal division of agential labor. Either way, if I agree to function as your deliberator and you agree to function as my executor, and you thereby defer to me, I count as making decisions for you even though at any point you might choose to “disobey” my instructions. To paraphrase Ferrero, this is the price that must be paid to exercise our interpersonal normative powers.²³ The upshot then, is this: Because distal causal control is unnecessary for intrapersonal diachronic decision-making, it isn’t necessary for interpersonal decision-making either in a division of agential labor. The fact that executors in

²⁰ (Ferrero, 2010, p. 12). ²¹ (Ferrero, 2010, p. 21). ²² For more on the connection between intrapersonal diachronic decision making and interpersonal activity, see for example (Rovane, 1997, p. 146). ²³ Indeed, were it not for that price, the Razian service conception of authority would be unable to reconcile the commands of law with our decision to freely abide by those commands. See (Raz, 1979).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

   

33

a division of agential labor are capable at any point of disobeying the commands of their deliberators does not vitiate their status as such.²⁴

1.3 Conclusion I developed in this chapter machinery enabling us to characterize a normative capacity each of us has: to establish divisions of agential labor with another person. I argued that is possible for you to “outsource” the executory functions of your practical agency to me. Return to the example in which you ask me to take you to the airport tomorrow; I make that promise and you accept it. The promise yields what each of us takes to be a protected reason for me to take you to the airport. That is, as far as the two of us are concerned, your practical claim against me that I take you to the airport normatively “settles the matter” with respect to whether I should take you to the airport. In this way, the promise that I made and you accepted establishes an agential division of labor in which you function as the deliberator and I function as the executor. When we establish this division of agential labor, we thereby change the object at which your ends and my actions are teleologically directed. Normally my conduct has the function of enacting my ends, and your ends guide your own conduct. But when I function as an executor for you and you function as a deliberator for me, my conduct also has the function of enacting your ends, and your ends also have the function of guiding my conduct. My conduct and your ends dovetail in this way by virtue of the protected status of the reason that you confer upon me and that I accept—or so I argued. I then addressed whether it is possible for an individual to function as an executor given that he has the ability to “disobey” his deliberator’s instructions at any point in time. It seems that he cannot be an autonomous agent if he is truly functioning as an executor. But this is not so. The possibility of intrapersonal diachronic decision-making explains why. In exercising diachronic agency, an agent confers authority on her past self when it comes to deciding how to act now. Her past self, in effect, issues orders to her present self, where those orders confer a Razian protected reason to comply, thereby normatively (rather than causally) binding her future self. In this way, her past self has practical authority over the present self—despite that her present self is fully capable of gratuitously

²⁴ I believe that the strategy I’ve adopted in the section can also help dissolve what has come to be known as the “control problem” in group agency (see (Pettit, 2007), (Searle, 2010, pp. 50–55), (List & Pettit, 2013, pp. 160–163), (Roth, 2014)). According to this problem, theories of group agency must reconcile the fact that a group is an agent in control of its actions, with the fact that such a group comprises individuals who carry out the group’s actions, where those individuals retain control over what they do. But as I’ve argued, one can have normative control over another even if she does not have causal control over him.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

34

, ,  

“disobeying” her past self. If the ability to “disobey” her past self does not vitiate the authority that her past self has over her, then the same goes for an executor and a deliberator in a division of agential labor; the executor’s ability to “disobey” the deliberator does not vitiate the deliberator’s authority over the executor. In the next chapter I argue that the deliberator in a division of agential labor can be accountable in a special way for what the executor does irrespective of the deliberator’s causal influence on the executor’s conduct. But before moving on, one might raise the following worry about the account I have developed. The idea that your practical claim against me ought to settle the matter for me—in that I ought to take myself to have limited deliberative say in the matter—might seem to overstate the role that the promise can be expected to play in my motivational economy. Suppose that I planned on going in the direction of the airport anyway, and if those plans changed, I would renege on my promise to take you to the airport. Or suppose I decide to go through with the promise I made only after deliberating about the merits of doing so, in that I re-consider the costs to me of undertaking the drive, whether I will earn any good will by taking you to the airport, whether I feel like doing it, and so on. In these cases, I seem to be violating the deliberative injunctions imposed by my protected reasons, even if I ultimately decide to fulfill the promise I made. Nonetheless, so long as the promise provides a protected reason to drive you to the airport, then you have the practical authority to settle the matter for me, and to exclude further deliberation about the matter. If I deny that, then I have not made a sincere promise (a possibility I consider in section 4.1.1 (chapter 4). So long as we both believe that you have a practical claim against me where that claim yields a protected practical reason for me to drive you to the airport, then the functional relationship characterizing the division of agential labor between us remains in place, even if my commitment to the protected status of that reason is less than perfect.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

2 Authority-Based Accountability So far, I have explained how we establish interpersonal divisions of agential labor. I outsource to you the executory function of my rational agency as it relates to my deliberations pertaining to ϕ, when I confer upon you, and you accept, the active function of enacting my instructions regarding ϕ. I do that when you confer upon me and I accept practical authority yielding a protected reason for you to do what I say pertaining to ϕ. Inasmuch, I count as a deliberator and you count as an executor. In developing this account, I’ve focused on a morally benign example of promising. Recall, though, that my goal is to show that when the executor does something morally wrong in accordance with her role in a division of agential labor, the deliberator is accountable for the fact that what the executor does is wrongful (for reasons apart from the deliberator’s causal influence on what the executor does). We need, then, a different example—one involving moral wrongdoing. I will use the example I presented in the introduction: Olympic Sabotage Contender is vying with Victim for first place in an upcoming marathon. Contender consequently hires Goon to maim Victim in a way that will put her out of the running. Goon promises to do so and Contender accepts that promise. Obviously, this is an immoral promise—i.e., a promise to do something morally wrongful. Do immoral promises count as promises? If so, do they generate protected reasons? In section 3.1.4 (chapter 3, p. 81), I argue that immoral promises yield protected practical reasons, which is enough to establish a division of agential labor between the parties to the immoral promise. But for now, I will simply stipulate that Goon takes himself to have a protected practical reason to commit the assault, and that Contender takes herself to have the practical authority to demand that Goon commit the assault, which will be enough to show that Contender bears a special kind of accountability for what Goon does. What do I mean by “accountability”? Gary Watson famously distinguished between two senses of “responsibility”.¹ An agent is responsible for conduct in the attributability-sense just in case the conduct is properly attributed to her in that it

¹ (Watson, 1996).

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0003

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

36

, ,  

reflects morally relevant characteristics, such as her virtues and vices. She counts as responsible in the attributability-sense for the conduct in question even if she is not at fault for coming to possess the characteristics that the conduct reflects. So, for example, if the characteristics are the product of childhood trauma, of a bigoted upbringing, of psychopathy, and so on, the agent is nonetheless responsible in the attributability-sense for the conduct reflecting those characteristics. We might still aptly blame such an agent in the sense that that we might negatively evaluate the character-traits that her conduct evinces; but this kind of blame does not imply that we hold the agent accountable for what she has done. For example, the agent’s vicious character-traits do not warrant subjecting her to the kind of adverse treatment characterizing retributive justice. That kind of treatment might be unfair if it is through no fault of her own that she harbors the poor charactertraits that her wrongful conduct reveals.² But when an agent can indeed be properly held to account for what she has done, she is responsible in the accountability-sense. To hold an agent accountable is to blame her in a more robust way. Not only do we negatively evaluate the poor character-traits her conduct reveals, but we also take a particular stance toward her: we regard her as a proper subject of potentially adverse treatment because she violated a moral norm. We might fairly express negative reactive attitudes, including anger, resentment, and indignation, and we might inflict upon her the kind of deprivations characterizing corrective justice in general and, possibly, retributive justice specifically.³ The claims I make about accountability should be understood not merely in terms of the attributability-sense of responsibility, but moreover in terms of the accountability-sense. Clearly, then, both Contender and Goon are severally and fully accountable for the ensuing assault, should it occur; Contender is accountable partly in virtue of the causal role she plays in intentionally motivating Goon to commit the assault. But in this chapter, I will argue that Contender is also accountable on different grounds: in virtue of the role she plays as the deliberator and the role Goon plays as the executor. Here is the argument in brief. When Goon commits the assault, Victim is entitled to an explanation of what happened. She has an interest in ascertaining the purpose Goon had for acting in the way that he did. But I will argue that because the reasons Goon took there to be in favor of the assault are protected, Victim must repair to the reasons Contender took there to be in favor of the assault, in order to divine Goon’s purpose. There is, then, a sense in which Victim needs to “look through” Goon’s conduct to Contender’s motivating reasons in order to evaluate what Goon has done. That is, Contender’s aims constitutively determine the content of the protected reasons she confers upon Goon. Contender

² (Watson, 1996, pp. 273, 281).

³ (Watson, 1996, p. 262).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

37

is thereby accountable for a wrong-making feature of what Goon does: the purpose of Goon’s conduct. We advert to that purpose in evaluating what Goon has done. Inasmuch, what Goon has done counts as wrongful partly because of the purpose which Contender furnishes. That is, Contender bears authority-based accountability for what Goon does. But before I lay out this argument in detail, I first need to address the type of reason morally relevant to evaluating conduct. In section 2.1, I argue that morally evaluating wrongful conduct requires ascertaining the wrongdoer’s motivating reasons—i.e., the reasons (if any) that the wrongdoer took there to be in favor of the conduct in question. But since, in a division of agential labor, the executor has the function of enacting the practical reasons that the deliberator takes there to be in favor of an action (or inaction), we must repair to the deliberator’s motivating reasons in morally evaluating the executor’s conduct. The result is that the deliberator will bear authority-based accountability for what the executor does. The argument for authority-based accountability might seem to share important similarities with accounts of shared action that others have developed in the literature. Focusing on the sophisticated accounts that Abraham Roth and Christopher Kutz each develop, I argue in section 2.2 that the similarity between mine and theirs is largely superficial. Nonetheless, their accounts serve as helpful points of departure for better understanding the argument for authority-based accountability. Before ending the chapter, I anticipate and respond in section 2.3 to counterexamples and counterarguments.

2.1 Argument for Authority-Based Accountability I will present the argument for authority-based accountability in five steps. First, in section 2.1.2 I argue that when morally evaluating an agent’s conduct there are cases in which it’s not just her motivating reasons that matter. If she is an executor in a division of agential labor, then the motivating reasons belonging to the deliberator are relevant to evaluating what the executor has done. If the deliberator’s motivating reasons are wrongful, she can thereby be accountable for a wrong-making feature of what the executor does. The deliberator thereby bears authority-based accountability for what the executor does. Second, in section 2.1.2, I argue that a deliberator bears authority-based accountability for what an executor does only when the deliberator confers protected reasons on the executor. I make this point by considering cases in which a putative deliberator issues to a putative executor an unprotected reason to do as the deliberator says. And third, in section 2.1.3, I argue that executors in a division of agential labor can also at times be inculpated in the authority-based accountability of their deliberators. Fourth, in section 2.1.4, I apply authority-based accountability to harms

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

38

, ,  

cooperatively committed by more than two individuals. I end, in section 2.1.5, by contrasting authority-based accountability with “ordinary” accountability.

2.1.1 Whose Motivating Reasons? We morally evaluate actions in part by adverting to the actor’s reasons. For example, whether an act counts as a murder or, alternatively, a manslaughter depends in part on the wrongdoer’s reasons. But what kinds of reasons are relevant in morally evaluating wrongful conduct? Here I will argue that the relevant reasons include (but are not limited to) the practical reasons that the wrongdoer took there to be. These are the (putative) facts that the wrongdoer takes to favor her actions.⁴ I will call these “motivating reasons”. I argue that the deliberator’s motivating reasons can partly determine how we evaluate the wrongful conduct of the executor. (To be clear, the deliberator’s motivating reasons serve as a basis for evaluating what the executor does, rather than the executor herself. That is, the deliberator is accountable for features of the executor’s conduct, rather than for features of the executor as an agent.) Return, again, to Olympic Sabotage. Victim is entitled to demand that Goon divulge his motivating reasons. The reasons Goon took there to be in favor of maiming her, reveal or reflect the accountability-relevant aspects of Goon’s agency, which in turn affects how we morally evaluate the harm he inflicted. There are various ways that Goon’s motivating reasons might be relevant to morally evaluating the harm that he inflicts. It might matter morally whether the harm was intended or whether it was merely a foreseeable side effect of some other goal; or whether the harm was a means to an end or an end in and of itself; or whether the harm was committed in furtherance of a morally worthwhile goal or in furtherance of a malign goal; and so on. Victim cannot morally evaluate what has been done to her without addressing the practical reasons Goon took there to be in favor of committing the assault. That is, Victim cannot morally evaluate what has been done to her without addressing Goon’s motivating reasons. Such reasons reveal the role that the victim’s rights and welfare played (if they played any role at all) in the deliberation yielding Goon’s conclusion that the conduct in question is acceptable. The role that Victim’s rights and welfare played in these deliberations partly determines our moral evaluation of how Victim was treated. To the extent that the deliberations fail to respect Victim’s rights or otherwise wrongfully ignores Victim’s welfare, the conduct the deliberation yields mistreats Victim. (What if intentions—and hence motivating reasons are deontically irrelevant? I consider this possibility in section 2.3.1.)

⁴ For helpful discussion of this category of reason see (Darwall, 2003, pp. 442–443).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

39

I will argue that Contender’s motivating reasons serve as a basis for evaluating Goon’s conduct. More generally, the point is to show that a deliberator’s motivating reasons can serve as a basis for evaluating the executor’s wrongful conduct. Consider a paradigmatic case of wrongdoing. Suppose an individual, acting on her own, commits a theft motivated by greed. In such cases we advert to her motivating reasons—the reasons she took there to be—in evaluating her conduct. Why? The purpose of practical reasoning is, after all, to normatively guide conduct by determining what ends to pursue and how to pursue them. Concomitantly, the purpose of conduct (or at least intentional conduct) is to enact the practical reasons the actor takes there to be (where enacting a reason is just to do as the reason prescribes). One reason why conduct morally matters to us is that a moral assessment of agency matters to us. The conduct specifies whose agency matters by virtue of its function. Her conduct has the function of enacting her motivating reasons; we thus advert to her motivating reasons in evaluating that conduct. Return again to Olympic Sabotage. Clearly, Goon’s conduct had the function of enacting some of his own motivating reasons, of which there might be a variety. Goon might have maimed Victim partly because he promised to do so, and partly because he enjoyed it, and partly because he was paid to do so, and so on. Suppose, though, that like most people Goon thinks that when we make a promise, we make a commitment to the person to whom we make the promise. This gives us a protected reason to fulfill the promise irrespective of whether we benefit from doing so. This isn’t to say the commitment motivates Goon. Rather, the point is that Goon cannot felicitously deny that he has made a commitment—one which ostensibly confers a protected reason to act accordingly. As a result, when Contender accepts Goon’s promise, doing so is supposed to settle the matter for Goon as to whether he is to commit the assault. This is because Goon confers authority upon Contender by promising to commit the assault. Recall from the previous chapter that vesting authority by making a promise is one way to “outsource” the executory functions of our rational agency; if I make a promise to you that I will do ϕ if you want me to, and you accept that promise, then we are both supposed to treat your deliberations pertaining to ϕ as “settling the matter” with respect to what I should do. In such a case, you attribute to me the active function of enacting the practical reason you take there to be pertaining to ϕ. By accepting that function, we thereby change the object at which your practical reasoning pertaining to ϕ is teleologically directed. Its object is no longer your conduct, but rather mine. The practical reasons you take there to be have the function of guiding my conduct, and my conduct has the function of acting on the practical reasons you take there to be. Recall that this is how we establish an interpersonal division of agential labor. Where does this leave us? We morally evaluate Goon’s conduct in part by evaluating the motivating reasons that the conduct had the function of enacting. Because Goon’s conduct had the function of enacting Contender’s motivating

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

40

, ,  

reasons, such reasons are among those serving as a basis by which Goon’s conduct is morally evaluated. But if the relevant motivating reasons are those that the wrongful conduct in question had the function of enacting, then they will include not only Goon’s reasons, but Contender’s as well. This is because Contender has the function of specifying whether Goon is to commit the assault, and Goon has the function of acting accordingly, given the division of agential labor the two of them established. Recall that these dovetailing functions arise from the promise which Goon made and which Contender accepted; by accepting the promise, Contender thereby issues—and Goon thereby has—a protected practical reason to commit the assault. Recall also that the function of intentional conduct in general is to enact the practical reasons we take there to be. So, if Contender gives Goon the go-ahead (or otherwise refrains from instructing Goon to stand down), and Goon subsequently commits the assault, Goon’s conduct will have had the purpose of enacting the practical reasons Contender takes there to be in favor of committing the assault. If we evaluate conduct by addressing the motivating reasons that the conduct has the function of enacting, then we ought to evaluate Goon’s conduct by adverting to Contender’s motivating reasons. That is, we ought to evaluate Goon’s conduct by adverting to the practical reasons Contender took there to be in favor of having Victim assaulted. After all, Goon and Contender have established a division of agential labor in which Contender counts as the deliberator and Goon counts as the executor. Victim is entitled to morally evaluate the role that her rights and welfare played in the deliberation yielding the conclusion that the conduct in question is acceptable. As the deliberator, Contender’s reasons are ineliminably part of that deliberation. So, if we evaluate conduct by addressing the motivating reasons that the conduct has the function of enacting, and if Contender’s motivating reasons are problematic, then Contender has made it so that Goon’s conduct is morally problematic as well. This point can be put differently; because Goon’s motivating reasons include protected ones, which refer anaphorically to Contender’s motivating reasons, it turns out that evaluating Goon’s conduct requires evaluating Contender’s motivating reasons. By doing so, we determine the purpose of Goon’s conduct. Goon’s purpose, unlike his motivating reasons, might remain introspectively opaque to him; after all, it’s Contender—not Goon—that fixes the content of the purpose Goon has. Since the reasons Contender took there to be—the motivating reasons determining the purpose of Goon’s conduct—are morally problematic, Contender is therefore accountable for a wrong-making feature of Goon’s conduct: the wrongful purpose for which Goon acts. So, for example, suppose Goon doesn’t know that Victim and Contender are competing for the same medal. Rather, he is just given a face and a name, and instructed to maim the target. Nonetheless, Contender has made it so that Goon

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

41

has the purpose of maiming Contender’s rival. That is, Contender’s motivating reasons furnish the relevant description of Goon’s purpose. In this way, a deliberator can make what the executor does morally better or worse by furnishing different aims constitutively determining the purpose of what the executor does. This deliberator is thus accountable for a wrong-making feature of the executor’s conduct. To be clear, the claim is not that Contender is accountable in virtue of having influenced Goon’s motivations (though that too might be a basis for accountability). Rather, the basis of Contender’s accountability for Goon’s conduct is constitutive rather than just causal. Goon’s conduct has the function of enacting the practical reasons Contender takes there to be; so, by adopting morally problematic motivating reasons, Contender, from afar, constitutively determines the purpose of Goon’s conduct. Insofar as that morally affects an assessment of Goon’s conduct, Contender is on the hook for that difference she makes. So, in morally evaluating the conduct in question, the agency that matters belongs not only to those we metaphysically attribute the conduct to, but also to those whose ends the conduct has the function of enacting. If Contender is accountable for a wrong-making feature of the conduct in this way, then Contender (in addition to Goon) is a proper object of Victim’s reactive attitudes in response to the wrong she has suffered. In this sense, Victim “looks through” Goon to Contender in determining the reasons for which she was maimed. The result is that when Goon fulfills an authoritative claim Contender has against him, Contender can be accountable for a wrong-making feature of the harm Goon inflicts for reasons apart from the causal role Contender plays in that harm. The relationship between Goon and Contender specifically, or between a deliberator and an executor more generally, can be depicted in the diagram (Fig. 2.1). The motivating reasons the deliberator has in favor of ϕ

The deliberator adopts the function of deciding what the executor should do and why he should do it. The executor adopts the function of acting accordingly, yielding a protected reasonto so act.

The protected reasons the executor has in favor of ϕ

The executor’s protected reason licenses attributing to him a purpose, the content of which is determined by the deliberator’s motivating reasons.

The purpose the executor has with respect to ϕ

To be clear: when describing the executor’s purpose, it is not enough to say that he is purposed with performing some act, described extensionally. The description of the executor’s purpose makes ineliminable reference to the deliberator’s

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

42

, ,  

motivating reasons.⁵ In this way, the executor’s purpose co-varies with the deliberator’s motivating reasons. What underwrites this dynamic relationship between the deliberator’s motivating reasons and the executor’s purpose? The answer, again, lies in the division of agential labor between the deliberator and the executor. A result is that the deliberator can furnish an impermissible purpose for otherwise permissible conduct, thereby changing the deontic status of that conduct. Consider the following variation of a classical philosophical case: Strategic Bomber A captain in the air force orders a pilot to drop a bomb on a munitions factory. Doing so will cripple an enemy’s bomb-making capabilities which will help end the unjust war that the enemy is waging. But the ensuing explosion will also collaterally kill dozens of civilians in a nearby village. Given the moral costs and benefits, bombing the factory is morally permissible. But when the captain issues the order, he harbors the inimical intention of murdering the villagers. The pilot does as she is ordered with the permissible intention of destroying the munitions factory. According to what I’ve argued so far, we look to what purpose the captain furnishes for the pilot in evaluating what the pilot has done. Clearly, the captain furnishes the purpose of dropping a bomb on the munitions factory. But that description is incomplete. By hypothesis, the captain wants the civilians in the nearby village killed. Thus, the purpose of what the pilot does, as far as the captain is concerned, is to murder the villagers. The purpose the pilot takes herself to have is different: to destroy the munitions factory, simply. But since the captain is the deliberator, his motivating reasons also determine the purpose of the pilot’s conduct. He is accountable for conferring a wrongful purpose upon the pilot’s actions. If intentions are relevant to moral permissibility, then the captain bears what I call “authority-based” accountability for murder, even though what the pilot did was permissible relative to her own intentions. The moral here is this. Evaluating an executor’s actions requires adverting to the reasons her conduct had the function of enacting—which include the deliberator’s motivating reasons. The deliberator’s motivating reasons serve as a basis for evaluating the executor’s actions. If those motivating reasons are wrongful,

⁵ Recall from the Introduction that, technically, the deliberator furnishes a purpose for the executor’s conduct, rather than the executor himself. This is worth mentioning since the deliberator is accountable for features of the executor’s conduct, rather than for features of the executor as an agent. But for the sake of brevity, I sometime describe the deliberator as furnishing a purpose for the executor, rather than for the executor’s conduct.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

43

then the executor’s actions are wrongful as well. Thus, in such a case, the deliberator is accountable for making the executor’s actions wrongful. This isn’t to say that the deliberator is a but-for cause of the fact that the executor’s actions are wrongful. It might be that absent the deliberator, the executor would have adopted and acted on her own wrongful reasons. Or it might be that absent the deliberator, some other deliberator would have furnished for the executor a wrongful purpose. Rather, the claim I am making is simply that the deliberator is actually accountable for making the executor’s actions wrongful. (I return to this issue in section 2.1.5.) The argument I’ve laid out so far can be generalized as follows. The Argument for Authority-Based Accountability Morally evaluating some conduct ϕ requires morally evaluating the motivating reasons for committing ϕ. Where the motivating reasons are wrongful, ϕ is wrongful as well. The relevant motivating reasons for committing ϕ are those that ϕ had the function of enacting. The motivating reasons that ϕ had the function of enacting will typically belong to those intentionally causing ϕ. This is because the purpose of intentional conduct in general is to enact motivating reasons. But where P1 issues and P2 accepts a protected reason for P2 to do ϕ, they thereby establish a division of agential labor in which, vis-à-vis ϕ, P1 counts as the deliberator and P2 counts as the executor. Hence, P1’s motivating reasons for instructing P2 to do ϕ constitutively determine the purpose of P2’s conduct. If P2 commits ϕ, we must advert not just to P2’s but to P1’s motivating reasons in order to morally evaluate ϕ. This is because the motivating reasons that ϕ had the function of enacting belongs not just to P2 but to P1 as well. So, where P1’s motivating reasons are wrongful, she is accountable for the fact that P2’s actions are wrongful (which in no way indemnifies P2 against accountability for ϕ). That is, P1 bears “authority-based” accountability for what P2 does.

To be clear, authority-based accountability is not a sui generis kind of accountability. Rather, authority-based accountability is a species of ordinary accountability. We might think of ‘accountability’ as a predicate taking as its variables a basis, a subject, and an object. The basis is that in virtue of which the agent is accountable. The subject is the accountable agent. And the object is that for which the agent is accountable. Authority-based accountability differs from ordinary accountability simply in that it specifies a particular subject of, basis for, and object of accountability. The subject of accountability is the deliberator. The basis of accountability is the deliberator’s criticizable decision to furnish the executor with a wrongful purpose—a purpose aiming at enacting the deliberator’s wrongful motivating reasons. The object of accountability is the fact that what the executor does is wrongful.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

44

, ,  

Here is an analogy. Consider, for example, the accountability that an agent bears in cases where she violates duties of easy rescue—such as an agent who gratuitously refuses to call the authorities as she passes the scene of a lifethreatening car-accident. We might introduce a term for this kind of accountability—e.g., ‘aid-based accountability’. This kind of accountability is clearly a species of ordinary accountability. The term simply specifies the subject of accountability (an individual in a position to render aid), the basis for the accountability (a gratuitous failure to render that aid) and the object of accountability (the harm that results from having failed to render the aid). Likewise, authority-based accountability is a species of accountability; the term simply specifies the subject, the basis, and the object of accountability.

2.1.2 Are Protected Reasons Necessary for Authority-Based Accountability? On the account I’ve developed, the protected status of the practical reasons that the deliberator confers on the executor plays an integral role in grounding authority-based accountability. Recall from section 1.1 (chapter 1) that the protected status of the practical reasons is what allows us to say that the executor has the purpose of enacting the deliberator’s motivating reasons. Put differently, the protected status of the practical reasons is what allows us to characterize the executor as an executor and the deliberator as a deliberator. It is instructive, then, to imagine cases in which the reasons I take there to be to do as you say are not protected. Suppose I promise that if you direct me to do ϕ, I will take the very fact that you do so as a reason in and of itself to do ϕ. But suppose in making this promise, I also make it clear that I will not treat your directive as a reason to exclude competing considerations from deliberation. So, I am free to weigh the reason to do as you say against any and all competing reasons. For example, suppose I promise the following: if you tell me to do the dishes, I will take that as a reason to do the dishes, but I’ll weigh that reason against competing considerations, such as whether I feel like doing the dishes. Should I subsequently decline to do as you say on those grounds, I will have nonetheless kept my promise to you. Now take the inverse. Suppose I promise that if you direct me to do ϕ, I will treat that as an exclusionary reason to do ϕ, but not in and of itself as a first-order reason to do ϕ. For example, suppose I promise the following: if you tell me to do the dishes, I will ignore competing considerations, but I won’t treat your directive as a reason to do as you say. As a result, I might end up with no positive reason to do the dishes. I subsequently might decline to wash the dishes, not because there are considerations militating against it, but because there are no considerations in its favor. Yet, in such a case, I will have nonetheless kept my promise to you.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

45

In neither of these examples am I treating your directive that I do ϕ as a protected practical reason to do ϕ. Recall that a protected reason to do ϕ is a combination of a first-order reason to do ϕ and a second-order reason to exclude from deliberation certain competing considerations. In the first example, I meet the first criterion but not the second criterion, and in the second example the reverse is true. The result is that in neither of these cases have we established a division of agential labor in which you count as the deliberator and I count as the executor. Why is that? I have the function of enacting your practical reasons pertaining to ϕ only if I lack, to a substantial degree, the normative freedom to deliberate about whether to do ϕ. Rather, I’m just supposed to do as you say because you say it. By excluding competing practical reasons that would otherwise be normatively accessible to me, and by treating your directive as a practical reason in and of itself in favor of ϕ, we both make it the case that I’m supposed to treat your decisions as decisive. Absent that, we have not established a division of agential labor in which you count as the deliberator and I count as the executor. Suppose, then, that in the first case—where I am permitted to weigh considerations against ϕ—I nonetheless voluntarily elect not to do so, as a result of which I do ϕ. Likewise, suppose that in the second case—where I do not need to treat your directives as a positive reason to follow them—I nonetheless voluntarily elect to do so, as a result of which I do ϕ. In these cases, you bear no authority-based accountability for what I do, because I do not have the purpose of enacting your ends. I might have acted in a way that happens to accord with such a purpose; but in the absence of the requisite protected reasons, I lack such a function. This is because your decisions do not have the function of settling the matter for me. As a result, I do not count as an executor and you do not count as a deliberator. This also explains why you bear no authority-based accountability in cases where I fulfill your wishes absent authorization from you. Suppose I know that you want ϕ done—a wrongful act—but that you cannot do it yourself. I consequently do ϕ because I want to earn your respect. (If you had desired something else, I would have done that instead.) I evaluate what to do not by evaluating whether ϕ is in and of itself choice-worthy, but by simply determining whether you desire ϕ, and then treating your wish as a decisive reason to do ϕ. In this sense, your wish is my command, though you lack even a putative claim against me that I act accordingly. Since you played no role in establishing a functional relationship between my conduct and your desires, you bear no authority-based accountability for what I do.

2.1.3 Implications for Executor Accountability I’ve argued that a deliberator’s motivating reasons constitutively determines the executor’s purpose. Where the deliberator’s motivating reasons are wrongful, so is

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

46

, ,  

the purpose she furnishes, and thus what the executor does, should she comply. The deliberator is thereby accountable for the fact that what the executor does is wrongful. But is the executor at all accountable? I argue that he can be. Consider the following example. Performance Review Owner is in charge of a large private business. He wants a systemic overhaul of the criteria for promoting employees in his firm—criteria which emphasize seniority over merit. Owner develops this new set of criteria and subsequently instructs the department-heads to implement it, which they do. Owner, though, is a sexist who orders the new criteria in order to ensure that men rather than women attain positions of power in the firm. A seniority-based standard will help achieve that aim since the employees who have been there the longest tend to be men; it is only recently that social and legal pressure effectively forced Owner to hire more women. Manager, who works at the firm, is tasked with determining whether Senior, who is up for review, should be promoted. It so happens that Senior is a man and is one of the longer-serving employees at the firm. Based on Owner’s newly adopted promotion-criteria, Senior ought to be promoted. It turns out, though, that Senior has a stellar record; he would have been promoted even given the older, merit-based criteria. Manager promotes Senior. Suppose Manager is aware of Owner’s sexism and of the sexist purpose behind the newly adopted seniority-based promotion criteria—both of which the Manager recognizes as morally reprehensible. He takes solace, though, in the thought that he does nothing wrong in promoting Senior, specifically. “After all,” Manager says to himself, “my attitudes and goals are not sexist—and absent such goals, promoting Senior would be permissible given his meritorious record.” Is Manager correct in his belief that he does not act wrongly in promoting Senior? Let us grant Manager that promoting Senior would be permissible absent sexist aims. According to the argument for authority-based accountability, the deliberator’s motivating reasons constitutively determine the purpose of the executor’s conduct. So, whether Manager likes it or not, the purpose of his conduct, in promoting Senior, is to enact a sexist aim. The fact that those aims aren’t his does not undercut the basis for thinking that in promoting Manager he has the purpose of enacting a sexist aim. So, even if promoting Senior would have been permissible absent the sexist aims, and even though Manager does not harbor such aims, promoting Senior is wrongful (or, more conservatively, it is a pro tanto wrong that might be outweighed or overridden by possible negative consequences of refraining from promoting Senior). The moral here is that in a division of agential labor the purpose that a deliberator furnishes for an executor can affect the deontic status of what the executor does even if the executor’s motivating reasons oppose that purpose.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

47

By furnishing Manager with a wrongful purpose, Owner is accountable for a wrong-making feature of what Manager does, since, in virtue of that purpose, what Manager does counts as a pro tanto wrong. This is just to say that Owner bears authority-based accountability for what Manager does. But what about Manager? Is he accountable? Owner would be unable to furnish Manager with a wrongful purpose absent Manager’s voluntary participation. The result is that Manager is accountable as well—but less so. Though by hypothesis Manager knew he was enacting a sexist aim, he didn’t intend to enact that aim. The purpose with which he was saddled does not converge with his own aims. Though this does not get him off the hook, it makes him less accountable than he would be if the aims converged.⁶ The moral here can be generalized. An executor can, under certain circumstances, be accountable for enabling the deliberator’s authority-based accountability for the wrong that the executor commits. The deliberator furnishes the executor with a wrongful purpose and is thereby accountable for the fact that what the executor does is wrongful; but since the deliberator does this with the executor’s assistance, the executor is potentially accountable as well for that wrong.

2.1.4 Authority-Based Accountability in Groups One might argue that authority-based accountability might seem wholly otiose in cases like Olympic Sabotage, since ordinary accountability—that is, accountability for the foreseeable consequences of your conduct—is enough to put Contender on the hook for what Goon does. But the relevance of authority-based accountability becomes especially apparent in cases of the sort mentioned in the introduction— where many cooperants act together. Consider the following schematical example: The Bandits and the Beans⁷ There is a poor village of one thousand peasants each of whom has exactly one thousand beans. There are also one thousand well-off bandits who intend to steal the villagers’ beans in order to sell them. These bandits cooperate in the following way: each agrees to steal one bean from each villager. They agree that afterward, they will pool together and sell what they’ve stolen. The result is that one thousand villagers have all of their beans stolen; they consequently starve. If an agent’s accountability is limited to the harm she causes then any given bandit bears very little accountability since any given bandit makes only a negligible

⁶ I discuss this issue in further detail in chapter 8.

⁷ This case belongs to (Glover, 1975).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

48

, ,  

difference to the diet of any given villager. But in the example the bandits agree to work together in furtherance of depriving all the villagers of their beans. The result is that each bandit is both a deliberator and executor in a division of agential labor that they’ve established with one another. That is, each bandit has authority over every bandit that he does his part, and each bandit is subject to authority over every bandit that he do his part. The result is that each bandit bears authoritybased accountability for what every other bandit does, which means each bandit is accountable for far more than the harm she causes or could have prevented. This is, presumably, a conclusion we want. The argument for authority-based accountability, then, does work in this case that ordinary, causal accounts of accountability cannot. Suppose you and I are among the bandits. In accordance with the agreement that we have all made, I furnish you with the purpose of stealing one bean from each of a thousand peasants. Even if I am accountable for a wrong-making feature of what you did—i.e., the wrongful purpose for which you acted—I’m still not accountable for much, since, by hypothesis, stealing one bean from each of a thousand peasants doesn’t amount to much. But there are 998 other bandits each of whom respectively stole a bean from the same peasant; I am accountable for a wrong-making feature of what each of them did too. The result is that my accountability adds up to much more than what’s within my causal reach. What the other bandits do count as wrongful because of me; mutatis mutandis for your accountability, and the accountability for each of the other bandits. A skeptic might raise the following worry. By hypothesis, each bandit is a deliberator and an executor in a division of agential labor, which means no single bandit furnishes the sole purpose for what the other bandits do. Indeed, any given bandit is just one of a thousand each of which furnishes for the others a purpose for raiding the village. As a result, it might seem misleading to claim that a given bandit is accountable for furnishing the purpose for which all the other bandits act, when 999 other bandits are furnishing such a purpose as well. If a given bandit’s causal contribution is minimal, it seems that his normative contribution is minimal as well. This in turn would suggest that for any given bandit his authority-based accountability is no more substantial than his accountability for the harm he causes. This criticism, though, rests on a misunderstanding of the purpose that the bandits furnish for one another. The content of the purpose that each bandit furnishes is determined by the practical reasons that that bandit takes there to be in furtherance of raiding the village. One bandit might want the village raided because it will enable him to feed his starving family, while another bandit might want the village raided out of sadistic glee. Suppose though that each bandit has ostensibly the same practical reason: he wants the village raided so that he can sell his share of the beans and profit from it. Even in this case, each bandit harbors a different motivating reason in favor of the raid, since a description of the reason

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

49

includes an ineliminable indexical referring uniquely to the bandit harboring that reason. That is, each bandit wants himself to profit. So even in this case, each bandit qua deliberator furnishes a unique purpose for each of the other bandits. Each has made it so that, for each and every other bandit, he has a protected reason to steal a thousand beans—a protected reason that he would not have otherwise had. This makes sense, given that each bandit has a unique practical claim against every other bandit that he does his part. Suppose again you and I are two such bandits. I illicitly abandon my role and then attempt to rationalize my decision to you by saying that others will keep their end of the agreement they made with you. But you could rightly point out to me that the fact that others will abide by the agreements they made does not leave me off the hook. This suggests that my obligation to you—and the protected reasons you confer upon me—are directed. So, the many protected reasons and the many obligations they generate are not redundant, but instead unique in their content and their normative force. This is not, then, a case of bandits together furnishing one and the same purpose for one another, but instead bandits furnishing similar but importantly different purposes for one another. The skeptic, though, might then raise a related worry. If each bandit has literally a thousand different purposes furnished for what he does, then it seems any single purpose that a given bandit furnishes does little motivational work. After all, it’s not as if any given bandit has in mind one thousand different reasons for doing his part in the raid. But recall that the basis for a deliberator’s authority-based accountability lies not in motivating the executor to act in a particular way. Rather, the basis lies in constitutively determining the purpose for which the executor acts. The executor counts as acting for that purpose regardless of what actually motivates the bandit. A skeptic might, at this point, raise a different worry. The skeptic might grant that all the bandits are furnishing wrongful purposes for one another, where each wrongful purpose is unique. As a bandit, I am accountable for furnishing for all the other bandits a unique wrongful purpose. But it doesn’t seem to follow that I am accountable for the fact that what all the other bandits do count as wrongful. Suppose that I furnish for you the following purpose: to help steal and sell the peasants’ beans as a means of enriching me. It’s true that absent my participation, you wouldn’t have the purpose of enacting that motivating reason in favor of stealing the beans—a motivating reason that makes essential reference to me. But you would still have the purpose of enacting the motivating reasons of 998 other bandits. Again, it seems that my normative contributions don’t really make what you do much worse. Given this, it’s unclear what warrants the claim that any given bandit is accountable for the fact that what the other bandits do count as wrongful. It’s true that I am just one of 998 individuals furnishing a purpose for you, each of which enjoins you to steal from the peasants. But each purpose, including the one I confer, is normatively decisive on its own, in that it is sufficient as a

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

50

, ,  

normative basis for action. Each agreement you made yields a protected reason for you to act accordingly—it’s supposed to settle the matter for you. It’s true that I am not a but-for cause of the fact that what you do is wrongful. Nonetheless, what you do is wrongful because of me. The situation here is structurally analogous to cases of concurrent causal overdetermination. Suppose each of two villains places one lethal drop of poison in an innocent victim’s coffee, for a total of two lethal drops. The victim drinks the coffee and dies. Though neither of the perpetrators is a but-for cause of the murder, each of the villains caused the victim’s death on a INUS (an insufficient but necessary element of an unnecessary but sufficient set) test⁸ or a NESS (necessary element of a sufficient set) test,⁹ according to which an event c causes an event e if and only if c is a necessary element in a set of conditions sufficient for e where the set itself need not be necessary for e. In cases of concurrent overdetermination such as the case of the dual poisoners, each lethal dose is a necessary element of its own sufficient set. Thus, each lethal dose kills the victim. The same goes no matter how many poisoners there are. We can imagine a thousand such poisoners; each counts as killing the victim so long as the dose that each administers is on its own lethal. Back to the bandits: the normative relationship that I and the rest of the bandits bear to you is structurally analogous to the causal relationship the poisoners bear to their victim. Each poisoner causes the death of her victim even though she is not a but-for cause of the victim’s death; likewise, each bandit causes you to have the purpose of committing the thefts even though he is not a but-for cause of the fact that you have such a purpose. The result is that as one of the bandits I am indeed accountable for the fact that what you do is wrongful, even though I am not a butfor cause of the fact that what you do is wrongful. To summarize: there are three reasons why it might seem at first that the purpose that a given bandit furnishes for each of the other bandits does not really matter morally. First, the purpose that a given bandit furnishes seems the same as the purpose that every other bandit furnishes. In this respect, the given bandit seems to furnish a purpose that the other bandits already had. But as I argued, this is not so—each bandit furnishes a unique purpose, since each bandit’s motivating reasons are unique. Second, the purpose that a given bandit furnishes does little to no motivational work for the bandit to whom the purpose is furnished. This might seem to suggest that a given bandit’s motivating reasons do not serve, in any significant sense, as the object of what the other bandits do after all. But as I argued, this is not so—a deliberator’s motivating reasons determine the purpose of the executor’s conduct even if that reason does little to no motivational work for the executor.

⁸ See (Mackie, 1980).

⁹ See (Hart & Honoré, 1958).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

51

Third, though each bandit furnishes a uniquely wrongful purpose, the moral status of what any given bandit does (i.e., whether a given bandit’s conduct is wrong) is concurrently overdetermined, since nearly a thousand bandits are furnishing such purposes. Absent any given bandit, what the other bandits do would still be wrongful. It is hard to see, then, how we can say of any given bandit that the purpose he furnishes is what makes the conduct of the other bandits wrongful. But each bandit is, in effect, a concurrent cause of the fact that what the other bandits do count as wrongful. Thus, each bandit is individually and severally accountable, for the moral status of what the other bandits do. What if the bandits were to try avoiding mutual inculpation by dissolving itself as a group, while still carrying out their plan to rob the villagers? Such a possibility is, I suggest, incoherent. The bandits are only nominally dissolved if they’re still intentionally acting together in furtherance of their antecedent plan to rob the villagers. To better understand how the argument for authority-based accountability functions, it is helpful to consider how a victim of the bandits might react to the wrongdoing. Recall that in Olympic Sabotage, when Victim demands reasons from Goon, there comes a point at which Goon must refer her to Contender, since she is the one who furnished him with the purpose of attacking Victim. Likewise, if one of the peasants wants to know the purpose behind your actions when you steal from her, there is a point at which you will have to refer her, serially, to those who furnished you with that purpose. That is, there is a point at which you would have to refer her to every other bandit, including me. This is because, in accordance with the agreement we made, each bandit serves a) as a deliberator who has authority over every other bandit, and b) as an executor who is under the authority of every other bandit; each can demand of any other that he do his part, and each is likewise subject to such a demand. Thus, the peasant demanding answers can direct her reactive attitudes of resentment and indignation resulting from your actions toward me, because I furnished you with the purpose of stealing one of her beans, via a protected reason to act accordingly. It’s true that 998 other bandits furnished you with roughly the same purpose. But because the purpose each of us individually furnished for one another derives from a protected and thus a decisive reason to comply, the peasant can licitly direct to each and every bandit the entirety of her moral resentment and indignation for what you did. Of course, The Bandits and the Beans is a recherché case, to say the least. It includes a host of idealizing assumptions. Here are a few. • Each bandit is party to an agreement with every other bandit as a result of which she serves as both a deliberator and an executor. • No bandit has unilateral authority over any other bandit—there is no chain of command.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

52

, ,   • Each bandit is committed to stealing the beans as a means to achieving his own goals; none of the bandits are alienated or otherwise indifferent to the success of his task. • None of the bandits are ignorant of or deceived about what they’re doing together. • All the bandits are cooperating voluntarily—none of them are coerced. • All of the bandits are performing precisely the same task.

The example’s unrealistic simplicity serves as an initial stepping-stone to more complex cases which I address later on. In particular, I haven’t specified any common-knowledge conditions between the deliberator and the executor. What must be “out in the open” in order for a deliberator to confer protected reasons and thus a purpose on the executor? I discuss this issue specifically in section 4.3. Though the example is rudimentary, it helps reveal how the cooperants mentioned at the outset of this book—the Al-Shabaab militants, the soldiers in the SS Wachbattaillon, and the PMI outreach workers—might inculpate one another, as a result of which the individual cooperants can end up accountable for more than what is within their own causal reach. In Bandits and the Beans, though, individuals confer purposes upon one another, which serves as a basis for authority-based accountability. What about cases in which a group qua group confers a purpose? How do we determine the content of an agreement in cases of collective decisionmaking? The purpose of this book is not to settle that issue, but instead to provide an account of the moral implications of such decision-making.¹⁰ Of particular concern here is: what happens when the official purpose that a group confers qua group is different from the purpose that the individual constituents confer? For example, suppose the board members of a corporation running a chain of restaurants vote on whether to direct a manager to shutter a particular branch which happens to be located in a Black community. The vote is unanimous in favor of closing the branch. Some board members vote in favor of the closure on the grounds that their business needs to downsize. Others cast their vote on racist grounds. The “official” reason in favor of closing the branch, and the only reason discussed prior to the vote, is the need to downsize. Even if this purpose is the purpose that the group confers qua group, it’s unclear why we should think that it supplants or otherwise silences the individual purposes that the board members confer. To be clear, the claim is not that individual board members severally determine the content of what the manager is supposed to do. The judgment-aggregating decision-procedure determines that. Rather, the claim is that the individual board

¹⁰ The content of a collectively made decision obviously depends on the procedure by which the decision is reached. For more on judgment aggregation in the context of shared decision-making, see (List, 2005). For an economist’s perspective, see (Bacharach, 1999).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

53

members severally determine why the manager is supposed to shutter the branch. The official purpose the group confers does not vitiate the individual purposes that the board members confer. If the former employees of the now shuttered branch want to know why they were fired, their former manager could of course reveal her motivating reasons. But she might add: “If you want the whole story, you have to ask the board member chairing the meeting in which this decision was made.” Upon doing so, we can imagine that the chair cites the board’s official reasons for the vote, but then adds “If you want the whole story, you’d have to ask the individual board members.” The lesson here is that the official purpose that the group qua group confers can coexist with the various purposes that the individual members severally confer. And each board member can be accountable not only for serving as one of many who jointly determine the official purpose (via whatever judgment-aggregating decision procedure the board adopts) but also for conferring a purpose on her own. Where that purpose is morally problematic, the board member will bear authority-based accountability.

2.1.5 Authority-Based Accountability Versus Ordinary Accountability Return now to the Bandits and the Beans. If each bandit bears authority-based accountability, as I have argued, then is each bandit accountable for the harm inflicted on all the peasants? Suppose, contrary to what was assumed, one superbandit, acting alone, stole all the beans from all the peasants. How do we compare the super-bandit’s accountability with that of an ordinary bandit in the example’s original version? Is what the ordinary bandit does morally tantamount to what the super-bandit does? No. An act’s purpose is only one potentially wrong-making feature of that act. There are other potentially wrong-making features, including, most notably, the harms that the act causes. In morally comparing the super-bandit with the ordinary bandit, we have to determine which wrong-making features of their respective situations we can pin on them. The super-bandit is accountable for the following wrong-making feature: the wrongful ends for which he acts. The superbandit is also accountable for another wrong-making feature: causing the death of all the villagers. Now compare this to the ordinary bandit. The ordinary bandit is accountable for the following wrong-making feature: the wrongful ends for which all the other bandits act. In this respect, the ordinary bandit and the super-bandit are alike in that they both determine the aims in furtherance of which the peasants starve. The super-bandit does so in that he is a deliberator of his own actions, and the ordinary bandit does so in that he is a deliberator of the actions that the other bandits commit. But, unlike the super-bandit, the ordinary bandit is not accountable for causing the death of all the villagers.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

54

, ,  

An upshot is that, even given the ordinary bandit’s authority-based accountability, what the super-bandit does is worse. But how much worse is it? I confess that I am not sure. I therefore relegate myself to the claim that authority-based accountability is accountability for making conduct wrongful by giving that conduct a wrongful purpose. I make no further claims about how wrong this wrong is in comparison to wrongfully causing a harm.

2.2 Contrast with Other Views Other theorists have developed accounts of individual accountability in the context of shared action,¹¹ and in the context of authority.¹² Here I will contrast mine with others. I will focus on the accounts that Abraham Roth and Christopher Kutz develop respectively since they bear important similarities to the account I have developed here.

2.2.1 Abraham Roth Abraham Roth has argued that when individuals act together, they act “directly” on the intentions of one another in a way analogous to the manner in which an agent acts on her own intentions.¹³ Suppose we’re driving together to a conference. We’ve agreed that you’ll serve as the navigator by reading off directions, and I’ll serve as the driver by heeding your instructions. Roth contends that when you issue instructions—say, for me to turn left—and I follow those instructions, I thereby take up and act on your intentions without “re-issuing” the intention in my own mind. Instead, I am “preserving and executing” your intentions.¹⁴ In this respect, “one might act directly on someone else’s intention, just as one might act directly on one’s own prior intentions.”¹⁵

¹¹ Margaret Gilbert has written extensively on this issue. See also (Feinberg, 1968), (McGary, 1986), (May, 1987), (French, 1998), (Tollefsen, 2003), (Sadler, 2006), (Miller, 2006), (Smiley, 2010), (Mellema, 2016), (Lepora & Goodin, 2013), among others. ¹² Larry May, for example, (May, 2005, pp. 139–148) argues that we can analyze accountability in hierarchical groups in the way we analyze accountability in conspiracies. He emphasizes the role that intention and negligence plays in accountability for wrongdoing in the context of hierarchical activity (May, 1987, pp. 89–106). On this view, a superior is accountable because “she had knowledge of the criminal activity of subordinates and failed to stop it or to punish those who so acted” (May, 2005, p. 142). His account is compatible with the thesis I develop in this book. Tracy Isaacs, on the other hand, argues that a superior is accountable for what her subordinates do at least partly because the propositional content of superior’s order can be described in terms of the causal upshots of that order (Isaacs, 2011). In developing this argument, Isaacs draws on what Joel Feinberg called the “accordion effect” (Feinberg, 1970). ¹³ (Roth, 2004). ¹⁴ (Roth, 2004, p. 384). ¹⁵ (Roth, 2004, p. 383).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

55

This account, Roth urges us, does not entail any sort of psychological “action at a distance.” He makes clear that your intentions still need to be successfully communicated to me. But he maintains “this is just like the individual case wherein my acting directly on my prior decision/intention is conditioned on the successful transmission of the intention, usually via memory.”¹⁶ At first, this account might seem similar to mine. Roth says that when I act on your intention, your decision “settles the matter” for me in that your decision “does not merely provide a strong reason for me to act” as you intend. Your instructions do not serve as “a reason that is supposed to enter into my deliberation about what to do.” Instead, in such a case, “I do not deliberate at all, or at least not enough to undermine this other individual’s ability and authority to settle what it is that I’m to do.”¹⁷ But Roth’s account is much more radical than mine. Like Roth, I claim that when we establish (what I call) an “interpersonal division of agential labor” in which you count as the deliberator and I count as the executor, your directives have the normative function of “settling the matter” for me. But I do not go so far as to claim that I am thereby acting directly on your intentions. That is, I do not claim that “intentions and not merely their contents” are “transmitted and preserved between individuals through communication.”¹⁸ Roth later expands on this account but in terms of reasons rather than intentions. On his view, when I follow your instructions by turning left, the best normative explanation of my conduct adverts in part not to reasons I have but to reasons you have. On this proposal, “sometimes one may act for reasons that one doesn’t possess.”¹⁹ In such a case, I turn left when you instruct me to do so not in order to comply with your instructions. Rather, to determine why I am turning left, we need to advert to your reasons. “I am in this way,” Roth says, “taking you as a source and repository of reasons for what I’m doing.”²⁰ This is because we aim to go to the conference together, which means the reasons you have in favor of turning left serve an end that we share. Again, this is more radical than what I claim. Roth contends that what serves as my motivating reason in favor of turning left when you say “turn left” is literally the motivating reason you take yourself to have in favor of turning left. On this view, there is no space between my motivating reasons and yours—they are one and the same. This view is stronger than the one I accept. I’ve argued that the reasons the deliberator takes there to be determine what the executor’s protected reasons are and thus what the executor’s purpose is. But this claim falls short of the claim that the deliberator’s motivating reasons are the executor’s motivating reasons. I make no such claim. Many think that the reasons I take there to be must, in some way, ¹⁶ (Roth, 2004, p. 383). ¹⁹ (Roth, 2017, p. 76).

¹⁷ (Roth, 2004, p. 385). ²⁰ (Roth, 2017, p. 85).

¹⁸ (Roth, 2004, p. 408).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

56

, ,  

be accessible to me. But if the reasons you take there to be determine the reasons I take there to be, which is what Roth claims, then I do not necessarily have access to the content of the very reasons I take there to be. On this view, in a division of agential labor, the executor might be introspectively ignorant of the reasons she putatively takes there to be. Any theory countenancing this possibility is questionable. This is because there is a normative connection between the practical reasons we take there to be and the act of reasoning. If you take yourself to have a reason to φ, then presumably the reason in question can at least in principle serve as a premise in your reasoning toward doing ϕ. As Jonathan Way puts it, “Reasons are meant to guide us to act, believe, desire, or otherwise respond. But to be guided by reasons just is to engage in reasoning, broadly construed.”²¹ Or, to quote Mark Schroeder, “When an agent is reasoning well, the kinds of things about which he should be thinking are his reasons.”²² Some theorists have accordingly defined practical reasons in the following way: a reason to do ϕ just is a premise of good reasoning.²³ This is what Way calls the “Reasoning View” of practical reasons. The Reasoning View is a view about the reasons that there are, rather than the reasons that we take there to be. But part of what makes the Reasoning View compelling is that responding to the reasons that there are or that we think there are means reasoning in a way that rationalizes our actions. As Donald Davidson put it: A reason rationalizes an action only if it leads us to see something the agent saw, or thought he saw, in his action—some feature, consequence, or aspect of the action the agent wanted, desired, prized, held dear, thought dutiful, beneficial, obligatory, or agreeable.²⁴

If the reasons the deliberator takes there to be determine (or otherwise just are) the reasons the executor takes there to be, then the executor does not necessarily have access to the very reasons she takes there to be. And if she does not know what those reasons are, they cannot serve as premises in her reasoning. The putative reason fails to rationalize the executor’s actions since the putative reasons reveal nothing that the executor saw or thought she saw in favor of the action in question. So, on the Reasoning View of reasons, these putative reasons do not count as reasons at all. I make no claims either way about whether the Reasoning View of practical reasons is correct. Still, the Reasoning View poses a problem for Roth’s account. He seems to be aware of this; he writes that his view “will no doubt strike many as

²¹ (Way, 2015, p. 251). ²² (Schroeder, 2007, p. 26). ²³ Others who seem to endorse this sort of approach include (Gibbons, 2010), (Hieronymi, 2005), (Raz J., 1978), (Setiya, 2007), (Setiya, 2014). ²⁴ (Davidson, 1963).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

57

controversial” partly because “it challenges some conceptions of the agent’s nonobservational epistemic access to the reasons for which she acts.”²⁵ The account I developed avoids this issue since I do not claim that the reasons the deliberator takes there to be are (or even determine) the reasons the executor takes there to be, but rather that the executor has the purpose of enacting the reasons that the deliberator takes there to be. It is perfectly felicitous to claim that the executor’s purpose, unlike her motivating reasons, might remain introspectively opaque to her. Such an account is compatible with the Reasoning View of reasons. Still, one might raise the following objection. Return to the example Olympic Sabotage. There, Contender pays Goon to maim Victim so that Contender can win an upcoming competition. Suppose Goon has no interest in whether Contender wins. He is only concerned about getting paid. To do that, he needs only maim Victim. Given a commitment to the Reasoning View of reasons, Goon seems to lack a protected reason to help Contender win. Rather, he only has a protected reason to maim Victim. It’s true that doing so will help Contender win. But the reasons we have are not closed under causal or logical implication. That is, if I have a reason to do ϕ, and ϕ causes or entails ψ, I do not thereby have a reason to do or to cause ψ. So, from the fact that Goon has a protected reason to maim Victim we cannot conclude that he has a protected reason to help Contender win. But this is perfectly compatible with the argument for authority-based accountability. We do not read an executor’s purpose off her conception of the protected reasons she has. Instead, we read her purpose off the deliberator’s motivating reasons. When Contender confers and Goon accepts a protected reason to maim Victim, the practical reasons that Contender takes there to be serve as a basis for determining the purpose of what Goon does in maiming Victim. As far as Contender is concerned, the whole point of maiming Victim is to eliminate her from the competition. That is, then, Goon’s purpose in maiming Victim, whether he knows it or not. Thus, Contender bears authority-based accountability for furnishing for Goon that purpose. The point here is that even though Goon lacks a motivating reason to help Contender win, he still has that purpose. Suppose, though, that Roth’s view is correct after all. The account I develop of authority-based accountability could easily accommodate such a view. If the executor is literally acting on the deliberator’s intentions, or literally acting out of her reasons, and the deliberator’s intentions or reasons are morally problematic, then the deliberator is accountable for a wrong-making feature of the executor’s conduct: the intentions with which, or the reasons out of which, the executor acts. The result, again, is that the deliberator bears authority-based accountability for what the executor does.

²⁵ (Roth, 2017, p. 90).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

58

, ,  

2.2.2 Christopher Kutz Kutz develops a sophisticated account of individual accountability for harms in the context of jointly intentional collective action. On his account, jointly intentional collective action consists of “overlapping, individual participatory intentions”.²⁶ A participatory intention is just a regular intention but with “collectivized” content; it is an intention to accomplish a primary individual task which, in turn, contributes to a collective act.²⁷ To possess a participatory intention, the participant need not intend the collective end itself; but she does need to intend to do something that she knows will contribute to it. In this respect, the participants need only “regard themselves as acting for the sake of some joint goal; no more content is necessary,” in that they need not know what the goal is or with whom they’re acting.²⁸ Intentions overlap when a single event satisfies the intentions in questions.²⁹ Hence, a collective act is jointly intentional only under a suitably generalized description. “You may believe,” Kutz says, “we are going to a friend’s house for a quiet dinner, while I believe we are going for a surprise party. While our going to the surprise party is not jointly intentional, our going to the friend’s house is.”³⁰ On what Kutz calls the “Complicity Principle” I am accountable for what others do independent of the effects my actions have when my participatory intentions overlap with theirs. That is, I am accountable for what others do, independent of the effects my actions have, when I act as part of a jointly intentional collective action. This is because, when we act together in this way, what you do can be ascribed to me, and what I do can be ascribed to you.³¹ In this respect, each of us is what he calls an “inclusive author” of what the other does. Our actions are mutually attributable in this way because, on his account, my actions teleologically explain yours, and your actions teleologically explain mine. A function of my actions when we engage in jointly intentional action is to help you achieve your aim within the scope of our overlapping participatory intentions; a purpose of your actions is to help you achieve my aim. Kutz puts it this way: “If a set of agents’ participatory intentions overlap, then the will of each is represented in what each other does qua group member, as well as what they do together. The logical overlap permits us to say they manifest their attitudes through one another’s actions.”³² And this, in turn, serves as a basis for mutual accountability. This account shares similarities with mine. On my account, an interpersonal division of agential labor is what grounds authority-based accountability; I analyze an interpersonal division of agential labor that an executor and deliberator

²⁶ (Kutz, 2000, p. 75). ²⁹ (Kutz, 2000, pp. 76, 94). ³² (Kutz, 2000, p. 141).

²⁷ (Kutz, 2000, p. 82). ³⁰ (Kutz, 2000, p. 95).

²⁸ (Kutz, 2000, p. 83). ³¹ (Kutz, 2000, pp. 138–139).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

59

establish by adverting to the executor’s executory function of implementing the deliberator’s will, and the deliberator’s deliberative function of determining what it is that the executor is to do. Likewise, on Kutz’s account, inclusive authorship is what grounds the complicity principle; he analyzes inclusive authorship in terms of the functional role my will plays with respect to your actions, and vice versa. But Kutz’s account, I believe, rests on a somewhat dubious or at least mysterious view of actions. Suppose I see you attempting, in vain, to push a car out of the mud. I decide to help you out, for which you are grateful. We both thereby possess overlapping participatory intentions of freeing the car. But what licenses Kutz’s claim that your actions manifest my will, and my actions manifest yours? It’s true that I see myself as sharing an end that you have, and you see yourself as sharing and end that I have, and it’s true that this shared end serves as a goal for me and as a goal for you (indeed, that’s just what it means for the end to be shared). But for your conduct (or even just its upshot) to be attributable to me, it is not enough for our ends to be shared. In addition, I must be a cause of what it is you are doing. Indeed, Kutz himself says earlier in his book that he “will assume intentional action is action (body movements) that is both causally and teleologically explained by an agent’s goals” (my emphasis).³³ Yet Kutz explicitly disavows the claim that a causal role is necessary to ground inclusive authorship.³⁴ Be that as it may, I think Kutz is right when he says that there is a sense in which we are “the authors” of each other’s actions when we act together. The problem, though, is that Kutz never adequately spells out this notion. If authorship is just action-attribution, then we run afoul of a basic tenet in action theory—that actions are caused by the actor. If authorship is just responsibility for an action, then the Complicity Principle ends up viciously circular. The trick is to explain the concept ‘authorship’ in a way that avoids the Scylla of literal action-attribution and the Charybdis of responsibility. The account I develop avoids this problem. When an executor acts on a deliberator’s instructions in a division of agential labor, the function of what the executor does is to implement the deliberator’s will. This doesn’t mean that the deliberator literally does what the executor does. But it does mean that the deliberator is accountable for a feature of what the executor does—the deliberator is accountable for the purpose for which the executor acts. To the extent that the purpose the deliberator furnishes is morally problematic, the deliberator is accountable. And when we work together, each of us is both executor and deliberator, which means each of us is accountable for the purpose for which the other participants act (absent excusing conditions, of course, which I discuss in chapter 4). Perhaps this is a way to make sense of the intuitively appealing claim

³³ (Kutz, 2000, p. 72). ³⁴ “My inclusive accountability,” Kutz says, is “independent of the difference I make or the control I have.” (Kutz, 2000, p. 141).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

60

, ,  

that when we work together, we are authors of each other’s actions. For this reason, I see my account as in keeping, at least in spirit, with Kutz’s account.

2.3 Counterarguments and Arguments from Casuistry Before I address how we establish divisions of agential labor, I will respond to some possible criticisms of the account I’ve developed. I begin in section 2.3.1 by considering the possibility that intentions—and with it, motivating reasons—are irrelevant to the moral permissibility of action. In the remaining sections I consider some apparent counterexamples in which it might seem that the deliberator is not accountable, contrary to what the argument for authoritybased accountability suggests. There are a host of other potential challenges to the argument for authority-based accountability. What happens when the deliberator or the executor doesn’t take the authority-relation seriously? What happens when the deliberator or executor is coerced into their respective roles? What happens when either of them is ignorant of what falls under the aegis of the deliberator’s authority? I address these worries later on, in chapter 4.

2.3.1 Deontic Relevance of Motivating Reasons The argument for authority-based accountability presumes that the practical reasons an individual takes there to be in favor of an (in)action—i.e., her motivating reasons—are relevant not just to the evaluation of that individual’s character but to the evaluation of the act she performs. But this presumption is contentious. It is closely related if not identical to the tendentious claim that intentions with which an act is performed can affect the moral permissibility of that act. Consider the following case: Trolley 1 A trolley is heading toward five persons. The trolley will kill the five unless someone pulls a lever that will divert the trolley toward one other person whom the trolley will then kill. Boss, who works on the train tracks, is witness to these unfolding events. Suppose in one version of this case, Boss is a morally upstanding individual. Call him “Good Boss”. He takes there to be a sufficiently strong reason to save the five, even at the cost of regrettably killing the one as a side effect. Good Boss consequently pulls the switch with the intention of saving the five. In another version of this case, Boss is a morally reprehensible person. Call him “Evil Boss”. He normally would want as many people as possible to die; but he takes there to be

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

61

a special reason for killing the one individual, even at the cost of regrettably saving the five others. Evil Boss consequently pulls the switch with the intention of killing the one. Those who claim intentions are morally relevant to permissibility will claim that Good Boss but not Evil Boss is permitted to pull the switch. Those who deny that intentions are morally relevant to permissibility will claim that both Good Boss and Evil Boss are permitted to pull the switch. To see how this disagreement affects the possibility of authority-based accountability, we can expand on the example in the following way. Trolley 2 Boss cannot get to the lever in time. He knows that a subordinate of his, Worker, who also works on the train tracks, is situated closer to the lever. But Worker isn’t aware of what’s going on. So, Boss hails Worker on his walkie-talkie. Boss doesn’t have time to explain his reasons; he instead demands that Worker pull the lever now. Worker abides by that demand. According to the argument for authority-based accountability, the deliberator’s motivating reasons constitutively determine the purpose that the deliberator furnishes for the executor. Given that Boss has issued to Worker a protected reason to pull the lever, Boss thereby furnishes for Worker a particular purpose in pulling the lever (even if Worker would not agree with or would otherwise disavow that particular purpose). But the purpose furnished differs depending on whether Good Boss or Evil Boss is the deliberator. Good Boss furnishes Worker with the purpose of saving the five at the regrettable cost of killing the one as a side effect. Evil Boss furnishes Worker with the purpose of killing the one at the regrettable cost of allowing the five to live as a side effect. If, in Trolley 1, Boss’s reasons for pulling the switch are morally irrelevant to our evaluation of what Worker is permitted to do, then, in Trolley 2, the reasons Boss furnishes for Worker are likewise irrelevant to our evaluation of what Worker is permitted to do. The claim that Boss’s reasons are morally irrelevant to our evaluation of what Worker is permitted to do is tantamount to denying that Boss bears authority-based accountability for what Worker does. After all, to say that a deliberator bears authority-based accountability for what his executor does, is to say in part that the deliberator is accountable for a wrong-making feature of what his executor does: a purpose for which the executor acts. But if the purpose that the deliberator furnishes is morally irrelevant to the permissibility of what the executor does, then the deliberator is “accountable” for a morally irrelevant feature of what the executor does. The upshot is that if the reasons we take there to be are morally irrelevant to the evaluation of our actions, then it seems that the argument for authority-based

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

62

, ,  

accountability falls apart. There are, though, two strategies in avoiding this outcome. I will proceed by developing both. The first strategy is to defend the view that motivating reasons are indeed deontically relevant. The second strategy is to argue that even if they are not, the argument for authority-based accountability survives that outcome. Though I cannot present here a knockdown argument for the view that motivating reasons are deontically relevant, I can defend such a view by responding to a popular argument against it. Though the argument in question purports to show that intentions are deontically irrelevant, it also threatens the deontic relevance of motivating reasons as well. To understand the argument claiming to show that intentions are irrelevant to moral permissibility, return to Trolley 1. On the view that intentions are morally relevant to permissibility, Good Boss but not Evil Boss is permitted to pull the lever, even though both cases are the same when described extensionally. In favor of such a view, T. M. Scanlon argues that when you believe that an action ϕ has reason x but not reason y in its favor, you lack the ability to choose between doing ϕ-for-reason-x and doing ϕ-for-reason-y.³⁵ This means Evil Boss cannot choose between pulling the lever to kill the one, and pulling the lever to save the five. If acting for the right reasons is a prerequisite for acting permissibly, then Evil Boss is incapable of acting permissibly. Such a conclusion seems to violate the claim that we can be morally required to do only what we are capable of doing—that ‘ought implies can’. Thus, we ought to conclude that acting for the right reasons is not a prerequisite for acting permissibly. Or so Scanlon argues. Scanlon’s argument, though compelling, is mistaken. Victor Tadros helps explain why.³⁶ Scanlon argues that there is a sense in which we cannot choose what to see as reasons; we are instead confronted with what we take to be reasons. Once an agent comes to see certain reasons in favor of φ as decisive, she cannot rationally decide to act for some different set of reasons.³⁷ Absent a relevant change in the circumstances, re-litigating the reasons in favor of ϕ will yield the same result as the one she reached earlier.³⁸ Tadros makes a similar point but in terms of values: for the individual to change her mind at t₂ about the reasons in favor of ϕ would means that she “gives up” valuing ϕ in one way and “comes to value it in some other way,” which, he says, “seems hard to do.”³⁹ But as others point out,⁴⁰ we determine what reasons there are in favor of an action in the course of deciding what to do rather than after that decision. Typically, we decide what to do by considering the reasons for and against candidate courses of action; once we have decided what to do, the reasons for which to do it are already normatively settled, as far as we’re concerned. Put

³⁵ (Scanlon, 2008). ³⁶ See (Tadros, 2019, pp. 116–117). ³⁷ (Scanlon, 2008, pp. 59–60). ³⁸ See discussion in section 1.2 (chapter 1). ³⁹ (Tadros, 2019, p. 116). ⁴⁰ See (Kolodny, 2011) and (Tadros, 2019, p. 117).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

63

differently, the reasons we take there to be are action-guiding. Such reasons are relevant to moral permissibility when those reasons play the role of informing us what we ought to do. When we claim that Evil Boss did the right thing but for the wrong reasons, we are not claiming that he should have a) decided to pull the lever, and then b) adopted the right reason for doing so. Scanlon is correct in surmising that this would indeed be a bizarre requirement. But this is not what we are requiring of Evil Boss. Rather, we are claiming that he should have a) adopted the right reasons in deciding whether to pull the lever, then b) decided to pull the lever. And this requirement is perfectly appropriate. The upshot is that Scanlon’s argument for the view that we cannot choose the reasons for which we act, fails to undermine the claim that the reasons for which we act are relevant to moral permissibility. I suspect that proponents of the view that intentions are morally irrelevant will have other reasons for remaining committed to that view. This, in combination with the fact that I have presented no positive reasons for that view, might suggest that I have not done enough to show that intentions—and thus motivating reasons—are morally relevant. Fortunately, though, there is reason for thinking that even if intentions are not morally relevant, that the argument for authoritybased accountability still goes through. To see why, it’s important to distinguish three senses in which intentions might be morally relevant. First, intentions might be morally relevant in the agentneutral sense. This is the view I’ve been defending so far. It suggests that all things being equal everyone has reason to prevent anyone from acting on wrongful intentions (even if the action “itself”—i.e., construed extensionally—is perfectly justified). Second, intentions might be morally relevant in the agent-relative sense. This suggests that all things being equal, only those who harbor the wrongful intention in question have a reason not to act on that intention. Third, intentions might be relevant in the aretaic sense. This suggests that wrongful intentions are relevant only to our characterological evaluation of the intender and are ultimately irrelevant to our evaluation of what she does. Suppose, then, that I am wrong in thinking that intentions are morally relevant in the agent-neutral sense. (Hence, the best impersonal axiology of the world does not evaluate actions by reference to the intentions with which they are done.) This still leaves open the possibility that such intentions might be morally relevant in the agent-relative sense. This is, famously, Thomas Nagel’s view.⁴¹ On this view, I ought not to commit (otherwise permissible) acts with wrongful intentions; ditto for you. But I have no reason to ensure that you refrain from committing (otherwise permissible) acts with wrongful intentions, and you have no reason to ensure that I refrain from committing (otherwise permissible) acts with

⁴¹ (Nagel, 1989, pp. 180–183).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

64

, ,  

wrongful intentions. On this view, bad intentions are relevant only from a firstpersonal standpoint, in that they provide reasons only to those who harbor the intentions in question. Does the argument for authority-based accountability go through if intentions—and with it, motivating reasons—have only agent-relative (and not agent-neutral) deontic relevance? At first, it seems like the answer has to be “no”. After all, the deliberator is one person, and the executor is another. If motivating reasons have only agent-relative deontic relevance, then one party cannot be held accountable for the purpose with which another party acts. But recall that deliberator and executor share a division of agential labor; the aspects of agency normally wrapped up in a single individual are, for practical purposes, “distributed” between the deliberator and the executor. Inasmuch, the deliberator constitutively determines the purpose of the executor’s actions. It is literally the deliberator’s own motivating reasons that determine the executor’s purpose. So, to suggest that the deliberator is accountable for executor’s purpose isn’t to suggest that the deliberator is accountable for someone else’s motivating reasons, but rather that she is accountable for her own. Hence, the argument for authoritybased accountability goes through even if motivating reasons are morally relevant in only the agent-relative sense. What if, though, intentions have only aretaic relevance? If so, performing an act with bad intentions doesn’t affect the deontic status of the action; rather, it reflects badly on the actor as a person. The action itself, even when described in terms of the bad intentions that motivated it, can be morally permissible (contrary to the view that intentions have deontic relevance). The argument for authoritybased accountability can accommodate this view, though it will have different implications. In particular, it would be misleading to say that the deliberator is accountable for a “wrong-making feature” of what the executor does since motivating reasons do not make actions wrongful on the aretaic view. Instead, what the executor does reflects badly on the deliberator’s character. When a deliberator tasks an executor with enacting the deliberator’s wrongful motivating reasons—that is, when the deliberator confers a wrongful purpose upon an executor—the executor’s actions, saddled with an inimical purpose, reflect badly on the deliberator, given the division of agential labor they’ve established. The deliberator, after all, is the one who conferred that inimical purpose upon what the executor does. Think again, then, of The Bandits and the Beans; if the aretaic view is correct, any given bandit’s moral character is impugned by what every other bandit does. This makes any given bandit quite a bad person, indeed. But what about cases of executor accountability, as discussed in section 2.1.3? Think of Performance Review in which Manager is tasked with promoting Senior, which would be morally unproblematic if not for Owner’s sexist reasons for doing

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

65

so. If the aretaic view is correct, Manager’s actions impugn Owner’s character, despite that Manager isn’t himself sexist. But what about Manager? It seems that because Owner is the one who saddles Manager’s actions with a sexist purpose, the action reflects badly on Owner rather than Manager. Of course, Owner accomplishes his sexist goals via Manager. Does this put Manager on the hook? Not in and of itself, given the aretaic view. This is because, on that view, Owner’s sexist attitude is irrelevant to whether Manager’s actions are morally permissible. Indeed, it is characteristic of such a view (as well as the view that motivating intentions are morally relevant in only the agent-relative sense) that we can permissibly enable others to act on wrongful motivations, provided that what they do is otherwise morally permissible. (So, for example, we can permissibly enable a trolley-switcher to pull the lever away from the five and toward the one, even if the trolley-switcher’s intentions are murderous.) It seems, then, that Manager is free of recrimination given the aretaic view. But this is not so. To see why, recall the primary function of practical reasoning and of intentional conduct. The function of practical reasoning is to normatively guide conduct by determining what ends the actor is to pursue and how to pursue them. Concomitantly, the purpose of intentional conduct is to enact the practical reasons the actor takes there to be. Owner and Manager are in a division of agential labor, which means Owner, as a deliberator, has the function of deciding what Manager should do and why, within a given domain of conduct. Likewise, Manager’s actions, within that domain of conduct, have the function of acting accordingly. Normally, when acting on his own, Manager’s private motivating reasons constitutively determine what she is supposed to do and why; when Manager acts as Owner’s executor, it’s Owner’s motivating reasons that constitutively determine what Manager is supposed to do and why. This means the relationship Manager bears to Owner is functionally analogous to the relationship in which Manager bears to his own self when he is acting in his own capacity. As such, Manager can no more disavow the Owner’s motivating reasons than he can his own, provided that a) Manager is at least partly responsible for establishing the functional relationship between himself and Owner, and b) he knows or is in a position to know what Owner’s motivating reasons are. Put more generally, an executor doesn’t just help the deliberator achieve her goals; in addition, the deliberator’s goals functionally count as the executor’s. The moral here is that the argument for authority-based accountability has important implications even if motivating reasons are not deontically relevant. That being said, I will continue to proceed as if such reasons are indeed deontically relevant (since I believe that they are). Those who deny as much can translate my arguments and their implications accordingly into claims about the moral character of the deliberators and executors.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

66

, ,  

2.3.2 Causally Inert Promises One might raise the following sort of example in response to the argument for authority-based accountability. Double Promise Mastermind puts out an ad for a hitman to kill Politician but offers only paltry remuneration. Assassin takes the job, but only because she has already agreed to kill Politician for someone else who promised to pay much more. The first promise was sufficient motivation for Assassin to kill Politician. By the time she took on the second promise, Assassin had already settled on a plan for murdering Politician; the second promise has no effect on that plan. Indeed, after accepting the promise from Mastermind, Assassin promptly forgets doing so. On the account I’ve developed, Mastermind bears some accountability for the murder, even though he does not serve as a but-for cause or even as an INUS/NESS-condition for the murder’s occurrence. Is this, intuitively, the right answer? It might be tempting to think that a deviant causal chain is at work in this example. On such an analysis, Mastermind would be accountable for a mere attempt, in that Mastermind is a beneficiary of good outcome moral luck. By hypothesis, he didn’t serve as a cause of the murder. Though Assassin achieved Mastermind’s aim, he got lucky in that Assassin forgot about the promise she made to him; it was not because of Mastermind that Assassin achieved his aim. Our intuitive reaction to this example, then, might depend on whether we think outcome luck is morally relevant. Those who think it is relevant might conclude that Mastermind is accountable merely for an attempt, whereas those who think it isn’t relevant might conclude that Mastermind is accountable for the murder after all. The moral here is that proponents of the view that outcome luck is morally relevant might point to examples like Double Promise in an attempt to show that the argument for authority-based accountability yields unintuitive consequences. But such a view is predicated on the moral relevance of outcome luck, which is itself a contentious position. Hence, cases like Double Promise hardly serve as decisive evidence against the argument for authority-based accountability. Given the contentious status of moral luck, I believe we ought to take the argument for authority-based accountability at face value: Mastermind does indeed bear authority-based accountability for the murder. Even though Assassin forgot that she made a promise to Mastermind, she still retains that purpose so long as their agreement remains normatively in effect. Hence, Assassin acts in accordance with that promise even though she wasn’t aware of it at the time

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

67

of acting. Insofar as Assassin successfully serves as an executor and Mastermind successfully serves as a deliberator, Mastermind bears authority-based accountability for what Assassin does by furnishing him with a purpose to so act.

2.3.3 Consequentialist Reasons to Cooperate in Wrongdoing Here is a second argument from casuistry. Consider the following possible counterexample. Baker’s Dozen Each of a dozen individuals agrees with one another to add a single drop of poison to a carafe out of which twenty innocent people will drink (all of whom are enemies of the dozen conspirators). Ten drops are enough to kill all of the drinkers. Before executing their plan, they ask a thirteenth individual, Baker, to join in on the plan by adding a thirteenth drop of poison. However, Baker has already set plans into motion to kill an innocent enemy of hers. The only way for Baker to interrupt those plans, thereby preventing her innocent enemy’s demise, is by participating in the cooperative act of killing the twenty drinkers. What should, morally, Baker do? By hypothesis, as a thirteenth contributor, she won’t make a difference to whether any of the twenty innocents will die. But she will thereby make a difference to whether she kills the one innocent. On the other hand, by refraining from contributing, she won’t thereby save any of the twenty, but she will thereby end up killing the one. To some, it will seem intuitive that Baker should go ahead and participate in the cooperatively committed act. How does the account I have developed accommodate this example? If Baker refrains from participating in the cooperative act, she won’t bear authority-based accountability for killing twenty innocents—but she will be accountable for killing one innocent. If, on the other hand, she participates in the cooperative act, she won’t be accountable for killing the one innocent, but she will bear authoritybased accountability for killing twenty innocents. It seems, then, that we have to weigh authority-based accountability for the death of twenty innocents versus ordinary accountability for the death of one innocent. If the argument for authority-based accountability is to accord with our pre-theoretic intuitions, then ordinary accountability will have to be worth more than twenty times what authority-based accountability is worth. Indeed, it doesn’t matter how many drinkers there are—we can make the number arbitrarily high. Suppose there were twenty million drinkers; again, if Baker’s contribution doesn’t make a difference to their deaths, then it seems she ought to go ahead and contribute if the alternative means killing one person who

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

68

, ,  

would otherwise live. This seems to diminish the comparative importance of authority-based accountability to the positive limit of zero. But this just shows that the moral reasons we have to avoid furnishing others with a wrongful purpose are not the only kind of reason that matters in morality; other moral reasons might have lexical priority. It might be that as important as it is to avoid furnishing others with a wrongful purpose, it is always more important to avoid killing other people. That being said, the argument for authority-based accountability is still relevant in such a case in that it reveals what we owe to the victims of the cooperatively committed harm. Even if in Baker’s Dozen she ought to give (lexical) priority to refraining from killing, she still has a pro tanto reason to refrain from furnishing the other participants in the cooperative harm with a wrongful purpose. By acting contrary to that pro tanto reason, she thereby owes compensation to the victims of that cooperative harm even though Baker a) acted in the way that she should, and b) did not make a difference to whether any of the victims were harmed. Thus authority-based accountability retains its relevance even in cases where it is lexically overridden.

2.3.4 Group-Size and Authority-Based Accountability Finally, consider this third possible counterargument. Conspiracy Choice In Conspiracy A, each of twelve individuals agrees with one another to add a single drop of poison to a carafe from which twenty innocent people will drink. Ten drops of poison are enough to kill all those who drink from the carafe. Before executing their plan, they ask a thirteenth individual, Chooser, to join in on the plan by adding an additional drop of poison. Meanwhile, in Conspiracy B, each of 120 individuals agrees with one another to add a single drop of poison to a second carafe from which a (different) twenty innocent people will drink. Ten drops of poison are enough to kill all those who drink from the carafe. They invite Chooser to join in on the plan by adding an additional drop of poison. Supposing that Chooser has no choice but to agree to participate in one of these two cooperative harms, which one should she choose? Regardless of which Chooser selects, she makes no difference to who lives and who dies. Some might think that, intuitively, she ought to select Conspiracy B, because her drop of poison constitutes a smaller overall percentage of the poison in the carafe. The argument for authority-based accountability, however, seems to suggest that Chooser ought to participate in Conspiracy A. After all, by participating in Conspiracy A, Chooser furnishes a wrongful purpose for twelve other individuals. But by participating in Conspiracy B, Chooser furnishes a wrongful purpose for 120 other individuals.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

69

Recall from section 2.1.4 that the purpose a deliberator furnishes for her executor is unique and directed—thus it isn’t made redundant by other deliberators who furnish an extensionally equivalent purpose. Assuming that furnishing a wrongful purpose for 120 executors is worse than furnishing a wrongful purpose for twelve executors, the argument for authority-based accountability suggests— counterintuitively—that participating in Conspiracy B is worse than participating in Conspiracy A. But, again, the reason we have for refraining from furnishing a wrongful purpose for others is not decisive. It can be outweighed by competing considerations. Take the causal influence we have on the motivations of others. Though such causal influence does not play a role in the argument for authority-based accountability, it does indeed play a role in morality. By hypothesis, each conspirator agrees with every other conspirator to participate. This potentially motivates them to act in accordance with such agreements. The greater the number of conspirators, the less the causal influence that Chooser’s agreement has on the motivations of any other given conspirator; the greater the number of conspirators, the greater the number of individuals that Chooser’s agreement motivates. But the relationship between the quantity and quality of the motivational effect that an agreement confers might not be linear; thus, it might turn out that Chooser has more of an effect overall in Conspiracy A than she does in Conspiracy B. If that’s the case, it would be a reason for Chooser to select Conspiracy B. But this doesn’t suggest that the argument for authority-based accountability is mistaken. Rather, it just suggests that authority-based accountability is but one morally relevant factor among several. There is another factor in Conspiracy Choice competing with authority-based accountability. I stipulated that more than ten drops of the poison makes no difference to whether the victims die. The result is that in both Conspiracy A and B, the victims’ deaths are overdetermined. But in Conspiracy A, the participants (absent Chooser) administer only twelve drops—which is two more drops than a lethal dose. In Conspiracy B, the participants (absent Chooser) administer 120 drops—which is 110 drops more than a lethal dose. The death of the victims in Conspiracy B is far more modally robust than the deaths of the victims in Conspiracy A. So, when Chooser adds another drop of poison in Conspiracy A, she makes more of a modal difference than she does when she adds another drop of poison in Conspiracy B. Some might think this matters morally, in which case we have another reason for thinking that Chooser should select Conspiracy B rather than Conspiracy A. But again, this doesn’t speak against the argument for authority-based accountability. This argument presents only one reason to refrain from furnishing a wrongful purpose to others. Though this reason is important, it can be outweighed by competing considerations, which is what happens in Baker’s Dozen and Conspiracy Choice.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

70

, ,  

2.4 Conclusion On the account I developed, when two individuals, P1 and P2, establish a division of agential labor in which P1 is the deliberator and P2 is the executor, P2 has the active agentive function of enacting particular ends belonging to P1. We can characterize this division of agential labor and the concomitant agentive function in terms of the protected status of P2’s reason to do as P1 says. P2’s conduct thereby has the purpose of enacting P1’s ends, and P1’s ends determine the purpose of what P2 does. Where that purpose is morally problematic, P1 is thereby accountable for furnishing that purpose. This dyadic example, I argued, serves as a building block for accountability in more complex forms of cooperative activity. Though this machinery is complex, the phenomenon it underwrites is perfectly pedestrian; it includes (I will argue) promises, agreements, requests, and shared acts. All are based on divisions of agential labor, even if there is no “leader” in the activity at issue. Suppose you and I agree to go on a hike together. In this shared activity, each is a party to a division of agential labor in which each counts as a deliberator and an executor. In agreeing to go for a hike together, I thereby have authority over you in that I have a practical claim against you that you go for a hike with me, and you thereby have authority over me in that you have a practical claim against me that I go for hike with you. The word “authority” might seem overblown in this example. All it means, though, is that each of us has a protected reason to act in accordance with the agreement we made, where each of us has the standing to release the other from that protected reason. Likewise, the claim that there is a “division of agential labor” in this example might seem inapt considering that we’re both doing the same thing—going for a hike. But by agreeing to go for a hike, I have vested in you the authority to decide what I am to do in that you have the normative power to demand that I hold up my end of the agreement; you have vested the same authority in me. Each of us is, then, both a deliberator and an executor. By refraining from releasing you from your commitment to me, I decide that you will hike with me; and by refraining from releasing me from my commitment to you, you decide that I will hike with you. Suppose, though, that the decision to go on a hike together was wrongful. Perhaps we were both slated to comment on a presentation at a conference at that time. Suppose the person we have wronged—the speaker—wants to know why we did what we did. I can give her the reasons I took there to be for my decision to go for a hike, but they would be incomplete. I’d also have to refer her to you, since you furnished a purpose for my hike. Likewise, if she asked you, you could give her the reasons you took there to be, but they too would be incomplete. You’d have to refer her to me, since I also furnished a purpose for your hike. We thereby furnish purposes for each other. Indeed, we might look to each other and think: “Why did

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

- 

71

we go for a hike?” And those reasons, in turn, affect how accountable I am for what you did, and how accountable you are for what I did. So, when we wrongly go for a hike together, I am accountable for the fact that what you did is wrong (without thereby vitiating your accountability) and you are accountable for the fact that what I did is wrong (without thereby vitiating my accountability). This is because each of us furnish for the other reasons insufficiently important to morally justify skipping out on our commenting-duties. The relevance of authority-based accountability becomes apparent when individuals are cooperating together in a way that causes an overdetermined wrong. In such cases, authority-based accountability provides a basis for thinking that an individual cooperant can be accountable for more than what is within her causal reach. Consider this example: Conference Hike A local conference has a dozen members in attendance, excluding the invited keynote speaker. During the lunch break, each of the dozen members agrees with one another to go hiking in lieu of attending the keynote session. As a result, no one attends the keynote session. Without at least two attendees the keynote will be canceled. Since no one attends, the speech is canceled. Each hiker bears, qua deliberator, authority-based accountability for what every other hiker does qua executor, in virtue of having furnished each with the purpose of going on a hike. That is, each hiker bears authority-based accountability for the actions of eleven other hikers, in virtue of having made agreements with each of them. The result is each hiker bears the authority-based accountability for what every other hiker does—and that accountability aggregates. So, each hiker bears authority-based accountability for what they together do. That is, each hiker bears authority-based accountability for the cooperative act of going on a hike together—a cooperative act that caused the keynote speech to be canceled. This means each hiker bears authority-based accountability for a harm that she herself did not cause and (we can assume) could not have prevented. It might seem that authority-based accountability cannot add up to much considering that each hiker furnishes one and the same purpose for every other hiker. Thus, even normatively, a given hiker seems to make no difference in that the purpose she furnishes seems otiose. But I argued that in such a case each deliberator furnishes for each executor a unique protected reason, thereby grounding a unique purpose, deriving from the deliberator’s own motivating reasons in favor of the action in question. Each hiker doesn’t just have a protected reason to go on the hike, simpliciter; rather that reason is directed in that it refers uniquely to each and every other hiker. The result is that each hiker does indeed make a difference to the purpose every other hiker has. As one of the hikers,

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

72

, ,  

I don’t simply have the purpose of going on a hike, but rather going-on-a-hikefor-you, going-on-a-hike-for-her, going-on-a-hike-for-him, and so on. Moreover, the protected reason grounding that purpose is decisive; thus, each hiker is severally accountable for what every other hiker did—or so I argued. This example shows how a dyadic case in which one individual has authority over the other, or in which they both have authority over each other, can serve as building-blocks for organizing progressively complex cooperative arrangements. In the Appendix, I consider how these various formal arrangements affect authority-based accountability. One upshot is that authority-based accountability is not necessarily transitive. P1’s authority-based accountability for what P2 does, and P2’s authority-based accountability for what P3, does not necessarily entail P1’s authority-based accountability for what P3 does, even if they are all acting intentionally toward one and the same cooperative end. Nonetheless, P1’s authority-based accountability for what P2 does will often result in P1’s authoritybased accountability for what P3 does. This is because P1 bears authority-based accountability for what P2 does in foreseeable furtherance of the promise she makes. If that includes enlisting P3’s assistance in fulfilling that promise, then what P3 foreseeably does in providing that assistance falls within the scope of activities for which P1 bears authority-based accountability. See, again, the Appendix for further discussion. I hope to have shown that in a division of agential labor the deliberator bears what I’ve called “authority-based” accountability for what the executor does qua executor. But how do we establish the requisite division of agential labor in the first place? I turn to this issue next.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

3 Establishing a Division of Agential Labor The account I developed helps us make sense of a common phenomenon: we “give” each other reasons. Suppose an assailant attacks an innocent who is consequently in need of immediate medical attention. There is a straightforward sense in which the assailant has “given” you a reason to aid the innocent. Here, the assailant creates circumstances serving as the manifestation-conditions for your disposition to act on a standing reason you have to lend aid.¹ But by “giving” a reason, I mean something else: an event in which the very fact that I have asked you to do something provides a basis for you to so act. Suppose a bodyguard promises to his employer that he will “rough up” whomever his employer wants and for whatever reason his employer wants. The employer, who accepts this promise, asks the bodyguard to accost an innocent who looked at the employer “the wrong way”. The employer thereby gives the bodyguard a reason to act accordingly, quite apart from the merits of accosting the innocent. That is, the employer’s directive has the function of “settling the matter” for the bodyguard. I argued that in such a case, the person giving wrongful reasons (the deliberator) can be accountable in a special way for the conduct of the person who accepts and acts on the wrongful reason given (the executor). This kind of reason-giving should also be contrasted with advice. Suppose P1 advises P2 to do ϕ, where ϕ is some wrongful act, as in the following case: Drug Theft Talker has learned of an easy way to steal the entire supply of narcotics from a drugdealer. This would effectively put the drug-dealer out of business, which would benefit Talker since he also sells narcotics. Talker, though, is unable to pull off the theft herself. So, she advises an acquaintance, Listener, to commit the theft. Talker emphasizes the ease with which the theft can be accomplished, the low risk that it involves, and the benefits of success. At no point does Talker request or demand that Listener commit the theft; Talker knows that any such move will likely dissuade Listener. Instead, Talker limits herself to presenting the reasons in favor of committing the theft. Based on Talker’s advice, Listener comes to appreciate the practical reasons in favor of the theft, which he subsequently commits.

¹ See (Enoch, 2011, pp. 4–5) and (Enoch, 2014, p. 299) for discussion of ‘triggering reasons’ versus ‘robust’ reason giving.

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0004

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

74

, ,  

In this case, Talker is not issuing protected reasons for Listener to commit the theft. Instead, her advice throws light on what Listener takes to be reasons favoring the theft.² Talker thereby affects the practical reasons Listener takes there to be, and potentially the explanatory reasons for Listener’s actions. Talker does so by convincing Listener that certain facts are normatively relevant to his deliberations. Listener accordingly adopts motivating reasons that he might not have otherwise adopted. Insofar as Listener commits the theft because of the advice Talker gave, Talker might be accountable for what Listener does as an intended outcome of giving that advice. But Talker does not bear authority-based accountability for what Listener does. In attempting to determine what prompted Listener to act in the way he did, his victim might indeed need to address the causal role that Talker played in Listener’s deliberations. But to ascertain the purpose of Listener’s action, Listener’s victim needn’t refer to Talker’s motivating reasons, since Listener’s conduct did not have the function of enacting those reasons; Talker’s advice did not serve for Listener as a protected reason to act accordingly. The point can be put differently: though the practical reasons Talker takes there to be might have causally influenced the reasons Listener took himself to have, they do not constitutively determine Listener’s purpose. Suppose, alternatively, that Listener does as Talker suggests not because Talker’s reasons convince Listener, but out of respect for Talker’s judgment. For Listener, the very fact that Talker suggests committing the theft, provides for Listener a protected reason to proceed accordingly—even though Talker did not take herself to be conferring a protected reason, and had no basis for thinking that Listener would interpret her suggestions that way. Does Talker bear authority-based accountability? No. Even if Talker’s motivating reasons constitutively determine the purpose of Listener’s subsequent conduct, Talker can’t be held accountable for this state of affairs, since by hypothesis she neither intended to confer a protected reason, nor had any basis for thinking that she would end up doing so. The upshot of Drug Theft generalizes. It is characteristic of advice that it does not yield protected reasons. And it is characteristic of advice that it does not yield authority-based accountability. In an interpersonal division of agential labor, the deliberator doesn’t merely causally influence the sorts of practical reasons the executor takes there to be. In addition, the deliberator controls what counts as the purpose of the executor’s conduct by constitutively determining the content of the executor’s protected reasons. In this respect, there is no “gap” between the deliberator’s motivating reasons and the purpose of the executor’s conduct; the deliberator can, “on the fly’, modify her own motivating reasons and thereby alter ² David Enoch calls this phenomenon, in which we reveal to someone a reason that was already there, “epistemic reason-giving.” He contrasts this with the kinds of reasons that authorities give by way of issuing commands: they create reasons ex nihilo. See (Enoch, 2011) and (Enoch, 2014).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

75

the purpose for which the executor is acting. This is in contrast to the adviser— once the adviser has drawn the actor’s attention to facts that the actor takes to be normatively relevant, the adviser no longer has any constitutive control over the purpose of what the actor does. When a deliberator, on the other hand, gives a wrongful reason by instructing the executor to commit a wrong which the executor consequently commits, the deliberator not only is a cause of what the executor does but also determines the purpose of what the executor does. That is, the executor has the agentive function of enacting the practical reasons that the deliberator takes there to be—i.e., the deliberator’s motivating reasons. Those motivating reasons thereby determine the purpose of the executor’s actions. As a result, the deliberator can be accountable not just for serving as a cause of what the executor does, but also for a wrongmaking feature of what the executor does: the wrongful purpose for which the executor acts. It is accordingly because of the deliberator that the executor’s actions count as wrongful. But how does someone get to be a deliberator or an executor? That is, how do we establish a division of agential labor? I argued that persons P1 and P2 establish a division of agential labor when P2 makes and P1 accepts a promise to do ϕ. The protected status of the reason that the promise yields can be recharacterized in terms of a division of agential labor. But do promises in fact yield protected reasons? This is an important question since there is no authority-based accountability if the deliberator does not furnish a protected reason for the executor. (Recall from section 1.1 that the protected status of the reason that the deliberator confers upon the executor is what grounds their respective agentive functions in a division of agential labor.) In what follows, I first consider whether, on the various accounts of promising on offer, promising yields protected reasons. I will argue that they do. Then I will consider other ways of establishing an interpersonal division of agential labor: by making requests, by forming agreements, and by engaging in shared action. Even immoral promises, I will argue, confer reasons on the promisor—albeit practical rather than moral reasons. With respect to requests, I will argue that even though, by its very nature, a mere request generally does not require compliance, requests will standardly yield protected reasons. This means the individual presenting the request can bear authority-based accountability for the wrongful conduct of the individual complying with the request. I will also argue that agreements yield protected reasons. Obviously, this follows if agreements are nothing more than interdepending promises. But I will argue that agreements yield authority-based accountability even if we do not analyze agreements in terms of promises. Finally, I will argue that shared action yields protected reasons for all the parties involved. In arguing for this view, I will consider two competing accounts of shared action—Margaret Gilbert’s and Michael Bratman’s. It will be relatively easy

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

76

, ,  

to show that on Gilbert’s account shared action yields protected reasons. Bratman’s account will present more of a challenge since he denies that shared action necessarily yields obligations to act accordingly. I will argue that though, on his view, shared action need not yield protected reasons, it typically will—which means that the parties will bear authority-based accountability for one another’s actions. Throughout this chapter, I will focus on examples involving only two persons. I do this to keep the discussion tractable. I end Part I of this book by returning to more complex cases of cooperative action.

3.1 Promises Recall that one way to establish a division of agential labor is by accepting a promise. If you promise to me that you will do ϕ, and I accept that promise, then I have authority over you with respect to ϕ, by deciding whether or not to hold you to that promise. If I decide to hold you to that promise, then I have a practical claim against you that you do ϕ, where that claim provides you with a protected reason to do ϕ.³ So, if I hold you to the promise you made, then that in itself provides you with a first-order reason to do ϕ, and a second order-reason to exclude from your deliberations a certain range of competing considerations. Put differently, we have established an interpersonal division of agential labor in which I count as the deliberator, and you count as the executor. There are, though, multiple competing accounts of promises developed in the literature. It is not immediately clear that all of them yield protected reasons. In what follows, I consider three of the most prominent accounts of promising, and I argue each of them either entails or implies the view that promises confer protected reasons. But to show as much, I will need to unpack each of these three accounts. I will end by considering whether and how immoral promises confer reasons.

3.1.1 Normative Power Views According to normative power accounts of promises, we have the ability to generate obligations by promissory utterances, where this ability is a species of a more general ability we have: to alter our normative relationships with others.⁴ On

³ See (Hart, 1982, p. 255) and (Raz, 1986, pp. 35–37). Also see Owens (2008), (2012) for an extended discussion of promise-making as conferring authority. ⁴ Proponents include (Hart, 1955), (Raz, 1977) (Raz, 2014), (Watson, 2004), (Owens, 2008), (Owens, 2012), (Shiffrin, 2008), (Shiffrin, 2012), and (Rosati, 2011).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

77

normative power accounts, our powers to obligate are unmediated, rather than mediated by convention or by evoking expectations in others. Normative power accounts ground the power in terms of our interests, the appeal to which divide such accounts into two types: authority-based and autonomy-based. David Owens has done the most to develop authority-based accounts of promising.⁵ He argues that apart from familiar non-normative interests—such as an interest in how others behave toward us—we also possess normative interests—which are interests in how others ought to behave toward us. These normative interests ground a wide range of moral practices, which includes promising. On authority-based versions of normative power views, promisors, specifically, lend promisees “the right to decide” what we do.⁶ On this view, if I promise to do something, the authority-interest underwriting my obligation to act accordingly is not an interest in avoiding the harms that breaking the promise will cause. Rather, the promissory obligation is grounded in an interest we have in regulating the normative status of the act promised. We have an interest in determining whether we possess practical authority over others. This account explains a variety of features of a promise I make to you (such as the validity-conditions of promises). Alternatively, on autonomy-based versions of normative power views, promises are grounded in our status as autonomous agents. Seana Shiffrin, the primary proponent of this view, argues that an autonomous life requires the ability to engage in meaningful, moral relationships with others. These relationships depend, in turn, on our ability to formulate binding promises, construed as the ability to intentionally create an obligation through the exercise and expression of will alone.⁷ But why believe that meaningful moral relationships require the ability to formulate binding promises? On Seana Shiffrin’s account, when I make a promise to you that I will do ϕ, I transfer to you my right to decide whether I should ϕ. By doing so, I convey my willingness to abdicate my moral right to act unilaterally, in that I am no longer permitted to base my decision on my assessment of the merits of ϕ per se. I thereby acknowledge your involvement in the situation by authorizing you to make the decision for me regarding ϕ by empowering you to demand ϕ of me or to otherwise release me from that obligation. Merely expressing my intention to do ϕ leaves you a bystander normatively at the mercy of my deliberative whims. A promise provides a framework for building trust, which, as Shiffrin points out, is essential to moral relationships. The upshot is that on both authority-based and autonomy-based versions of normative power views, we can analyze promises as yielding protected reasons. ⁵ See (Owens, 2012). ⁶ (Owens, 2012, p. 74). ⁷ See (Shiffrin, 2008). See also Joseph Raz, for whom promissory obligations are grounded in the value of voluntary special bonds (Raz, 1986, pp. 173–176).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

78

, ,  

This is because, on both versions, the function of my promise to you that I will do ϕ is to grant you the practical authority to decide whether I am normatively required to do ϕ, by transferring to you the right to make that decision for me. Hence your instructions to me regarding ϕ settle the matter for me with respect to whether I should ϕ. This is, after all, the point of normative power views. On authority-based versions, our ability to normatively empower others via promising serves a general interest we have in selectively regulating the normative status of what others do. On autonomy-based versions, our ability to normatively empower others via promising falls out of our status as persons capable of establishing trusting relationships with others. Though I believe that normative power views of promising are correct, and that competing views—notably, conventionalism and expectationalism are mistaken— I do not want my account to rest on a controversial view of promising. For this reason, I will argue that if conventionalism and expectationalism are correct, they too yield accounts in which promising yields protected reasons.

3.1.2 Conventionalism According to conventionalism, promises are based in rule-governed practices facilitating cooperation and coordination, thereby conferring benefits to those living in the society in which the rule-governed practice is in force. Conventionalists argue that violating promises weakens the convention of promising as a whole. Insofar as we have reasons to sustain the coordinative and cooperative benefits of the convention, we have a reason not to violate promises.⁸ Conventionalism has come under sustained criticism. T. M. Scanlon points out that conventionalism requires a rule-governed practice before promising is possible, which precludes promises between those who lack a shared convention. Scanlon goes on to argue by example that in a state of nature, binding promises are perfectly possible even absent a shared institution of promising.⁹ He also argues, convincingly, that conventionalism fails to accommodate the directed nature of ⁸ Conventionalists, though, need to explain how the social value of the convention as a whole underwrites an agent’s obligation to keep specific promises. Contractarians attempt to fill this lacuna by arguing that the coordinative and cooperative benefits of sustaining the promising-convention yields reasons for individuals to keep their promises. For example, David Gauthier argues that to the extent we are rational we will keep our promises since doing otherwise will vitiate our ability to resolve prisoner’s dilemmas (Gauthier, 1986, p. 167). Following Richard Brandt, Rule Consequentialists such as Brad Hooker (Hooker, 2011) similarly argue that individual promissory obligations fall out of the coordinative and cooperative benefits of the promising-convention; insofar as the rules composing the convention yield the best consequences when followed, the duty to obey them is not merely pragmatic but moral. Contractualists, such as John Rawls, argue that the duty to uphold the convention of promising includes a duty of fairness to refrain from free-riding on that institution by violating promissory obligations (Rawls, 1971, p. 112). ⁹ (Scanlon, 1998, p. 297ff).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

79

the wrong in which breaking a promise consists. On conventionalism, the wrong of violating a promise consists in the harm done to a social convention and to the individuals benefiting from it. But this fails to accommodate the wrong done to the promisee specifically, who is specially positioned to rebuke the promisor.¹⁰ Suppose, though, that in spite of these criticisms¹¹ that a version of conventionalism is correct. If so, I believe it too can accommodate the view that promises yield protected reasons. Presumably, the convention in question will yield coordinative and cooperative values only if the promissory obligations are categorical in the following sense: when I make a promise to do ϕ for you, and you demand that I act in accordance with that promise, the basis for complying lies not in the merits of doing ϕ but in the very fact that you have a practical claim against me that I so act. It is no surprise that under conventionalism the reasons we have to abide by promises do not derive from the merits of the act in question. Suppose, by reductio, that when deciding whether to abide by a promise, the promisor bases that judgment not on the requirement of keeping promises per se, but on whether she deems what is promised choiceworthy. Under these conditions, the promise itself does no normative work—it does not add to the reasons the promiser has. The promisor retains the prerogative to unliterally rescind the ostensible promissory obligation she has, should she decide that the content of the promise is not choiceworthy. Clearly, then, promise-making under this impoverish regime does not promote coordinative and cooperative values any more than stated intentions do. If promises are to serve those ends, then we must have a reason to abide by them qua promises. But recall that for promises to yield protected reasons, they must yield not only a first-order reason to abide by them per se, but a second order reason to exclude from deliberations certain competing considerations. Doing so seems to facilitate the coordinative and cooperative values which promises serve to promote. Suppose, by reductio, that promises do not have exclusionary force. So, when I make a promise to you, we both know that this does not settle the matter for me; that is, I do not thereby take myself to have a decisive reason to abide by the promise, in that, even absent new information, I might, upon evaluation of the reasons I take myself to have, accordingly decide to violate the promise. To the extent that, under this hypothetical regime, promises do not yield normatively decisive reasons, there is substantially less basis for thinking that any given promise guarantees compliance. Thus, such a regime would less effectively promote the coordinative and cooperative values which promises are supposed to serve. If promises are to serve those ends, then the reasons to abide by them must

¹⁰ Some have attempted to resolve this problem by combining elements of conventionalism with expectationalism. See (Kolodny & Wallace, 2003). ¹¹ And more besides: see, e.g., (Shiffrin, 2008) and (Owens, 2012).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

80

, ,  

be exclusionary. It follows, then, that conventionalism can accommodate the exclusionary character—and thus the protected nature—of promises.

3.1.3 Expectationalism When we make promises, we engender certain expectations in those to whom the promises are made. Acting contrary to these expectations generally causes harm. The reasons we have to avoid this harm are what ground our promissory obligations and are what explains the wrong of violating those obligations, according to expectationalist accounts. Although there are various versions of expectationalism,¹² I will focus on T. M. Scanlon’s approach, which is the most comprehensive and influential of those on offer. On Scanlon’s account, promissory obligations are grounded in more basic moral obligations to refrain from unfairly manipulating other people. If I make a promise to you that I will do ϕ, I thereby lead you to expect that I will do ϕ; if I should subsequently violate that promise, I thereby not only frustrate your expectations, but in doing so, I have deceived you. Given that deception is morally verboten, we have a moral duty to keep our promises. Scanlon grounds promissory obligations in terms of what he calls the “Principle of Fidelity” or Principle F: If (1) A voluntarily and intentionally leads B to expect that A will do X (unless B consents to A’s not doing so); (2) A knows that B wants to be assured of this; (3) A acts with the aim of providing this assurance, and has good reason to believe that he or she has done so; (4) B knows that A has the intentions and beliefs just described; (5) A intends for B to know this, and knows that B does know it; (6) B knows that A has this knowledge and intent, then, in the absence of special justification, A must do X unless B consents to X’s not being done.¹³

On this view, there is nothing special about promises per se in that the wrong of violating a promise is no different from the harm of misleading others by lying or deceiving—in all these cases, we engender and then violate trust.¹⁴ A problem for expectationalism is that it seems viciously circular.¹⁵ But supposing this challenge can be overcome, does expectationalism yield the view that promises yield protected reasons? As it stands, the answer seems to be “no’. ¹² See (McNeilly, 1972), (MacCormick, 1972), (Anscombe, 1981), (Foot, 2001, pp. 5–24), (Mason, 2005). ¹³ (Scanlon, 1998, p. 304). ¹⁴ See (Mason, 2005). (Freiderich & Southwood, 2011) defend a version of expectationalism by arguing that what grounds promissory obligations is not necessarily the presence of the trust that the promise engenders, but instead the “invitation” to trust implicit in the promise. ¹⁵ By making a promise to you, I engender in you the trust that I will act in the way that I promised. You should trust me presumably because you believe that I have placed myself under a moral obligation

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

81

According to the Principle of Fidelity, when I make a promise to you that I will do ϕ, the reason I have to fulfill that promise just is a reason to refrain from deceiving you. The reasons we have to refrain from deceiving others are not protected. They are instead just ordinary moral reasons. We might, however, augment expectationalism in the following way. What animates expectationalism is the assurance that promises implicitly provide. Assurances range in strength, of course. Promises provide assurance of a particular and peculiar strength. Arguably, my promise provides you assurance of the strength characteristic of promises in general, only if by making it I engender not only the expectation that I will fulfill the promise, but also the expectation that the promise will provide me with a particular kind of reason. Specifically, I will treat your claim against me that I fulfill the terms of the promise as a decisive reason, in and of itself, to act accordingly, thereby settling what I should do independent of whether what happens to be promised is choiceworthy (absent new evidence to the contrary). That is, you can reasonably expect that I will treat the promise in a particular way: it yields for me a protected reason to act accordingly. You can reasonably expect me to treat the promise in this way, precisely because doing so provides you with the assurance of the strength characteristic of promises in the first place. In this way, expectationalism can and should accommodate the view that promises yield protected reasons.

3.1.4 Immoral Promises I have argued that several of the most popular accounts of promises can accommodate the claim that promises yield protected reasons. But what about cases in which the promise is immoral? Consider again the example Olympic Sabotage (chapter 2). Recall that Goon makes a promise which Contender accepts: that Goon will maim Victim thereby increasing the chances that Contender will receive first place in an upcoming competition. Clearly, this is an immoral promise. Is there any sense in which this promise is binding? Does it even count as a promise?

by having made the promise. Provided you also believe that I comply with my moral obligations, you have a good reason to believe that I will do as I promised. See (Robins, 1976), and (Prichard, 1940). On this story, putting myself under an obligation explains the trust I engender. But according to expectationalism, the trust I engender is what explains the obligation I have. That is, trust is the explicans, not the explicandum. The problem for expectationalism, then, is that my promissory obligation cannot serve as a reason for you to trust me, since this would make the obligation logically prior to the trust itself, whereas according to expectationalism, the trust is what grounds the obligation. So, when you try to find a reason to trust me, you cannot invoke the fact that my promise puts me under an obligation. A social convention governing promises might serve as the basis for trusting promisors; but expectationalists such as Scanlon deny that conventions are necessary to explain promissory obligations. See (Kolodny & Wallace, 2003).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

82

, ,  

This is a fraught issue. The most popular strategy is to deny that there is any sense in which immoral promises yield obligations.¹⁶ This is not to say that immoral promises are not possible, but rather that such promises yield no obligations.¹⁷ Such a view seems to cut at the heart of the argument for authority-based accountability. After all, if Goon’s promise to Contender is not binding, then that promise seems to yield no reasons—and a fortiori, no protected reasons—in favor of enacting it. Absent any such protected reason, there is no division of agential labor between Goon and Contender, in which case Contender bears no authority-based accountability for what Goon does. More generally, authority-based accountability is virtually useless if it cannot accommodate immoral promises. Contrary to appearances, though, immoral promises pose no problem for authority-based accountability. There’s no denying that there are decisive moral reasons for Goon not to commit the assault; those reasons override or silence whatever moral force the promise he made would have otherwise had. Hence, Goon has no moral reasons to commit the assault. When Goon promises that he will assault Victim, and when Contender accepts that promise, Contender thereby enjoys practical rather than moral authority over Goon. What does it mean for Contender to have “practical authority” over Goon? As explained in chapter 2, it means that the practical reasons Contender takes there to be in favor of committing the assault have the function of guiding Goon’s conduct, and his conduct has the function of enacting the practical reasons Goon takes there to be. That is to say, the practical reasons Contender takes there to be constitutively determine the purpose of Goon’s conduct. They establish this division of agential labor—in which Contender counts as a deliberator and Goon counts as an executor—by way of the protected status of the promise that Goon makes and Contender accepts. And though the protected status of the promise fails to protect against decisive moral reasons for Goon to refrain from committing the assault, it succeeds in protecting against competing practical reasons. It is an unfortunate fact of life that we can have practical reasons to act

¹⁶ See, for example, (Thomson, 1990, p. 12), (Scanlon, 1998, pp. 295–327), (Shiffrin, 2008), (Kolodny & Wallace, 2003), (Tognazzini, 2007), (Watson, 2009), (Rawls, 1971, p. 305). See also (Prichard, 1940) and (Hart, 1958). For general discussion of immoral promises, see (Altham, 1985) and (Raz, 1977, p. 212). For a discussion of the view that promissory obligations are prudential, see for example (Baier, 1985, pp. 174–206), (Baier, 1994, pp. 110–120), (Sheinman, 2008). A related strategy claims that promises yield obligations concomitant with the type of promise proffered, so that moral promises yield moral obligations, legal promises yield legal obligations, and so on. See, for example, (Raz, 1977) and (Owens, 2008). ¹⁷ Proponents of this view would need to explain, though, why the promisee still seems to have a basis for complaint against the promisor should she violate an immoral promise. (Anscombe, 1978) and (Altham, 1985) make this point. Those who think that immoral promises are in some sense morally binding might claim that Goon has a pro tanto, directed obligation toward Contender. This is the sort of strategy that Searle and Gilbert seem to adopt (see (Searle, 2001, p. 194) and (Gilbert, 2011)). I will not discuss it here because it introduces no special challenges for authority-based accountability.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

83

in immoral ways. The decisive moral reasons there are against committing the assault do not touch the practical reasons that the promise confers. The result is that Contender retains authority over Goon—albeit prudential rather than moral authority—in that Contender controls Goon’s purpose. Because the authority is “merely” practical, Contender is not only morally permitted but morally obligated to violate the promise he made. But if he doesn’t—if he complies with the practical authority Contender has over him—then Contender bears authority-based accountability for what Goon does. Taking a step back, the idea that a deliberator can have this sort of practical authority over an executor should come as no surprise given the division of agential labor between them. A basic thesis of this book is that personal agency, normally wrapped up in a single person, can be in practice “distributed” across multiple individuals. The result is that it is possible for one person to have practical authority over another insofar as the deliberator can settle what the executor is practically supposed to do and why. So, when we say that the deliberator has a claim against the executor that he do some wrongful act φ, the claim is not moral, but rather practical. The deliberator has practical authority over the executor in the same way that your prior self can have practical authority over your present self. If yesterday you settled on doing φ today, then, absent a relevant change in circumstances, your past self has a claim— albeit a practical rather than a moral one—that you do φ today. A deliberator can have, analogously, a practical claim over an executor. The upshot, then, is that immoral promises do not threaten the argument for authority-based accountability, because the authority that the deliberator has over the executor is fundamentally practical rather than moral. Suppose, Goon (wrongfully) acts on the protected practical reason to fulfill the promise. The result is that Contender is accountable for a wrong-making feature of what Goon does—viz., the purpose for which he acts. When the victim asks Goon for a full account of what he did, Goon would have to point the victim to Contender in elucidating the purpose of his own actions. The fact that the promise Goon made was patently immoral does not vitiate Contender’s authority-based accountability for what Goon does. The same goes not just for promises, but for requests, agreements, and commitment in shared action—or so I will argue. Each case can be analyzed in terms of a division of agential labor, in which a deliberator confers upon an executor a protected reason, and thus, a purpose, which, if acted upon, inculpates the deliberator at the bar of authority-based accountability—even in cases where it is immoral for the executor to act in that way.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

84

, ,  

3.2 Requests and Agreements So far, I have argued that promises—including immoral promises—confer protected reasons (even if only non-moral, practical reasons). A mere request, though, can do the same thing, as can an agreement. In what follows I analyze each in turn.

3.2.1 Requests Suppose that I want you to do ϕ for me. I accordingly try to solicit a promise from you. You indicate that you’ll do ϕ, but then add the caveat that you “make no promises.” In this case, I do not take myself to have protected reasons to do ϕ; as a result, we have not established a division of agential labor. To see why not, consider again a case of intrapersonal, diachronic decision-making. Suppose I decide the following: “I will do ϕ unless I change my mind.” This seems to be no decision at all. It is tantamount to deciding to do ϕ unless I don’t do ϕ.¹⁸ Should I decide on a whim not to do ϕ after all, I do not transgress any norm of rational deliberation. This is because my ostensible decision failed to settle whether I should do ϕ by making it so that literally any reason—or even no reason at all—can be normatively sufficient for not doing ϕ. Now return to the interpersonal case: you say that that you “make no promises” with the intention of indicating that it is not settled for you whether you will defer to me. To establish a division of agential labor, my reasons must normatively guide your conduct. Your relationship to my decision is analogous to your relationship to your own “decision” in the intrapersonal example. In both cases, the decision settles nothing. So, if the caveat “I make no promises” signals that you are normatively free to do whatever you want, then you have indicated that I have no authority over you, and that my request fails to yield protected reasons—and thus, a purpose—for you. But we might imagine a less extreme case: your caveat that you “make no promises” might mean that my request that ϕ conditionally settles whether you will do ϕ. That is, “I make no promises” might mean “I make no unconditional promises.” Suppose you make it clear that you will treat my request as settling the matter for you, but that if complying proves overly onerous, you will opt out. Suppose I acknowledge this. Does my request that ϕ yield protected reasons for you? It does. To see why, repair again to the intrapersonal case. Suppose I make this decision: I will do ϕ unless it proves onerous. This does in fact settle something for me: that I should ϕ under the condition specified, i.e., given that

¹⁸ This might suggest that self-promises are impossible. But see (Rosati, 2011).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

85

doing ϕ is not onerous. So, for example, if it turns out that doing ϕ is not onerous, there is no basis for me to re-deliberate whether I ought to do ϕ; the decision for those conditions has already been made. The same goes for the interpersonal case; when you accept my request, it yields protected reasons for you under the condition specified. So long as doing ϕ is not overly onerous, there is no basis for you to re-deliberate whether my request settles the matter for you. The decision has already been made. There are, then, several ways to read the caveat “I make no promises” when you accede to my request regarding ϕ. Where doing so indicates a mere conditional promise to defer (where the conditional has narrow-scope), the caveat is perfectly compatible with establishing a relationship in which it is agreed that your request yields protected reasons for me. But where the caveat is meant to indicate that you have not decided even conditionally to defer to me, then you have thereby rejected any such protected reasons. So even if you should subsequently happen to do ϕ, I bear no authority-based accountability. So far in this section I have considered cases where you respond to a request by explicitly disavowing any protected reasons that such a request would yield if you had responded to it by agreeing to comply. But there is the separate issue of whether the request itself—independent of your reaction to it—yields a protected reason for you to comply. In addressing this question, it is instructive to contrast requests with commands. On Raz’s view, the function of a request I make to you, is to provide you with a reason to comply where the request itself serves as that reason.¹⁹ You are thereby permitted to decide whether to comply. Should you ultimately refuse the request, you do not thereby wrong me.²⁰ (That is unless the content of the request itself requires compliance, as is the case if I request easily rendered life-saving aid from you.) The purpose of a command, on the other hand, is to impose an obligation by virtue of having made the command.²¹ When I command to you that you do ϕ, the imperatival character of the command itself has the function of providing you with a reason to do ϕ, independent of whether ϕ itself is choiceworthy. In addition, the command functions by excluding from your practical deliberations competing reasons that would normally weigh against compliance. The upshot here, then, is that on Raz’s view, commands yield reasons in their favor, whereas requests do not. Other thinkers, however, demur. Owens, in particular, argues that commands and requests both yield protected reasons. He says: If I am requested to do something, then it makes sense for me to do it without my deliberating about whether to do it, without my weighing the pros and cons. The ¹⁹ See (Raz, 1986, pp. 35–36) and (Raz, 1999, pp. 83, 100–101). ²⁰ See (Feinberg, 1970, pp. 4–5) and (Darwall, 2011). ²¹ (Raz, 1999, p. 193).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

86

, ,   point of someone’s asking me to give them a light is precisely to by-pass the issue of the desirability of my giving them a light, is precisely to take the matter out of my hands. And the more urgent the request, the wider the range of countervailing reasons that it makes sense for me to exclude from my deliberations.²²

In this respect Owens follows Shapiro who writes: “If, for example, Able asks Baker to help him with his homework, Able’s request gives Baker a reason to act that is independent of whether Baker ought to help Able. It is the fact that Able asked, rather than what he asked, which gives Able a reason to act.”²³ The difference between a request and a command, on these accounts, is that disobeying an order is verboten whereas disobeying a request is not. At first it might seem that I needn’t take a stand on whether Raz’s or Owens’s accounts of requests is correct. It is built into Owens’s analysis of requests that they yield protected reasons. Thus, if you indicate to me that you have decided to comply with my request, then that just means you take my request to settle the matter for you. On Raz’s account, if you accept my request by indicating to me that you have decided to comply with it, then you thereby grant me the practical authority to demand of you that you comply. If I request that you give me a ride to the airport, you needn’t comply; but once you indicate that you accept my request, I can thereby legitimately demand compliance should you do otherwise. In this respect, accepting a request is or is tantamount to making a promise. But the issue isn’t whether agreeing to abide by a request yields a protected reason. The issue, rather, is whether the request itself yields a protected reason to comply. And the answer, it seems to me, is “yes”. I suggest that there is a sense in which we all have a standing authority over one another whereby we can confer protected reasons simply by making a request. The authority is practical, not moral, in that there is nothing morally wrong with refusing such a request (absent attendant moral reasons to comply, as in a case when the request is to call an ambulance following an accident). And the standing authority we have yields protected reason that are only weakly exclusionary; the result is that you can refuse most requests without necessarily acting irrationally. So, if I request that you defer to me regarding ϕ, I thereby provide you with a protected reason to comply—albeit one that you can refuse without acting immorally or irrationally. Should you nonetheless subsequently choose to comply with my request, you thereby undertake the function of enacting my practical reasons which have the concomitant function of normatively guiding your conduct. That is, in virtue of soliciting your assistance and in virtue of complying with my request, we establish a division of agential labor in which I count as the deliberator and you count as the executor. The result is that I thereby bear

²² (Owens, 2012, p. 85).

²³ (Shapiro, 1998, p. 493). See also (Marmor, 1995, p. 355).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

87

authority-based accountability for what you do, irrespective of my causal influence on your conduct. Having argued that we can establish divisions of agential labor via both promises and requests, I now turn to agreements.

3.2.2 Agreements We sometimes use the verbs “agree” and “promise” interchangeably, as in “You agreed to do the laundry!” and “You promised to do the laundry!” Still, agreements might be thought to have a “jointness” to them that promises lack— agreements are shared in a way that promises are not. In standard cases, an agreement between me and you binds both of us, whereas a promise binds only one of us. It is no surprise, then, that some have analyzed agreements in terms of interdependent promises. Suppose we agree that I will wash the dishes and you will dry them. This might be analyzed as a pair of promises: I make a promise to you, and you make a promise to me. But such promises will together compose an agreement only if they interdepend in a particular way. Perhaps, then, the promises must be mutually conditional. I make this promise to you: “I will wash the dishes if you (promise to) dry them.” Mutatis mutandis for your promise to me. Alternatively, the content of the promises might be unconditional while the promises themselves motivationally interdepend, in that I promise because you do, and vice versa.²⁴ Regardless of whether the interdependence is located in the conditional content of the promises, or in the motivational conditions for promise-making, this strategy is to analyze agreements as joint promise-making. In this respect, the strategy invokes theories of shared agency in an analysis of agreements. Clearly, if agreements are just appropriately related promises, and promises yield protected reasons (as I have argued), then agreements do so as well. Several theorists, however, deny that agreements are composed of interdependent promises. For example, several legal theorists deny that legally binding contracts can be composed of interdependent promises.²⁵ Though their focus is on legally binding contracts, their arguments seem to generalize to agreements simpliciter. Margaret Gilbert, though, does the most to develop the argument that agreements are not interdependent promises. On her view, an analysis of agreements succeeds only if the agreement under analysis possesses the following three features. First, the agreement must give each party an obligation to do her part of the agreement. Second, the agreement must

²⁴ This is Sheinman’s strategy in (Sheinman, 2011). See also (Bach, 1995). ²⁵ See for example (Smith, 2004, pp. 180–181), (Penner, 1996, p. 326), (de Moore, 1987, p. 122), (Black, 2004).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

88

, ,  

give these obligations simultaneously. Third, these obligations must interdepend. Attempts to analyze agreements as promises fail to satisfy these three conditions together, Gilbert argues. No exchange of promises, no matter how complex, will yield simultaneous, unconditional, interdepending obligations.²⁶ Gilbert consequently analyzes agreements as joint decisions.²⁷ I will argue, though, (in section 3.3.1) that Gilbert’s account of shared agency yields protected reasons for the parties involved. So even if Gilbert’s criticisms of agreements-as-promises is correct, it still follows that agreements yield protected reasons. So far, I’ve canvassed views that analyze agreements as interdepending promises. But some views proceed in the opposite direction by characterizing promises in terms of agreements. Promises, after all, cannot be established unilaterally (barring promises to oneself). Rather, they consist of an offer which in turn requires acceptance. In this respect they resemble agreements. So, when I say “I promise to wash the dishes,” I do not thereby promise to wash the dishes; instead, I am making an offer the content of which is either to wash the dishes or to promise to wash the dishes. Only if and when you accept this offer either explicitly or implicitly (depending on the circumstances) is the promise established.²⁸ Acceptance requires intentionally communicating consent pertaining to the offer. This is because my promise to you imposes upon me a directed duty to you and it affords you with a corollary claim-right against me, which requires your consent.²⁹ On this view, an agreement is nothing more than an offer which, once accepted, establishes the agreement.³⁰ The offer I make to you, once accepted, yields protected reasons to comply. The result is that we can establish divisions of agential labor via such agreements, which in turn grounds authority-based accountability in cases where, in accordance with the offer you accept, you do something morally problematic. I now turn to the fourth and last of the primary ways to establish a division of agential labor—via shared action.

3.3 Commitments in Shared Action Shared actions often comprise interpersonal divisions of agential labor. There are several senses in which action can be “shared”. If I am brushing my teeth, as are you, there is a sense in which this action is shared in that we’re both engaged in tokenings of the same activity-type. But that’s not the relevant sense of sharing ²⁶ See (Gilbert, 2006, p. 220), (Gilbert, 2000, p. 61), (Gilbert, 1996, p. 328). See also (Mintoff, 2004, p. 50). ²⁷ See (Gilbert, 1996, pp. 313–338), (Gilbert, 2006, pp. 215–237), (Gilbert, 2011). ²⁸ See (Owens, 2012, p. 228). ²⁹ See, for example, (Thomson, 1990), (Owens, 2012), (Watson, 2009, pp. 156–157). ³⁰ See also (Black, 2007).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

89

action. We share an action in the relevant sense when we coordinate and combine our actions in furtherance of doing something together, as when we walk together or paint a house together (to take canonical examples from Gilbert and Bratman). In such a case, the shared action is a result of a shared intention, which itself supervenes on individual intentions interdepending in the right way. (It is hard to be more specific about what shared actions are without committing ourselves to particular and controversial accounts of shared action.)³¹ In what follows, I focus on the two most influential accounts of shared action— the accounts developed by Margaret Gilbert and by Michael Bratman. It is not my goal to claim that all instances of shared action yield protected reasons for the parties to the shared action. There will be some instances of shared action in which no participant bears authority-based accountability for what any other participant does. My goal here is instead to present the conditions under which the commitments characteristic of shared action yield authority-based accountability. My conclusion will be that, typically, shared action includes commitments yielding mutually protected reasons among the actors, which in turn means that the participants in the shared action will end up furnishing purposes for one another. Where that purpose is wrongful, the actors end up mutually inculpating one another, given authority-based accountability.

3.3.1 Margaret Gilbert On Gilbert’s account, shared activity requires that each participant take on a “joint commitment” to “act as a body” as a result of which they form a “plural subject’.³² I will use an example to explain what each of these terms means. Suppose you and I agree to bake a cake for our colleague’s potluck. Each of us does not individually agree to bake our colleague a cake; that would result in two cakes. Instead, we each individually agree to make it the case that we bake a cake together. It is in this respect that we “act as a body” —we coordinate our efforts in furtherance of baking the cake as would a single individual. By doing so we together form a “plural subject” —an entity to which intentional action and propositional attitudes can be attributed.³³ On this account, a joint commitment is not the aggregate of a commitment I make to you and of a commitment you make to me.³⁴ Instead, a joint commitment is a commitment we both become party to, when I express my willingness to bake the cake with you and you express your willingness to do the same. Though

³¹ For helpful recent discussion see (Blomberg & Hindriks, 2020). ³² (Gilbert, 2006, p. 134). ³³ (Gilbert, 2006, p. 145). ³⁴ For helpful discussion of the role of commitments in collective responsibility more generally see (Arruda, 2020).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

90

, ,  

the commitment is held by a plural subject, this does not mean that the commitment is held by a “single center of consciousness” or a “distinctive form of ‘subjectivity’.”³⁵ Instead, the commitment is held, simply, by those bound to it— me and you. On Gilbert’s view, joint commitments entail a normative relationship between the individuals who are party to the commitment. By forming a joint commitment, each of us is obligated to do his or her part in furtherance of baking the cake together. And each of us has a practical claim against the other that he or she do the same. If I neglect to live up to those obligations, you would be justified in rebuking me.³⁶ Rescinding the joint commitment or otherwise releasing any individual from the requirements of participation requires unanimous concurrence from all the persons party to the commitment.³⁷ Gilbert makes clear that relevant obligations are “directed” in that if I fail to do my part in baking the cake, you have an agent-relative complaint against me; but it isn’t a moral complaint.³⁸ The relevant obligations are not moral obligations.³⁹ Instead, in emphasizing the directedness of these obligations, Gilbert analyzes them in terms of ownership: the obligation I have to do my part in baking the cake entails that only you are “owed” the obligated act.⁴⁰ This helps explain why only you can release me from my obligations to act in furtherance of the joint commitment—because only you can waive your claim on my contributions. The result is that agents party to a morally wrongful shared activity have claims against each other that they do their part even though these claims generate no moral reasons in favor of doing their part. It seems, then, that on Gilbert’s view the merits of the joint commitments do not affect my obligation to you that I act in accordance with the joint commitment; neither does it affect your practical claim against me that I act in that way. By agreeing to be part of a joint commitment, I thereby grant you authority over me which grounds a claim that you have against me: the claim that I do my part in the shared activity. Once I have agreed to be party to the commitment, I have a reason to act accordingly. Moreover, your practical claim against me that I do my part gives me a reason to exclude from the zone of deliberation certain contrary reasons that might otherwise be relevant. For example, the fact that I do not feel like doing my part ought not to serve in my deliberations as a reason against doing my part. The result is that once I agree to do my part, no further deliberation is practically required from me in the absence of new information. ³⁵ (Gilbert, 2006, p. 134). ³⁶ (Gilbert, 1990); (Gilbert, 1989, pp. 162, 409, 411). ³⁷ (Gilbert, 1999); see also (Gilbert, 1990). Gilbert uses the mutual obligation condition to criticize accounts that analyze shared activity in terms of “personal intentions” (Gilbert, 2008), such as that defended by Bratman (see below). ³⁸ (Gilbert, 2009), (Gilbert, 2008, p. 497). For more on the “bipolar normativity” characteristic of directed obligations, see (Darwall, 2004). See also (Roth, 2004) who characterizes the obligations characteristic of shared activity in terms of “contralateral commitments.” ³⁹ (Gilbert, 2009, p. 178). ⁴⁰ (Gilbert 2008, 497), (Gilbert, 2009, p. 183).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

91

Of course, as a matter of morality, more deliberation would be required if the shared activity or my role in it is morally wrongful. But there would be nothing practically problematic in treating the matter as settled given that I have already made myself party to the joint commitment. I have a reason to act in accordance with the claim you have against me, and a reason to ignore competing reasons. The result is that my reason to do my part is protected. It might seem strange to think that equals—i.e., individuals in a nonhierarchical relationship—can have authority over one another.⁴¹ Isn’t a nonhierarchical relationship precisely one in which no one has authority over anyone else? Such a view is tempting but isn’t quite right. A non-hierarchical relationship isn’t a relationship absent authority; rather, it is a relationship in which no one has any greater authority than anyone else. This is consistent with a situation in which everyone has equal and mutual authority over one another. This is the sort of relationship Gilbert envisions in non-hierarchical shared action: we each have the practical authority to demand of the other that she do her part, whatever that might be. The relationship counts as non-hierarchical in that the authority is mutual and roughly equal in strength and scope. Of course, we need a way to determine what exactly counts as “your part” in the shared activity.⁴² However it is we do this, the authority is not maximally specific. That is, absent some further agreement, I needn’t do my part in precisely the way that you specify. If, in baking the cake together, you tell me to melt the butter before mixing, but my mother always told me to use it cold, you needn’t cease to deliberate and do as you’re told, since the authority I have over you does not extend to specifying how you are to mix the butter. The upshot is that on Gilbert’s view, joint commitments by default establish relationships in which individuals party to that arrangement enjoy authority over one another which grounds mutual claims against each other that they do their part in the shared action. Such claims yield protected reasons to conform with the joint commitment they have made. In this way, parties to a joint commitment furnish purposes for one another, as a result of which each can bear authoritybased accountability for what the other does.

3.3.2 Michael Bratman Bratman has developed a competing, reductionist account of shared activity.⁴³ It is reductionist in the sense that he analyzes shared activity in terms of shared

⁴¹ On Gilbert’s view, hierarchically organized shared activity would itself be established by further shared activity (Gilbert, 2008, p. 180). ⁴² For more on this, see the discussion of Bratman in the next section. ⁴³ (Bratman, 1993, p. 113); (Bratman, 1999, pp. 108, 129).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

92

, ,  

intentions, which in turn are analyzed wholly in terms of first-personal intentions held by individuals. On this view, shared intentions are instantiated in an arrangement of individually held intentions across multiple persons. These shared intentions reflect the commitment each individual has to what they are doing together, rather than merely a commitment to her own role in those plans. Inasmuch, each participant intends the entirety of the shared activity. So, if you and I intend to bake a cake together, our respective intentions do not refer to our own respective acts but to the joint act of baking a cake. Put more generally, each participant has an intention of the form I intend that we J, where “J” refers to the joint activity in which the intending agents participate.⁴⁴ Though such an intention refers in its content to an action performed by a collective, the intention is still personal, in that the intention is held by an individual rather than a group.⁴⁵ As participants in shared activity, the ordinary norms of individual planning agency—such as consistency and coherence—apply to our interpersonal intentions. So, for example, suppose again that you and I share an intention to bake a cake. We share this intention only if each of us intends that we bake a cake. But the norms of consistency and coherence require more than this. We need to formulate a complex of interrelated personal intentions that interdepend in the right way. In particular, our plans must mesh. That is, our respective intentions instrumental to or constitutive of our shared intentions—our “subplans” —need not be identical, but they do need to be mutually compossible as a result of intending as much.⁴⁶ The resulting set of interdepending individual attitudes composes a shared intention. Each participating agent must intend that they achieve the activity together, in part because of the other participant’s corresponding intention; and it must be common knowledge between them that their subplans in furtherance of that activity are consistent. As Abraham Roth points out,⁴⁷ the process of meshing subplans is normatively laden in that it is a requirement of rationality. If you and I are baking a cake, your plans normatively constrain mine, and my plans normatively constrain yours. Roth argues that the intentions yielding the plans are subject to those constraints. In this respect, I treat your intentions and plans as the backdrop against which I formulate my own. Your intentions and plans constrain the subsequent

⁴⁴ (Bratman, 1992). ⁴⁵ (Bratman, 1999, pp. 122–123). ⁴⁶ For example, if my subplan is to bake the cake with frosting, and your subplan is to bake a chocolate-flavored cake, and I have no preference about the cake’s flavor and you have no preference about frosting, then our subplans mesh. The result is the following account of shared intention. We intend to J if and only if: 1. (a) I intend that we J and (b) you intend that we J. 2. I intend that we J in accordance with and because of 1a, 1b, and the meshing subplans of 1a and 1b; you intend the same. 3. 1 and 2 are common knowledge. ⁴⁷ (Roth, 2004).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

93

intentions and plans that I might formulate in the same way that my own do. Roth calls this characteristic of shared activity “practical intersubjectivity’. Critics have argued that Bratman’s account neglects the intrinsic normativity implicit in shared activity. Gilbert, in particular, argues that shared intentions are fundamentally obligations. Recall that on her view, if we are engaged in a shared activity—like taking a walk together—and you stop or veer off without further explanation, you are subject to justified rebuke. The justification of this rebuke is grounded in the obligations intrinsic to our shared activity—an obligation your conduct violated. In discussing Gilbert’s view, Bratman rejects her claim that obligations are constitutive of shared action.⁴⁸ He admits that agents engaged in shared activities will often create mutual expectations which in turn might create moral obligations to fulfill those obligations, in accordance with something like Scanlon’s principle of fidelity.⁴⁹ But on Bratman’s view, these obligations are incidental consequences of, rather than essentially intrinsic to, shared action.⁵⁰ As such, he maintains that it is perfectly possible to embark on shared activities absent any such obligations. If something like Bratman’s account of shared actions is correct, then they will not necessarily establish relationships in which each participant has an authoritative claim against the other. This is because shared activity only contingently results in mutual promissory obligations among the participants. If the participants explicitly or implicitly promise to do their part in furtherance of their shared activity, then it is the promise, rather than the shared activity per se, that yields mutual authority among the participants. The fact remains, though, that typical instances of shared action will include implicit agreements among the cooperators. And since these agreements themselves yield protected reasons, the result is that on Bratman’s account, shared action typically yields protected reasons for participants to do their part.⁵¹ Against this, it might seem that shared activity does indeed, in and of itself, yield protected reasons. On Bratman’s view, the norms of practical reasoning, which include consistency and instrumental rationality, yield interpersonal normative constraints in the context of shared action. I intend that my intentions and plans mesh with yours; the constraints of practical reasoning governing my decisions will require that they remain coherent and consistent with yours. This ⁴⁸ (Bratman, 1999). ⁴⁹ See (Scanlon, 1998, p. 304). For discussion of Scanlon’s principle as it relates to shared activity, see (Shiffrin, 2008). For discussion of mutual obligation in the context of shared action, see (Roth, 2004). ⁵⁰ Some, in addition to Gilbert, have criticized Bratman’s account on this point. See for example (Roth, 2004). ⁵¹ There is space for views between Gilbert’s account (in which shared mutual obligations are constitutive of shared activity) and Bratman’s view (in which shared activity only contingently results in mutual obligations). For example, Facundo Alonso (2009) argues that we can accommodate the normative significance of shared intention without having to claim that interpersonal obligations are constitutive of shared intention; instead, on his view, shared intentions necessarily present a basis for interpersonal obligations.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

94

, ,  

will often require that I accommodate some of your decisions as they pertain to our shared activity. For example, suppose that we decide to bake a cake together, though we don’t go so far as to specify exactly who will do what. Instead, we just “play it by ear’. You decide to begin by mixing the batter. Because I decided to coordinate my contributory actions with yours, your decision to mix the batter imposes rational constraints on my subsequent decisions. For instance, I am rationally obligated to refrain from duplicating or stymying your efforts, because I have formulated an intention to coordinate my actions with yours. To the extent that my own intentions are decisively action-guiding for me, it is a requirement of practical rationality that I accommodate your conduct. The normative force of my requirement that I refrain from duplicating or stymying your efforts derives from my second-order intention to coordinate with your conduct. Coordinating with your conduct requires incorporating the decisions you make into my plans. Your decision to mix the batter helps normatively settle what I should do in that I should perform some other step in baking the cake. But your conduct normatively settles what I should do solely in virtue of an intrapersonal division of agential labor, in which I decide at t₀ to coordinate with your conduct at t₁.⁵² This imposes on me protected reasons to act accordingly, which is why your conduct normatively settles what I should do. But because the basis of the protected reason is my own earlier decision, you do not thereby have any authority over me. If you have a basis for complaint should I fail to coordinate with you, it lies in my obligation to fulfill the reasonable expectations I inculcated in you, rather than in having violated a commitment I made to myself. Recall that your decisions settle the matter for me only if you have the requisite authority over me, absent which I have merely pragmatic reasons to defer to you when it comes to deciding what to do. The upshot is that if Bratman’s account is correct, shared action can—and usually does—but need not engender authority-relationships among the participants.⁵³ Thus, if Bratman’s account is correct, participants in shared action can and usually do but need not bear authority-based accountability over one another.

⁵² See section 1.2 (chapter 1) for further discussion. ⁵³ Some have criticized Bratman’s account on precisely that point. On Roth’s view, “. . . one important purpose or role for shared intention is to allow for individuals to have some power or control over what the group does. Shared intention might have a function of distributing the authority or power to settle practical issues facing a group. The alternative to Bratman’s proposal, then, is that the function of shared intention extends beyond interpersonal coordination to include power sharing amongst the participants” (Roth, 2015, p. 47).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

     

95

3.4 Conclusion According to the argument I developed in the previous chapter, a deliberator bears authority-based accountability for what her executor does qua executor only if the deliberator confers a protected reason on the executor, thereby furnishing a purpose for the executor. In this chapter, I presented various ways that agents can confer protected reasons—through promises, requests, agreements, and through shared actions. Though there are a variety of accounts on offer for these various social phenomena, the argument for authority-based accountability applies to all of them—or so I argued. A result is that the kind of authority that a deliberator has in a division of agential labor over an executor is not limited to the kind exhibited in formal and regimented hierarchies characteristic of, for example, the military. Instead, the kind of authority in question is as banal as the authority that we have over each other when we agree to meet for coffee. I also argued that even immoral promises, requests, agreements, and shared actions confer protected reasons—though the reasons in question are practical rather than moral. This captures the sense in which, for example, a “caporegime” in the Italian mafia has authority over a “made man’. That authority yields a protected reason—albeit a protected practical reason—to comply with, say, an order to commit a theft despite that it is morally impermissible to comply. The same goes in general for immoral promises, requests, agreements, and so on. The result is that authority-based accountability applies in such cases. It is worth noting that the account I’ve developed so far can help make sense of accountability in the context of group agency, since on some analyses divisions of agential labor partly constitute group agents. For example, Peter French famously argued that when a group organizes itself in such a way as to include what he calls a “Corporate Internal Decision Structure” (CID structure) that group can act intentionally through that structure. A CID structure is composed of two parts: 1) an organizational flow chart that delineates positions of authority within the corporation, and 2) constitutive rules that determine which decisions count as corporate rather than the personal decisions of the persons who occupy the positions on the organizational flow chart. Such rules, French says, are typically embedded explicitly or implicitly in corporate policy, and serve as an identitycondition for the corporation which explains how it persists as its membership changes.⁵⁴ French argued that such corporations thereby qualify as moral agents. Later writers⁵⁵ develop this line of thought. Though their accounts vary greatly, they all emphasize the role that rules and relations of authority have in grounding

⁵⁴ (French, 1979, pp. 297–317), (French, 1984), (French, 1995). ⁵⁵ See (Copp, 2006), (List & Pettit, 2013), (Hess, 2014)¸ to name just a few. See also (Collins, 2019, pp. 153–179).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

96

, ,  

corporate agency.⁵⁶ On such views, corporate agency derives from a decisionmaking procedure that enables the group to functionally possess and enact representational and motivational states in a way satisfying the constitutive conditions of rationality. The organizational structure embedding this decisionprocedure includes rules by which the constitutive members of the group make decisions and undertake actions. Put in the language I developed, corporate agents are grounded, one way or another, in an interpersonal division of agential labor. The result is that the literature on group agency is fertile ground for applying the argument for authority-based accountability. This fills a lacuna in these accounts of group agency: when a group-agent commits a wrong, authority-based accountability helps us determine which individuals are accountable for which elements of that wrong. This is, though, a task for another time. So far, I’ve characterized divisions of agential labor in a way that includes a host of idealizing assumptions. I’ve assumed that the deliberator and executor regard themselves as such, and that they enact their roles freely, and that they know what they’re doing when they act accordingly. In what follows, I address how the authority-based argument handles cases absent these idealizing assumptions.

⁵⁶ For a helpful recent overview of the role that the discursive dilemma plays in group agency and collective responsibility, see (Szigeti, 2020).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

4 Imperfect Divisions of Agential Labor In the cases I’ve discussed so far, a deliberator furnishes for the executor a morally problematic purpose; the deliberator is accordingly accountable for that wrongmaking feature of the executor’s conduct. I have assumed so far that the participants in the division of agential labor a) are fully committed to what they’re doing together, b) are not coerced into participating, c) know what it is that they are doing together and why they are doing it. One might worry that the argument for authority-based accountability holds water only given these idealizing assumptions. In what follows, I address this worry. The upshot will be that authoritybased accountability survives cases in which the executor—but not necessarily the deliberator—suffers from alienation, coercion, and ignorance. I will begin with whether alienation affects authority-based accountability. Suppose a factory owner agrees to train a worker provided that the worker will, in return, subsequently work on the assembly line at the factory. The factory worker, though, is alienated from the owner’s reasons, in that the worker either a) does not care whether she keeps her promise to work on the assembly line (despite knowing that the promise is binding), or b) does not regard her promise to work on the assembly line as binding in the first place. I call these kinds of alienation “weak” and “strong”, respectively. I will argue in section 4.1.1 that the factory worker’s alienation does not affect the factory owner’s authority-based accountability. Put more generally, the executor’s alienation does not undercut the deliberator’s authority-based accountability. Suppose, though, that the factory owner is alienated from his worker’s reasons, in that he a) does not care whether the factory workers abide by the orders he issues, or b) does not regard the orders he issues to his workers as binding. The deliberator’s “weak” or “strong” alienation can indeed affect his authority-based accountability. This should come as no surprise, since, according to the argument for authority-based accountability, the practical reasons that the deliberator takes there to be determine the purpose of the executor’s conduct. I make this argument in section 4.1.2. What happens when one or more participants in a division of agential labor are coerced into participating? After presenting a preliminary account of coercion and its effect on accountability in general in section 4.2.1, I consider its effects on authority-based accountability specifically. Suppose the factory owner outright threatens the worker into taking the job. Even if she is not morally required to comply, by acceding the worker and the owner thereby establish a division of

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0005

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

98

, ,  

agential labor in which the owner is the deliberator and the worker is the executor. Put simply, the role of executor in a division of agential labor need not be voluntary. The result is that a deliberator bears authority-based accountability even when the executor’s participation is coerced. I make this point in section 4.2.2. I point out, though, that in forcibly establishing a division of agential labor, it is possible to coerce someone into adopting the role of an executor, it is more difficult to coerce someone into adopting the role of deliberator. I end the discussion of coercion by considering in section 4.2.3 examples of coercion in the context of more complex organizational relationships. Ignorance can also mitigate authority-based accountability, under the right conditions. Suppose the factory worker is unaware of what she is helping to build; or suppose the factory owner is uncertain of what exactly his newly hired laborers have agreed to do for him. Alternatively, the owner might not know who the workers are; likewise, the workers might not know who the owner is. How does ignorance of these sorts affect authority-based accountability? I address this issue in section 4.3.1. I argue that the executor’s ignorance of her task’s purpose does not itself do much to affect whether the deliberator bears authority-based accountability for what the executor does. This is in contrast to the deliberator; her non-culpable ignorance of what the executor agreed to do, who the executor is, or what the effects of the executor’s actions are, can indeed undercut the deliberator’s authority-based accountability. I end the discussion of ignorance in section 4.3.2 by considering its blame-mitigating effects in more complex organizational relationships. In section 4.4, I end the chapter by noting that there is a pattern in the analysis of how and whether alienation, coercion, and ignorance affect the division of agential labor and authority-based accountability: the deliberator’s alienation, coercion, and ignorance matter much more than the executor’s alienation, coercion, and ignorance.

4.1 Alienation On the argument for authority-based accountability, we might say that the deliberator’s bad motivating reasons “infect” the executor’s conduct. The deliberator is thereby accountable for that infection. The infection transmits through a division of agential labor, from deliberator to executor. But what if the participants in the division of agential labor are, psychologically speaking, less than fully “on board” with what they’re doing together? What if the executor or the deliberator does not take her role seriously? Does the infection still transmit from deliberator to executor in such cases? This is the issue I address in this section on alienation. I will argue that the deliberator’s alienation can affect authority-based accountability whereas the executor’s alienation does not. In doing so, I will focus on cases

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

99

in which the division of agential labor is established by way of a promise—i.e., P1 accepts a promise that P2 makes. Though promising is just one way to establish a division of agential labor,¹ I limit the examples of alienation to cases of promising for the sake of expository convenience. The conclusions of the discussion are applicable to cases of alienated agreements, requests, and joint commitments as well.

4.1.1 Alienated Executors Here I consider two ways in which the promisor—the executor—is alienated from the promise she makes to a promisee—the deliberator. The upshot of this section will be that even when an executor is, in the various ways to be discussed, alienated qua executor, the deliberator can still bear authority-based accountability for what the executor does. Sometimes we make promises but don’t actually care whether we fulfill them. This isn’t to say that we have no intention to fulfill them. Rather, it means we don’t take the promise as seriously as we should. If I make a promise to do ϕ, I promise to do ϕ in a range of situations. For example, if I promise to take you to the airport on Monday, I promise to do so even if I have a mild headache, or if it’s raining, or if I just don’t feel like it, and so on. In this respect, making a promise means adopting the right sort of disposition. Of course, the range of situations in which I’m committed to fulfilling the promise might be vaguely defined. Still, many situations will be clear-cut. That is, there are hypothetical situations in which it is pellucidly clear that the promise I made applies. Suppose, though, that in making the promise, I do not adopt the disposition to fulfill it in all or even most of the situations in which it is supposed to apply. Suppose I promise to drive you to the airport on Monday; I intend to fulfill this promise unless it means missing overtime in the football game playing that night. Yet the promise I make is not explicitly or implicitly conditional in that way. That is to say, in no way do I indicate to you (or is it otherwise implied) the following: I’ll drive you to the airport only if it means that I won’t miss the football game. Suppose the football game ends without going into overtime. I consequently keep the promise I made. You remain none the wiser that if things had gone slightly differently, I would have violated the promise I made. Was the promise I made in this case sincere? In making it, I knew there was a significant chance that I would break it. But I had no intention, at that point, to break it—only a prediction that I might very well do so. It is not wholly infelicitous to call the promise sincere. After all, I intended to keep the promise. Against this, it

¹ See chapter 3 for discussion of other methods.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

100

, ,  

might be said that I cannot felicitously intend to do what I suspect I ultimately will choose not to do. Suppose, though, I will violate the promise should keeping it prove at all costly for me—and I fail to predict this. I either mistakenly think that I will keep the promise should the going get tough, or I otherwise fail to consider whether I will do so. Again, such a promise is made sincerely. But where I fail to adopt the right sort of disposition (knowingly or not), the promise I made is defective in an important sense. After all, the promise imposes upon me a putative obligation to adopt the right sort of dispositions. It is only due to circumstantial moral luck that I happen to keep the promise. What we want from a promise-maker is a certain kind of modal robustness in their conduct, which my promise in these cases lacks. In this respect, the promise is what I will call “dispositionally defective”. Dispositionally defective promise-making characterizes what I call “weak alienation”. The executor is weakly alienated from a promise she makes if i) she promises to perform some task ϕ for the deliberator, ii) the conditions under which she intends to do ϕ are not the same as the conditions under which she ostensibly has a promissory obligation to do ϕ, and iii) the executor subsequently does ϕ, thereby fulfilling the terms of her promise. Take the following example: Hired Emails Deliberator wants threatening emails sent to an outspoken political figure in the hope that it will dissuade her from speaking at a public event. Because Deliberator lacks the skills to send the threatening emails anonymously, he offers to pay Executor to do so, but only if the emails successfully dissuade the political figure. Executor agrees to do so. Though she believes that dissuading the political figure is a worthy cause, and though she regards the promise she made as binding, she will violate it if sending the emails proves at all inconvenient for her. As it so happens, it doesn’t; she consequently sends the emails, successfully dissuading the political figure. Recall that when a deliberator confers and an executor accepts a protected reason, the protected status of that reason is what licenses attributing to the executor the purpose of enacting the practical reasons that the deliberator takes there to be. As a result, the deliberator bears authority-based accountability for what the executor does within the scope of the protected reason in question. Now, Executor’s failure to treat her protected reasons as a protected reason might seem to undermine her role as an executor in a division of moral labor. And this, in turn, would undermine Deliberator’s authority-based accountability for Executor’s actions. But this is not so. She retains that protected reason even if she fails to treat it as a protected reason. What grounds the protected status of the reason Executor has—and with it, her function of enacting Deliberator’s aims—is not how she treats the reason in question, but instead whether she made a promise that Executor accepted.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

101

The lesson here is that the argument for authority-based accountability can accommodate the executor’s weak alienation. But a still more radical kind of alienation is possible. According to what I will call “strong alienation”, Executor promises to do ϕ for P1, and subsequently does so, but denies that she has a promissory obligation to act accordingly. Consider the following case. Nihilist Emails Deliberator wants threatening emails sent to an outspoken political figure in the hope that it will dissuade her from speaking at a public event. Because Deliberator lacks the skills to send the threatening emails anonymously, he offers to pay Executor to do so, but only if the emails successfully dissuade the political figure. Executor consequently promises to do as much. But Executor does not think she has any promissory obligation to act on her promise. This is because she does not believe that promises obligate. But she fulfills her promise anyway because she wants to be paid. Although Executor fully intends on acting in the way that she happened to have promised, she denies that she has any protected reason to abide by the promise as such. More generally, she denies that the promises she makes provide even nonmoral, practical reasons for compliance. If there is any sense in which there’s a division of agential labor between Deliberator and Executor, she denies its normative force. She consequently takes herself to have only a prudential reason to fulfill the terms of the promise she made; she does so solely to get paid. It might seem that Deliberator’s motivating reasons cannot serve as a basis for evaluating Executor’s conduct since in such a case Executor does not construe herself as having the function of enacting Deliberator’s aims. But promises are binding even if the promiser makes the promise insincerely. And so it is with Executor, whose promise yields a protected practical reason despite her belief that it doesn’t. By ostensibly making that promise—a promise that Deliberator accepts—Executor establishes a division of agential labor with Deliberator even if Executor is disposed to deny that Deliberator has any authority over her. Of course, Executor ought not to comply with Deliberator’s demands since the authority she has granted Deliberator, and the protected reason that the authority yields, serves as a practical reason over which reasons of morality lexically prevail. But, if she does comply, Deliberator thereby furnishes part of the purpose for which Executor acts, in virtue of their respective roles in the division of agential labor they’ve established. Executor might disingenuously deny that they’ve established that relationship. But the ostensible promise she made has the function and the effect of conferring the relevant practical authority on Deliberator. The fact, then, that the promise Executor made was insincere does not vitiate Deliberator’s authority-based accountability for what Executor does.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

102

, ,  

4.1.2 Alienated Deliberators So far, I’ve considered only cases in which the promise-maker—i.e., Executor—is alienated from the promise she makes. I argued that these types of alienation do not mitigate Deliberator’s authority-based accountability for what Executor does. In what follows I consider cases in which the promise-accepter—i.e., Deliberator— is alienated from the promise Executor makes. I will argue that Deliberator’s alienation can indeed mitigate her authority-based accountability for Executor’s conduct. Start with the following case: Protégé Emails Executor wants threatening emails sent to an outspoken political figure. In addition, Executor hopes to impress Deliberator with her criminal behavior, and so promises to Deliberator that she will send the threatening emails. Deliberator accepts Executor’s promise, but just to placate Executor. Deliberator doesn’t actually care whether the emails are sent or whether the political figure is dissuaded. Here, Deliberator doesn’t care whether Executor fulfills the promise she made. Deliberator is weakly alienated from the promise she accepts. We can characterize a deliberator’s weak alienation in the following general way. Deliberator is weakly alienated from the promise Executor makes if i) Executor promises to Deliberator to perform some task ϕ, ii) Deliberator takes there to be no good reasons for Executor to do ϕ, yet iii) Deliberator accepts the promise Executor made, which Executor subsequently fulfills. In Protégé Emails, Deliberator has no reasons for Executor to send the emails. But he does indeed have a motivating reason to accept the promise Executor made: to placate Executor. So, the purpose Deliberator furnishes for Executor is not to dissuade the political figure, but rather to be placated. Such a purpose is still wrongful given its side-effects, but it is less wrongful. The upshot for authoritybased accountability is that a deliberator’s weak alienation, unlike an executor’s weak alienation, can mitigate the deliberator’s authority-based accountability for what the executor does. What happens if Deliberator doesn’t sincerely accept Executor’s promise? It is more or less clear what it means to make a promise insincerely; but what does it mean to accept a promise insincerely? Recall that on Normative Powers accounts of promising (see section 3.1.1, chapter 3), when Deliberator accepts Executor’s promise to do ϕ, Executor thereby grants Deliberator the practical authority to demand that Executor do ϕ. But if Deliberator accepts the promise insincerely, then at least within the confines of her own conscience, she refuses to

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

103

acknowledge or recognize the authority vested in her. In particular, Deliberator denies that she has a practical claim against Executor that Executor do ϕ. As a result, if Executor fails to do ϕ, then Deliberator won’t regard her as having done anything wrong. The point here is not that Deliberator will forgive Executor by letting her off the hook—rather, as far as Deliberator is concerned, Executor did nothing wrong for which to be forgiven. To be clear, Deliberator has not released Executor from the promise she made; instead, within the confines of her conscience, Deliberator denies that the promise Executor made is binding in the first place, despite having ostensibly accepted the promise Executor made. At first it might seem unclear why Deliberator would privately deny that the promise she accepted is binding. But people often privately disavow the authority they ostensibly have over wrongdoers. Consider this case: Benefactor Executor, an unscrupulous individual, has assumed the role of Deliberator’s benefactor contrary to Deliberator’s wishes. As an act of revenge, Executor promises to send nasty emails to a neighbor who slighted Deliberator. Deliberator doesn’t take there to be any practical reasons in favor of doing so. But to avoid a prolonged argument, Deliberator ostensibly accepts Executor’s promise. He accepts the promise insincerely in the following sense: within the confines of his own thoughts, Deliberator denies that he has a practical claim against Executor that she send the nasty emails to the neighbor. And if Executor reneges on her promise to do so, Deliberator will not regard Executor as having wronged him. Here, Deliberator is strongly alienated from the promise he accepted from Executor. Deliberator is strongly alienated if Executor promises to do ϕ for Deliberator, Deliberator ostensibly accepts that promise which Executor subsequently fulfills, but Deliberator denies that Executor has a promissory obligation to act accordingly. But accepting Executor’s promise—sincerely or not—yields a practical claim against Executor that she send the nasty emails. So, Deliberator has that non-moral practical claim against Executor regardless of whether he recognizes it or chooses to exercise it. Of course, Deliberator could release Executor from the claim that he has against her; but in Benefactor he refuses to do so. Inasmuch, ostensibly accepting the promise yields for Executor a non-moral protected practical reason to send the nasty emails. Executor and Deliberator have thereby established a division of agential labor. Of course, Deliberator might deny as much. But doing so would be disingenuous. Whether a promise is accepted sincerely does not affect the authority the promise confers—for the same reason that whether a promise is made sincerely does not affect the authority the promise confers.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

104

, ,  

So, the fact that Deliberator denies that he has authority over Executor does not vitiate his status as a deliberator in a division of agential labor. By ostensibly accepting Executor’s promise to send the emails, Deliberator furnishes that purpose, since it is included in the content of the promise he accepts. But his motivating reason for furnishing the purpose—that is, the object of the purpose he furnishes—isn’t to seek revenge on his neighbor. Rather, it is to avoid a prolonged argument with Executor. Though such a purpose is wrongful given its side-effect, it is less wrongful. The upshot is that the deliberator’s weak and strong alienation can each mitigate the deliberator’s authority-based accountability. So far, I’ve focused the discussion of alienation on examples involving two individuals. It is helpful to expand the examples to include more complex cooperatively committed harms. Consider the following case—a variant of those discussed so far in this section. Concerted Harassment 1 Two politically motivated individuals, P1 and P2, want a barrage of harassing emails sent to an outspoken political figure in the hope that it will dissuade her from speaking at a public event. They alone cannot send enough emails though, so they pay for the assistance of eight other individuals. The ten of them promise one another to send the harassing emails. P1 and P2 will pay the others regardless of whether the concerted attempt is successful. Among the eight individuals recruited, one of them, Dissuade, wants to help achieve the aim of dissuading the speaker. Another individual, Embolden, hopes that sending the harassing emails will embolden the speaker. The resulting barrage of emails ultimately dissuades the speaker. By virtue of the promises they make to one another, they establish a division of agential labor in which each counts as both a deliberator and an executor; they severally confer upon each other, and concomitantly accept, a protected practical reason to send the emails. The result is that each of them bears authority-based accountability for what they together do. However, Dissuade and Embolden have different private goals. The content of the promise they make to one another is one and the same: to send the harassing emails to the political figure. But they harbor radically different motivating reasons in favor of acting accordingly. For Dissuade, the point is to dissuade the political figure, whereas for Embolden the point is to embolden the political figure. The result is that each furnishes for the other a radically different purpose. Recall that according to the argument for authority-based accountability, the practical reasons a deliberator takes there to be can partly determine the purpose of what the executor does and can thereby count as a wrong-making feature of what the executor does. Dissuade accepts a promise from Embolden that has the

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

105

purpose of emboldening the political figure; thus, Dissuade has that purpose qua executor. Embolden accepts a promise from Dissuade that has the purpose of dissuading the political figure; thus, Embolden has that purpose qua executor. Each is accountable for the purpose that she confers upon the other. Since the purpose Embolden confers is presumably less bad than the purpose Dissuade confers, Embolden bears less authority-based accountability than Dissuade, all else being equal. The general lesson here is that in the context of cooperatively committed harms, a deliberator’s weak or strong alienation from the promise she accepts can diminish her authority-based accountability. But an executor’s alienation from the promise she makes does not mitigate the deliberator’s authority-based accountability. What happens, though, when the deliberator or executor is forced to adopt their respective roles in a division of agential labor? Does the argument for authority-based accountability still apply? I turn to such cases next.

4.2 Coercion As a participant, you might be coerced by others into participating in a cooperative activity. In such cases, your participation in the division of agential labor is less than fully voluntary.² Coercion can mitigate—and in certain cases, eliminate— accountability for wrongdoing. Here I will discuss whether and when it mitigates or eliminates authority-based accountability. In particular, I will argue that coercively forcing an agent into making a promise by subjecting her to a coercive threat does not vitiate her role as an executor in a division of labor. Suppose, for example, a mafioso credibly claims that he will kill a juror unless she promises to find a defendant “not guilty” in a criminal trial. The juror promises to do as she is told. The mafioso thereby counts as a deliberator and the juror counts as an executor even though the promise was not made freely and thereby fails qua promise. The result is that the mafioso bears authority-based accountability for what the juror does. However, I will argue that coercively forcing an agent into accepting a promise by subjecting her to a coercive threat fails to establish a cogent division of labor. Suppose a mafioso wants a grocer to be in his debt. So, he promises the grocer that he will “eliminate” the grocer’s local competition. The grocer refuses to accept the promise. But the mafioso makes it clear that failure to accept the promise would ² This conception of coercion stands in contrast to Harry Frankfurt’s account according to which coercion “requires that the victim of a threat should have no alternative to submission, in a sense in which this implies not merely that the person would act reasonably in submitting and therefore is not to be blamed for submitting, but rather that he is not morally responsible for his submissive action” (Frankfurt, 1973, pp. 38–39). I do not limit the concept to cases of this sort in which the agent is incapable of resisting a threat.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

106

, ,  

result in the grocer’s death. So, the grocer at least ostensibly accepts the promise that the mafioso makes. Here, the grocer does not count as a deliberator in a division of agential labor. Neither does the mafioso count as an executor. The result is that the grocer does not bear authority-based accountability for what the mafioso does. Before I begin, though, I will make some preliminary remarks about coercion in general. Coercion occurs just in case i) the coercing agent commits an act which imposes a cost C on an option O1 to which the coerced agent has a right, with the aim of motivating the coerced agent to choose an alternative option O2, ii) the coerced agent has a right to option O1 at some lower cost, and iii) the coerced agent chooses option O2 in order to avoid C. So, suppose a mugger points a gun at me and says, “your money or your life”; the mugger is attaching a cost—that of death—to an option I have—the option of keeping my money. Hence, C is death by murder, O1 is the option of keeping my money, and O2 is the option of handing over my money. I have a right to keep my money absent the cost of being killed; so, if I choose to surrender my money in order to avoid that cost, I count as coerced. This is obviously a “moralized” account of coercion insofar as it claims that coercion entails a rights-infringement.³ This account is capacious enough to include both coercive threats and coercive offers. Suppose I am drowning in the ocean through no fault of my own, and a fisherman on his boat has a chance to haul me on board. Assume the fisherman has a modest duty of rescue in that he is required to rescue me, though in doing so he can shift to me the cost that the rescue imposes on him—perhaps the value of the day’s catch of which he is deprived by rescuing me. But instead, he offers to haul me on board only if I surrender my life’s savings to him. In doing so, he goes overboard (as it were) in that he demands far more than he is entitled by way of compensation. In this case, C is death by refusing to rescue, O1 is the option of keeping my life’s savings, and O2 is the option of handing over my life’s savings. I have a right to keep my life’s savings; if I choose to surrender it in order to avoid the cost of death, then I count as coerced, even though he threatens me with merely doing nothing. This account of coercion is obviously incomplete. Since it is a moralized account, we need to know what costs can be attached to what options without infringing my rights. Determining as much would require a comprehensive account of rights, which is obviously beyond the ambit of this book. The buckpassing account of coercion I’ve outlined here will suffice for my purposes, which is to determine how and when in general coercion undercuts authority-based

³ The locus classicus for moralized accounts of coercion is (Wertheimer, 1987). “Hard choices which do not arise from injustice,” he says, “are not coercive in any important moral sense” (p. 234). See also (Wellman, 2005, pp. 132–138).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

107

accountability. After briefly discussing how coercion mitigates accountability, I turn to cases of coercion in divisions of agential labor.

4.2.1 Coercion and Accountability in General Sometimes we are not excused but rather justified in acceding to a coercive threat. Suppose a villain credibly threatens to kill me unless I pinch an innocent bystander. In this case, pinching the bystander is clearly justified as the lesser evil. I make the right choice in deciding to accede to the coercive threat. Of course, this is not to say that the villain acts justifiably in threatening me—far from it. Neither is this to say that the bystander’s rights weren’t infringed. She is owed compensation for the pro tanto wrong she suffered. But the right that the bystander had not to be pinched is significantly less weighty than the right I had not to be killed, which is what justified my actions. I will focus, though, on cases in which it is all-things-considered wrongful for you to accede to the coercive threat—yet doing so is partially or fully excused precisely because you are coerced into doing so. What is the basis for the excuse? Accountability-mitigating acts of coercion make it more difficult, psychologically, to do the right thing—i.e., to resist the coercive threat.⁴ The more psychologically difficult it is for an agent to do the right thing, the less accountable she is for failing to do so—provided that she isn’t accountable for the difficulty she faces. So, for example, akratic individuals might find it more difficult to resist coercive threats. Provided such individuals are accountable for their own akrasia, the greater difficulty they face in resisting the threat does not yield for them a greater excuse. Or suppose a virulent racist is faced with the following threat: if he doesn’t kill a random passerby, he will be forced to eat at a restaurant with members of the race he despises. No matter how psychologically difficult it is for him to do the right thing, he has no excuse—even a partial one—should he accede to the threat, insofar as he is presumptively accountable for the psychological difficulty he faces. The psychological account can, then, accommodate these cases. The claim that difficulty mitigates accountability might seem to yield the implausible conclusion that sufficiently tempting offers fail to partially excuse wrongful conduct. After all, such offers can exert psychological pressures of the same severity as coercive threats. There are a variety of temptations in life—love, power, fame—which make it quite difficult to do what morality requires of us. Yet no matter how tempting these goals might be, they do not yield even a partial excuse for wrongs done in pursuing them.

⁴ See (Feinberg, 1986, pp. 189–268) and (Murray & Dudrick, 1995) for a related account.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

108

, ,  

In response to this challenge, I maintain that an agent’s duty to forego benefits to herself is more stringent than her duty to accept harms to herself when in both cases the alternative is a pro tanto wrong. This point can be put differently. We all have an agent-centered prerogative to give extra weight to our own legitimate interests, which is outweighed or overridden when the harmful consequences to others are sufficiently dire. This threshold differs, from case to case, depending on whether the interests I am securing for myself consist in avoiding a harm or obtaining a benefit. More specifically, the threshold at which I am required to choose the interests of others over my own interests is harder to satisfy when the interests accruing to me consist in obtaining a benefit as opposed to avoiding a harm. This helps explain why the sheer difficulty of resisting coercive threats might serve as an excuse whereas the difficulty of resisting a tempting offer does not—even if the difficulty is of the same magnitude in both cases. In general, the difficulty required to excuse wrongfully acceding to a coercive threat is substantially greater than the difficulty required to excuse wrongfully accepting a tempting offer. With a preliminary account of how coercion diminishes accountability for wrongdoing, I now turn to the effect that coercion has on authority-based accountability. In doing so, I will focus on coercion in the context of promises and agreements. (For the sake of brevity, I will focus on coercive threats rather than coercive offers.)

4.2.2 Coerced Promises It is generally thought that promises are morally binding only if they are entered into freely.⁵ If a coerced promise is not binding, then such a promise does not generate additional moral reasons in its favor, above and beyond whatever reasons already exist for the coerced promiser to fulfill the promise she ostensibly made. (So, for example, the fact that a promise not to commit murder is coerced doesn’t make it permissible to commit murder.) But why believe that coerced promises aren’t morally binding? Perhaps it is conceptually impossible to coerce a promise. On this view, a coerced promise is no promise at all.⁶ Recall from section 3.1.1 (chapter 3) that the practice of promising is empowering because it provides a way for us to exercise a specific normative power: that of voluntarily putting ourselves under a moral obligation to others. But a moral power is transformative in this way only if I exercise it voluntarily. If I am coerced into making a promise, then ⁵ But see (Gilbert, 1993). ⁶ Many have traditionally argued for this view. See e.g., (Simmons, 1979, p. 82), (Simmons, 1984, pp. 812–816), (Raz, 1981, p. 126).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

109

my decision to put myself under a moral obligation to someone else is no longer voluntary. Rather, that moral obligation is imposed upon me. Since the definitive function of promises is to exercise the normative power to voluntarily put ourselves under moral obligations to others, coerced promises do not count as promises properly speaking, since these ersatz promises fail to fulfill that definitive function. There is an analogy here with consent. Clearly, coerced consent is invalid. This is because the practice of consent, as a normative power, allows us voluntarily to transform the impermissible into the permissible by granting permissions to others. It is the definitive function of consent to grant such permissions voluntarily. This is why coerced consent is invalid; mutatis mutandis for coerced promises.⁷ But I will argue that a coerced promise can still yield protected practical reasons, for the same reason that demands can do so. I will begin by discussing coercively made promises, after which I turn to coercively accepted promises. In both cases I explore whether and how such coercion affects authority-based accountability.

Coercively Made Promises Start with the following case: Coerced Promise 1 Villain is envious of Victim. Consequently, Villain doesn’t want Victim to hang out with a mutual friend of theirs. But Villain knows that Victim is perfectly entitled to hang out with their mutual friend. So, Villain impermissibly levels the following credible threat against Victim: unless Victim promises to stop seeing their mutual friend, Villain will spread false rumors about Victim (which Villain will also do if Victim reveals this threat to their mutual friend). Victim accedes to this threat by promising not to see their friend anymore. Suppose the promise Villain compels from Victim is in no sense binding. We can, however, recharacterize what Victim is doing. Though her threat fails to elicit a binding promise, it succeeds as a demand—a demand that Victim cease meeting their mutual friend. Demands can establish a division of agential labor, much in the same way that requests can. In section 3.2.1 (chapter 3), I suggested that there is a sense in which we all have a standing authority over one another whereby we can confer protected reasons simply by making a request. The authority is practical, not moral, in that there is nothing morally wrong with refusing such a request. So, when I make a

⁷ But see (Gilbert, 1993).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

110

, ,  

request of you, I thereby provide you with a protected reason to comply—albeit one that you can refuse without acting immorally or irrationally. Should you choose to comply with my request, you thereby undertake the function of enacting my practical reasons which have the concomitant function of normatively guiding your conduct. That is, in virtue of soliciting your assistance and in virtue of complying with my request, we establish a division of agential labor in which I count as the deliberator and you count as the executor. I bear authority-based accountability, qua deliberator, for what you do qua doer. In Coerced Promise 1, Villain doesn’t merely request that Victim do ϕ. Rather, Villain, in effect, demands it. But a demand can function like a request. When you make a demand of another person that she do ϕ, you insist that she treat that very demand as a protected reason to comply. In this respect, making a demand is like making a request. The difference is that a demand suggests either that you believe yourself to have a claim over the person that she do as you say, or that you will refuse to take “no” for an answer. Either way, demands yield a protected reason to comply. At first, this might not seem true of Victim. It might seem that what “settles the matter” for him regarding ϕ is not the putative authority Villain has to make the demand, but rather the threat that Villain credibly makes. It’s true (we can stipulate) that the desire not to suffer harm is what motivates Victim. But the threat is precisely why the demand settles the matter for Victim. In effect, Villain coerces Victim into establishing a division of agential labor with him, in which Villain counts as the deliberator and Victim counts as the executor. None of this means, of course, that Victim is morally required to accede. Nonetheless, if Victim chooses to comply, then he does so by ceding to Villain the authority to determine what she should do. The upshot, then, is that even if Villain coerces Victim into making a promise, and even if, as a result, that promise isn’t binding—or doesn’t count as a promise at all—Victim’s decision to accede to Villain’s demand establishes a division of agential labor between them. That is, as far as the two of them are concerned, Villain’s demand settles the matter for Victim. The result is that Villain bears authority-based accountability for what Victim does. The same goes for cases in which Victim is coerced by a third party into promising to do ϕ for Villain: Coerced Promise 2 Villain holds a gun to Victim’s head and credibly threatens to shoot Victim unless he makes a promise to Beneficiary: that he will commit a theft and then give the proceeds to Beneficiary. If Victim accedes to Villain’s demand, Villain will follow Victim around to ensure that he fulfills the promise Victim ostensibly makes to Beneficiary. Victim chooses to accede to Villain’s demand. Beneficiary, who is ignorant of the fact that Victim is coerced, accepts the promise Victim ostensibly makes.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

111

Suppose, as the result of the coercive threat, the promise Victim ostensibly makes is not binding—or does not count as a promise at all. Again, this does not matter as far as Villain’s authority-based accountability is concerned. This is because Villain is making a demand of Victim—a demand with which Victim complies. When we comply with a demand, the demand confers a protected reason. The content of Villain’s demand is, in effect, to establish a division of agential labor with Beneficiary—which, in turn, yields a protected reason commit a theft (albeit a non-moral, practical one). None of this means, of course, that Victim is morally required or even permitted to accede. Nonetheless, if Victim chooses to comply with Villain’s demands, then he does so by ceding to Villain the practical authority to determine whether he should cede to Beneficiary the practical authority to determine what Victim should do regarding the theft. Put differently, acceding to Villain’s demand results in a nested protected reason: a protected reason in favor of a protected reason. As far as the argument for authority-based accountability is concerned, Beneficiary bears authority-based accountability for what Victim does, since Beneficiary, by accepting Victim’s ostensibly made promise, establishes a division of agential labor in which Beneficiary counts as the deliberator and Victim counts as the executor, which yields for Victim a protected reason to commit the theft. And Victim, in virtue of acceding to Villain’s demand, counts as an executor. Recall from section 4.1.1 that the executor’s alienation from the promise he makes doesn’t vitiate the deliberator’s authority-based accountability. Thus, Beneficiary bears authority-based accountability for what Victim does. Notice though, that Villain ends up bearing authority-based accountability for the theft as well. Victim accedes to the demand Villain makes; that demand yields a protected reason in its favor. (See section 5.2.3, chapter 5, for more on this sort of case.) Suppose, alternatively, no one is trying to get Victim to make a promise to commit the theft. Rather, Victim’s circumstances impose a significant cost on refraining from making such a promise. Consider the following case: Coerced Promise 3 Villain offers to pay Victim a modest fee should Victim commit a theft in return. Unbeknownst to Villain, Victim is suffering from a rare disease, treatment for which requires an expensive life-saving surgery. The only way for Victim to obtain the necessary funds is by committing the theft for Villain, which Victim consequently promises to do. Here Victim is “coerced” by the disease from which she suffers; in effect, it imposes upon her a very high cost for refusing Villain’s offer. It is in response to this cost that Victim decides to establish a division of agential labor with Victim, in which Villain counts as the deliberator and Victim counts as the

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

112

, ,  

executor. The result, again, is that Villain bears authority-based accountability when Victim commits the theft. In canvassing how coercion affects authority-based accountability, I’ve focused on cases in which the promise is coercively made. I’ve argued that this doesn’t vitiate the deliberator’s authority-based accountability. I now turn to cases in which the executor is coerced in accepting a promise.

Coercively Accepted Promises The issue again is whether we can coherently force others, through a coercive threat, to adopt a role in a division of agential labor. I argued in the last section that if we force a victim, through a coercive threat, to defer to us when making a decision, we thereby force the victim into adopting the role of executor in a division of agential labor. It is questionable, though, whether it is possible to force a victim, through a coercive threat, into adopting the role of deliberator in a division of agential labor. To see why this is questionable, take the following case: Coerced Acceptance 1 Villain credibly threatens to kill Victim, unless Victim accepts Villain’s promise to do ϕ, where ϕ is a wrongful act. This case is hard to process. For me to accept a promise from you is for me to accept normative authority over you. This, in turn, means that I enjoy the discretion to decide whether you have a protected reason to do what you’ve promised to do, by deciding whether to hold you to that promise. The discretionary character of authority is what partly explains why, in Coerced Acceptance 1, Villain’s attempt to coerce Victim strikes us as bizarre. Suppose Victim accedes to the threat by accepting Villain’s promise to do ϕ; Victim can immediately thereafter exercise his discretion by unilaterally releasing Villain from that promise. It’s hard to know, then, what Villain could possibly gain by coercing Victim into accepting a promise. We might, as a result, elaborate on the example in the following way. Coerced Acceptance 2 Villain credibly threatens to kill Victim unless Victim a) accepts Villain’s promise to commit ϕ (where ϕ is a wrongful act) and b) holds Villain to that promise. In this example, Victim is coerced not just into accepting a promise from Villain, but into refraining from exercising her discretion to release Villain from that promise. That is, Villain demands of Victim that he accept a promise from Victim and hold Victim to that promise. In effect, Villain demands that Victim accept the role of “deliberator” in a division of agential labor in which Villain counts as the “executor”—and then Villain demands that Victim decide in a particular way.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

113

There is a straightforward sense, then, in which Victim is a deliberator in name only. She has no independent desire to accept Villain’s promise or to hold him to it. And she certainly has no desire for Villain to do ϕ. Rather, Victim accepts Villain’s promise and holds him to it, only on pain of death. When Victim ostensibly holds Villain to the promise she made, what “settles the matter” for Villain is not a decision Victim made, but rather a decision that Villain himself made. There is a straightforward sense in which Villain, rather than Victim, is “calling the shots” when it comes to what Victim should do. There is, then, an asymmetry in verdicts as to whether, in forcibly establishing a division of agential labor, it is possible to coerce someone into adopting the role of an executor as opposed to a deliberator. And this difference in verdicts makes intuitive sense: you can compel someone to defer to others, but you cannot compel someone to decide for themselves while simultaneously telling them what to decide. The upshot is that, in Coerced Acceptance 2, Victim does not bear authoritybased accountability for Villain’s actions when Villain does ϕ. This is because Victim has not established a division of agential labor with Villain in which Victim counts as a deliberator. Things are more complicated, though, in cases where the coerced promise is made to a third party: Coerced Acceptance 3 Villain credibly threatens to kill Victim unless she a) accepts Crook’s promise to commit ϕ (where ϕ is a wrongful act) and b) holds Crook to that promise. Recall that coercive demands yield protected reasons to do what is demanded. Villain coercively demands of Victim that she accept a promise from Crook and hold Crook to that promise. As I’ve argued, Victim does not qualify as a deliberator in this case. But, presumably, the purpose of Villain’s threat is to coerce Victim into motivating Crook to do ϕ. We can, then, recharacterize the coercive threat that Villain makes in the following way: Villain coercively demands of Victim that she motivate Villain to do ϕ. Victim accedes to that demand, thereby accepting the role of executor in a division of labor in which Villain counts as the deliberator. The result is that Villain is inculpated in what Victim does when she acts qua executor—i.e., when she motivates Crook to do ϕ. The upshot, then, is that according to the argument for authority-based accountability, Villain (and of course Crook) is accountable for ϕ, whereas Victim is not. Suppose, though, that Villain’s coercive threat is less severe. Perhaps he threatens to break Victim’s arm rather than to kill her. Recall from the discussion of Coerced Acceptance 2 that, because Victim was coerced, there is no division of agential labor between him and Villain; Victim cannot relevantly count as a “deliberator” when acting under threat of death. But as the severity of the threat decreases, it becomes less difficult for Villain to resist the threat. On the view of

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

114

, ,  

coercion discussion in section 4.2.1, its exculpatory effects are scalar; the worse the coercive threat, the greater the excuse, all things being equal. The result is that Villain’s status as a deliberator—and with it, the authority-based accountability she bears—co-varies with the severity of the coercive threat she faces.

4.2.3 Coercion in Cooperatively Committed Harm In discussing how the argument for authority-based accountability handles coercion, I’ve focused on dyadic cases of shared action. The upshot of this discussion is that the only circumstances in which coercion undercuts authority-based accountability are those in which the individual accepting the promise—i.e., the putative “deliberator”—is coerced into doing so. With this upshot in mind, let’s briefly consider the role that coercion can play in more complex cooperatively committed harms. Bigots 1 Suppose Bigot-A and Bigot-B agree that on Sunday night they’ll spray-paint antiimmigration epithets on a local immigrant-owned business. Bigot-B and Bigot-C make the same agreement with each other. So, Bigot-B is party to two separate agreements, whereas Bigot-A and Bigot-B are each party to only one agreement. Given the argument for authority-based accountability, Bigot-A bears authoritybased accountability for what Bigot-B does, and Bigot-C bears authority-based accountability for what Bigot-B does, while Bigot-B bears authority-based accountability for what Bigot-A and Bigot-B do. Suppose, though, that Bigot-C is notoriously violent and reckless (even for a bigot). As a result, Bigot-B wants nothing to do with Bigot-C. But Bigot-C coercively threatens Bigot-B into accepting his promise to engage in the act of vandalism on Sunday night. Provided that the threat is severe enough, Bigot-B will not bear authority-based accountability for what Bigot-C does—though he will still bear authority-based accountability for what Bigot-A does. The lesson, here, then, is that a coerced participant in a cooperatively committed harm can still be accountable for what other participants do. This is because the threat might be coercive only with respect to certain aspects of the cooperatively committed harm. In effect, BigotB is still a deliberator when it comes to Bigot-A, but not when it comes to Bigot-C. Sometimes, though, the consequences of selective coercion can reverberate through a cooperatively committed harm. Consider this case: Bigots 2 Burglar is a bigot who wants to break into and loot a high-end sporting goods store owned by a local immigrant. She lacks the expertise to deactivate the store’s

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

115

alarm system. So, she enlists Alarm Specialist’s help, who promises to deactivate the alarm. However, there are a host of surveillance cameras that Alarm Specialist cannot reach. So, he enlists Surveillance Specialist who agrees to deploy infrared emitters that will disable the security cameras. Burglar wants ϕ done but can’t do it herself since it requires accomplishing a number of smaller tasks. So, Burglar enlists Alarm Specialist to accomplish one of those smaller tasks, who in turn enlists Surveillance Specialist. This relationship describes a chain of command in which Burglar has authority over Alarm Specialist, who in turn has authority over Surveillance Specialist. The result is that Burglar bears authority-based accountability for what Surveillance Specialist does, even though Burglar has no agreement with Surveillance Specialist. This is because Burglar bears authority-based accountability for what Alarm Specialist foreseeably does in furtherance of disabling the alarm—and one thing Alarm Specialist foreseeably does in furtherance of that is to enlist Surveillance Specialist’s assistance in disabling the surveillance cameras. That is, Burglar bears authority-based accountability for what Alarm Specialist does in foreseeable furtherance of the promise she makes. The result is that Burglar bears authoritybased accountability for what Alarm Specialist does and what Surveillance Specialist does. And Alarm Specialist bears authority-based accountability for what Surveillance Specialist does. On the other hand, Surveillance Specialist bears no authority-based accountability for what anyone does. Suppose, though, that Burglar coerces Alarm Specialist into accepting Surveillance Specialist’s promise. Maybe Alarm Specialist doesn’t want to work with Surveillance Specialist, but Burglar effectively twists Alarm Specialist’s arm into doing so. If the coercive threat is severe enough, Alarm Specialist will bear no authority-based accountability for what Surveillance Specialist does after all. This means if Surveillance Specialist were to foreseeably enlist assistance from some further individual, then Alarm Specialist would not bear authority-based accountability for what that fourth individual foreseeably does either. (However, Burglar bears authority-based accountability for what they all foreseeably do since Burglar remains uncoerced in accepting Alarm Specialist’s promise.) The result is that if an agent accepts an authoritative role in a chain of command only because she was coercively threatened into doing so, then, provided that the threat is sufficiently severe, she will bear no authority-based accountability, not only for what her immediate subordinates do, but for what their subordinates do as well.⁸ These cases can be combined in interesting ways to yield increasingly complex cooperatively committed harms. We can imagine cases where a participant is

⁸ See Appendix for further discussion of such cases.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

116

, ,  

coerced into making and accepting promises, or coerced into accepting some whereas she accepts others freely, or coerced into making some whereas she makes others freely. I will not explore further permutations here, since I have already laid out the rules for doing so: a coercively made promise does not undercut authority-based accountability, whereas a coercively accepted promise does. Hence, the degree to which a participant in a cooperatively committed harm will be accountable for what others do depends on the degree to which that agent is coerced into adopting the role of a deliberator in the division of agential labor.

4.3 Ignorance Participants in cooperatively committed harms often don’t know what’s going on. Such a participant might not know exactly what he has agreed to do, or might not know what others have agreed to do for him. Alternatively, he might know all these things and yet remain ignorant of whom he has made promises to and accepted promises from. And he might still know all these things and yet remain ignorant of the motivating reasons for which the others are acting. How does ignorance of these sorts affect authority-based accountability? Here, I address this question. I argue that the executor’s ignorance has little effect on whether the deliberator bears authority-based accountability. Conversely, the deliberator’s non-culpable ignorance can indeed mitigate the deliberator’s authority-based accountability. Throughout this discussion (as in the discussions of alienation and coercion) I focus on dyadic cases. In light of this, I conclude by discussing the mitigatory effects of ignorance in instances of more complex organizational relationships. And though I focus again on the divisions of agential labor we establish by way of promises, the lessons here apply also to the divisions of agential labor we establish by way of requests, agreements, and shared action as well.

4.3.1 Ignorant Executors and Ignorant Deliberators Suppose the executor in a division of agential labor is ignorant in some respect of a wrongful promise she has made. If she is ignorant of precisely those features that make the promise wrongful, does the deliberator still bear authority-based accountability? She does indeed, regardless of whether the executor’s ignorance is culpable. Consider the following case: Ignorant Executor Deliberator asks Executor to drive a car to a junkyard—with the added instruction that he refrain from looking in the trunk. Executor isn’t sure why Deliberator

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

117

wants this done. Neither is he sure what is in the car’s trunk. Nor is he sure who Deliberator is exactly. He suspects though, that Deliberator is a person of some influence, engaged in some sort of illicit activity. Executor agrees to do as Deliberator says, since Executor wants to ingratiate himself to Deliberator. Executor’s ignorance is irrelevant to the argument for authority-based accountability. According to the argument, Deliberator is accountable for what Executor does because Deliberator is accountable for a wrong-making characteristic of Executor’s conduct; specifically, the purpose of Executor’s conduct. Deliberator furnishes that purpose even when Executor is wholly unaware of what he is doing, for whom he is doing it, or even what the point is of what he is doing. After all, it’s Deliberator’s role to “call the shots”, and Deliberator counts as doing so even if Executor is left largely in the dark about what’s going on. The result is that Deliberator bears authority-based accountability for what Executor does, when Executor drives the car to the junkyard. The upshot, then, is this. The executor’s ignorance of what she is doing does not itself do much to affect whether the deliberator bears authority-based accountability for what the executor does. Suppose, though, that the deliberator is ignorant in some respect of a wrongful promise she has accepted. If she is ignorant of precisely those features that make the promise wrongful, will she still bear authority-based accountability? I will argue that she does, provided that her ignorance is culpable. Consider the following case: Ignorant Deliberator Executor promises to do something special for Deliberator tomorrow, and Deliberator gleefully accepts without knowing what Executor has in mind, which turns out to be ϕ: the theft of a laptop which Executor subsequently gifts to Deliberator. Does Deliberator bear authority-based accountability for what Executor does? The answer, unsurprisingly, is that it depends. We can fill out the case accordingly: Variant 1a Deliberator suspected Executor might do ϕ but was hoping he wouldn’t. Executor subsequently does ϕ in furtherance of the promise he made, since Deliberator does nothing to exclude ϕ from the promise she accepts. Does Deliberator bear authority-based accountability for what Executor does in this variant of the case? Deliberator bears authority-based accountability for the purpose that Deliberator furnishes for Executor. The practical reasons Deliberator takes there to be constitutively determine that purpose. Since, by hypothesis,

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

118

, ,  

Deliberator does not want Executor to do ϕ, it seems that Deliberator furnishes no such purpose for Executor. So, it might seem that Deliberator bears no authoritybased accountability for what Executor does. But this is too quick. Deliberator does indeed take there to be a practical reason for Executor to do something special tomorrow. It’s true that Deliberator did not take there to be a practical reason in favor of ϕ. But she did indeed take there to be a practical reason for Executor to act in way that happened to risk him doing ϕ. That is, Deliberator gave Executor the go-ahead, despite knowing that this meant that he might do ϕ. If you know that that there is a significant chance a promise-maker will fulfill a promise in a way you don’t intend, and if you know that it is reasonable for the promise-maker to interpret the promise in such a way, then that means of fulfilling it falls within ambit of that promise unless you exclude it, explicitly or implicitly, when accepting the promise. Because Deliberator suspected that Executor is likely to interpret the promise Deliberator accepted as a promise to do ϕ, and nonetheless accepted the promise, it is hard to take seriously Deliberator’s claim that she took herself to have reasons against ϕ’s occurrence. The upshot, then, is that when Deliberator accepts a promise suspecting that Executor might fulfill the promise in a morally problematic way, Deliberator bears authority-based accountability should Executor act in that way, even if Deliberator didn’t want Executor to do so. When we “look through” Executor’s conduct, to the practical reasons that Deliberator took there to be—the practical reasons that Executor’s conduct had the function of enacting—they include practical reasons that risked ϕ’s occurrence. Again, this is because Deliberator believes that Executor might interpret the promise accordingly. So, Deliberator bears authority-based accountability for furnishing a morally problematic purpose. The lesson here is that if you don’t know but still suspect that a promise you accepted might be discharged in a way that you don’t want, you’re still accountable at the bar of authority-based accountability, though the degree of blame you bear will be less than it would be if you outright intended that outcome. Consider, though, the following emendation: Variant 1b Deliberator suspects Executor might fulfill his promise by doing ϕ. That is, Deliberator suspects Executor might fulfill his promise by stealing a laptop and gifting it to her. Deliberator does not want Executor to do this. So, she instructs Executor to fulfill his promise in some other way. But she also suspects that Executor will do ϕ, contrary to her expressed wishes, which is what happens. Deliberator took there to be no reason for Executor to do ϕ, and at least ostensibly, ϕ does not fall under the ambit of the promise Deliberator accepted—after all,

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

119

she explicitly ruled it out. It would seem, then, that Deliberator harbors no authority-based accountability for what Executor does when he commits ϕ. Again, this is too fast. It’s true that Deliberator did not take there to be a practical reason in favor of ϕ. But Deliberator gave Executor the go-ahead, despite knowing that this meant that he might do ϕ contrary to her expressed wishes. This, in turn, affects how we evaluate the motivating reasons constitutively determining the purpose she furnishes for Executor. Despite the explicit content of the instructions she provided, Deliberator cannot ingenuously claim that she took there to be strong reasons for Executor to refrain from doing ϕ, given that she accepted a promise she knew he might very well discharge by doing ϕ. So, in keeping with the interpersonal division of agential labor they’ve established, Deliberator furnishes a purpose for Executor’s conduct, which, in this case, is to enact reasons for acting in a way that happened to risk doing ϕ. The result is that, again, Deliberator is accountable for a wrong-making feature of Executor’s conduct in that Deliberator’s morally problematic practical reasons partly determines the purpose of what Executor does. What, if, however, Deliberator fails to recognize that Executor might do ϕ? Variant 1c Executor believes, reasonably, that ϕ falls within the ambit of the promise he made. That is, Executor believes that the promise he made and the promise Deliberator accepted was a promise that included the possibility of stealing a laptop for her. Deliberator culpably fails to recognize that Executor might steal a laptop, as a result of which she fails to exclude that act from the promise she accepts. Executor subsequently steals the laptop in accordance with the promise he made. The analysis of Variant 1c is nearly the same as the analysis of Variant 1a. In 1a, Deliberator suspected that Executor might do ϕ. But in 1c, Deliberator culpably failed to expect as much. In both cases, though, it was reasonable to think that Executor would interpret the promise in such a way that ϕ falls within its scope. Suppose that Executor’s victim wants to know why she was mistreated. Can Executor correctly claim that ϕ falls outside the scope of the promise Deliberator accepted? After all, Deliberator did not want Executor to commit ϕ and did not foresee that eventuality. Nonetheless, Deliberator cannot ingenuously deny that ϕ falls within the ambit of the promise she accepted. If you know or are in a position to know that that there is a significant chance a promise-maker will fulfill a promise in a way you don’t intend, and if you know or are in a position to know that it is reasonable for the promise-maker to interpret the promise in such a way, then that means of fulfilling it falls within ambit of that promise unless you exclude it, explicitly or implicitly, when accepting the promise.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

120

, ,  

So, though Executor didn’t act in the way Deliberator wanted, Executor still acted in a way consistent with the promise Deliberator accepted. Deliberator is, then, accountable for a wrong-making feature of Executor’s conduct. In keeping with the interpersonal division of agential labor they’ve established, Deliberator furnishes the purpose behind Executor’s conduct, which, once again, is to enact reasons that risked doing ϕ. Deliberator might deny that she took there to be any reason for Executor to do ϕ; but because she was culpably ignorant of the fact that Executor is likely to interpret the promise Deliberator accepted as a promise to do ϕ, it is hard to take seriously Deliberator’s claim that she took herself to have strong reasons against ϕ’s occurrence. The upshot, then, is that when you accept a promise while culpably failing to realize that the promise might be reasonably taken in a way that you didn’t intend, you bear authority-based accountability should the promise-maker fulfill the promise that way, even if you didn’t want her to do so. But as in Variant 1a, you bear less authoritybased accountability than you would if you had intended that outcome. This is because the degree of authority-based accountability a deliberator bears depends on the motivating reasons that the executor has the purpose of enacting. The worse those reasons are, the greater the degree of authority-based accountability. In the examples discussed so far, Deliberator’s ignorance is culpable. Consider, in contrast, the following case. Variant 1d Deliberator non-culpably fails to realize that Executor might fulfill the promise Deliberator accepted by doing ϕ—i.e., by stealing a laptop. As a result, Deliberator non-culpably fails to exclude that possibility when accepting the promise. Executor consequently does ϕ. Does Deliberator bear authority-based accountability for what Executor does? Suppose she was deceived by a third party into thinking that there is no way that Executor will do ϕ in furtherance of the promise Deliberator accepted. In such a case, Deliberator is justified but mistaken regarding the content of the promise she accepted. She has, in effect, unknowingly accepted Executor’s promise to do ϕ. Deliberator does not take there to be any practical reasons for Executor to do ϕ; nor does she culpably fail to consider the victim’s wellbeing in having accepted the promise (unlike the previous example). The result is that the purpose Deliberator furnishes for Executor does not include or reveal any morally culpable motivating reasons. She consequently bears no authority-based accountability. Suppose, alternatively, that Deliberator’s failure is non-culpable for a different reason. Suppose that no one reasonable would have suspected that Executor would fulfill the promise by doing ϕ. In this case, Deliberator does not accept a promise that Executor will do ϕ, since it is by hypothesis unreasonable not only for Deliberator but for Executor—and indeed, anyone else—to interpret the promise

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

121

in that way. Since ϕ does not fall within the ambit of the promise Deliberator accepted, and since there is no reason for Deliberator to have thought that Executor might commit ϕ in furtherance of the promise he made, Deliberator again bears no authority-based accountability for ϕ. So far, I’ve considered cases in which Deliberator is ignorant—in one way or another—of the content of the promise she accepts. The upshot of these cases is that, unless Deliberator’s ignorance is non-culpable, she ends up bearing some authority-based accountability for what Executor does. Sometimes, though, a promise perfectly innocent in and of itself is wrongful in virtue of its contingent consequences. Suppose Deliberator accepts a promise from Executor that Executor will fly from overseas to see her. However, Executor knows that he is an asymptomatic carrier of a highly contagious, lethal disease. A prolonged flight will likely infect other passengers, putting them at grave risk. Thus, accepting the promise is wrongful—not because there is anything problematic about flying overseas in and of itself, but because of the wrongful harms doing so happens to cause. Likewise, whether a promise is wrongful can depends on who is performing it. For example, it is permissible for a parent to do things for his or her child, or a spouse to do things for his or her partner, that it would be impermissible for strangers to do. As a result, whether a promise is wrongfully accepted can depend on who is making the promise. Deliberator, who is running a daycare, might accept a promise from Executor, to pick up a child, thinking that Executor is the child’s guardian when in fact Executor is a stranger. Accepting this promise is wrongful, given who is doing the promising. In these cases, where a promise is wrongful not in and of itself but in virtue of its consequences or in virtue of who is making the promise, the analysis proceeds as it did for Ignorant Deliberator and its subsequent variants. In short, Deliberator bears authority-based accountability provided her ignorance of the relevant facts is non-culpable.

4.3.2 Ignorance in Cooperatively Committed Harms A lesson of the discussion on ignorance is that the deliberator’s culpable ignorance affects authority-based accountability in ways that the executor’s ignorance doesn’t. This affects accountability in the context of cooperatively committed harms when some of the participants are less than fully aware of what they’re doing together. Consider this case: Ignorant Cooperative Action 1 Person A hires seven others—B through H—to build a machine together. They have to cooperate in order to build this machine. Each of the eight individuals is given a small part of the machine to build. Since A will pay the seven only if they

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

122

, ,  

successfully build the machine, each of the seven makes a promise not only to A, but to one another, to do their part. The problem, though, is that they’re unsure what it is that they are together building. Nor are they quite sure who A is. Suppose the machine is a dangerous tool which A will use for nefarious purposes. The shared action they’re involved in can be represented as follows, where the horizonal arcs represent bilateral authority-relations, and vertical arcs represent unilateral authority-relations (Fig. 4.1). A

B

C

D

E

F

G

H

How does authority-based accountability disseminate in this case? Note that each of B through H is both a deliberator and an executor—unlike A who is only a deliberator. The result is that even though A has authority over every individual in this shared action, each of the participants potentially bears authority-based accountability for what they together do (see section 2.1.5, chapter 2). So, for example, H, by accepting a promise from each of B through G, establishes a division of agential labor with each of them, in which H counts as a deliberator; H has a claim over each of B through G—a claim yielding a protected practical reason—that they do their part in building the machine. The result is that H bears authority-based accountability for what B through H do; mutatis mutandis of each of them over what H does. By hypothesis, each of the seven is ignorant of what it is, exactly, they’re promising to do, in that they don’t know the nature of the machine the construction of which they’re contributing to. Nor are they aware of who A is or what will happen once they successfully construct the machine. Recall, though, from the discussion of ignorant promise-making, that the promise-maker’s ignorance does not itself vitiate the promise-accepter’s authority-based accountability for what the promise-maker does. So, A is still on the hook, despite the ignorance of B through H. The problem, though, is that each of B through H is not only a promise-maker, but also a promise-accepter. Suppose, then, that H, specifically, is non-culpably ignorant of the fact that the machine she is helping to build is a dangerous tool which A will use for nefarious purposes. All the evidence suggests—mistakenly—that the machine is benign and that A is no villain. Though building the machine is unjust, the practical reasons A took there to be in favor of doing so do not reveal any morally problematic attitude toward the potential victims of the machine. This is because H’s failure to realize that the machine is dangerous and that A will use it for ill, are nonculpable. So, the practical reason H furnishes as part of the purpose of what the

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

123

others do, does not make the actions of the others morally worse than they already are. (This scenario just describes a complex instance of Variant 2d. See section 4.3.1.) The result is that H bears no authority-based accountability for what B through G do; the same goes for each of B through G if they, too, are nonculpably ignorant in the way that H is. Given the discussion of ignorant promise-acceptance (in section 4.3.1), there are a variety of ways in which an agent participating in a shared action might be ignorant of what’s going on. If H suspects that something nefarious might be afoot in building the machine, she might qualify the promise she accepts by indicating that it is not binding if there is evidence that the machine is being used for nefarious purposes. A failure to do as much inculpates her at the bar of authority-based accountability in the ensuing harm. (This just amounts to an application of Variant 1a in section 4.3.1.) Suppose H does indeed explicitly indicate to the others that she does not want or expect them to do their part if evidence turns up suggesting that the machine will be used for wrongful purposes. In this case, if the other participants learn as much and nonetheless contribute to building the machine, H bears no authority-based accountability for what they do (provided it wasn’t reasonably foreseeable that they’d continue to so act). This describes an application of Variant 1d (in section 4.3.1); but the context of shared action adds the following important complication. Suppose that H learns that the other participants have discovered the true nature of the machine they are building, yet they continue to contribute to its construction anyway. Given the conditional nature of the promise H accepted, it follows that H has no practical claim against them that they do their part. Let us assume that the other participants did not analogously conditionalized the promise that they accepted from H. And let us assume that H never indicated, in making the promise to the others, that she’d do her part only absent evidence suggesting that the machine is to be used for nefarious purposes. It follows, then, that though H no longer has a practical claim against the others that they do their part, they still have a claim over her. The result is that they will bear authoritybased accountability for what she does, but not vice-versa. This might seem problematic insofar as it suggests that participants in a cooperatively committed harm can insulate themselves from authority-based accountability for what they’re together doing by including a “rider” in the promises they explicitly or implicitly accept, while simultaneously demanding no such rider in making promises to the rest of the group. This tactic, though, is unlikely to succeed. To see why, note the difference in the promise H accepts versus the promise she makes. H indicates that she will hold others to their promise only if there’s no evidence suggesting that the machine will be used wrongfully. Yet she promised to do her part regardless of whether there’s evidence suggesting that the machine will be used wrongfully. Though these two promises are logically consistent, there is

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

124

, ,  

something practically or prudentially in tension between them. H refuses to accept authority over others in the event that they are contributing to a wrongful harm; yet she accepts the authority of others over her in the event that she is contributing to a wrongful harm. Presumably, she refuses to accept authority over others when they are contributing to a wrongful harm, because she believes that, all-thingsconsidered, there is no good reason for others to contribute to a wrongful harm. But if this is why she divests herself of authority over others, then why does she unconditionally promise to do her part? That is, why does she invest authority in others with respect to doing her part in furtherance of a harm she thinks wrongful? Her reasons for divesting authority over others speak also in favor of refusing to invest authority in others. This suggests that either her stated reasons for unconditionally investing authority in others, or her stated reasons for divesting authority over others, are disingenuous at best or outright false at worst. The result is that it is difficult to accept at face value the simultaneous attempts at divestment and investment. One or the other has to give. If we reject the attempt at divestment, then H does indeed take herself to have a claim over others, despite her protestations to the contrary, that they contribute to building the machine even if the machine’s operations end up wrongful—a claim that yields a protected reason for the other participants to do their part. In such a case, H ends up bearing authority-based accountability for what they together do (and the other participants end up bearing authority-based accountability for what H does). Alternatively, if we reject the attempt at investment, then H does not take the others to have authority over her that she do her part, despite her protestation to the contrary. In such a case, H ends up bearing no authority-based accountability for what they together do; but neither do they end up bearing authority-based accountability for what she does. Indeed, in such a case, H would likely be viewed as a kind of interloper—an agent physically involved in building the machine, but who is not normatively involved, in that the other contributors do not see themselves as beholden to her, or her as beholden to them. In the case under consideration, the object of authority-based accountability—i.e., what it is for which the individuals involved are accountable—is a cooperatively committed harm. Consider, though, a case in in which the object of authoritybased accountability does not consist in what the participants together accomplish, but in what some of them do as a means or a side-effect of doing their part. Ignorant Cooperative Action 2 Each of A through H would like to do Σ together—a beach clean-up. So, each agrees to contribute to that end. H, in particular, agrees to procure the necessary equipment (extension grippers, bags, and vests). However, he does so by committing ϕ—he obtains the equipment they need by stealing it. None of A through G foresaw that H might do ϕ; nor did they want H to do ϕ.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

125

How does authority-based accountability work in this case? Again, each of A through H is both a deliberator and an executor. Any given participant potentially bears authority-based accountability for what all the other contributors do when they do their part in the shared action, so long as it falls under the scope of the authority she has over them. The result is that every participant potentially bears authority-based accountability for what H does. If they took H to have a protected reason to do ϕ, then they bear authority-based accountability for ϕ. But they didn’t, by hypothesis. Moreover, they reasonably didn’t suspect that H might do ϕ. If their ignorance in this respect is non-culpable, then they bear no authoritybased accountability for what H does. We can assume that there is no reason for any of them to believe that H might do ϕ. They do not furnish him with a purpose that includes or risks including theft. Consequently, they bear no authority-based accountability for what H does. Things are different, though, if their ignorance is culpable. From here things proceed analogously to how they did in the discussion of Ignorant Cooperative Action 1. If A through G suspect that H might contribute to Σ by doing ϕ, they might accordingly qualify the promise they accept by indicating that they do not want H to do ϕ. If they neglect to do so, and if there is indeed reason to believe that H might do ϕ, then that failure inculpates them in ϕ, at the bar of authority-based accountability. A through G might try to insulate themselves against authority-based accountability for what H does, by indicating when accepting a promise from him that they do not want him to do ϕ. But again, they cannot divest themselves of authority over H in this way, while simultaneously investing in H authority over them. That is, where it is reasonable to think that H might do ϕ, either A through G bear authority-based accountability for what H does and H bears authoritybased accountability for what they do, or, alternatively, A through G bear no authority-based accountability for what H does, and he bears no authority-based accountability for what they do. (Of course, we can imagine mixed cases in which some of A through G recognize that H might do ϕ, whereas others do not—and where for some a failure to recognize as much is culpable whereas for others it is not. The result is that some participants might be inculpated while others are not.) In summary: a cooperatively committed action might go wrong morally in a number of ways. The cooperatively committed action might be morally problematic in and of itself. Alternatively, the action, though morally unproblematic in and of itself, might become morally problematic if the wrong individuals are participating in it. Or the participants in the cooperatively committed action might contribute to it in morally problematic ways. Or the cooperatively committed action might have morally problematic consequences. Given these ways that a cooperatively committed action might morally go wrong, a participant might be ignorant of one or more of these ways. Though such ignorance can insulate the

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

126

, ,  

participant from authority-based accountability in cases where that ignorance is non-culpable, there are other ways to insulate a participant from accountability. The participant might specifically conditionalize the commitment she accepts from others by sincerely indicating that she does not want or expect them to participate in particular, morally problematic ways. But this method of forestalling authority-based accountability only works if the participant conditionalizes the agreement that she makes in the same way. There are other ways we might think that a participant might get out from under authority-based accountability. A participant might think that she avoids accountability of this sort if she does not want and would not knowingly accept an agreement from other participants in which they promise to contribute in a morally problematic way or in furtherance of a morally problematic shared act. But unknowingly accepting such an agreement can still serve as a basis for authority-based accountability if the ignorance involved is culpable.

4.4 Conclusion In this chapter I explored the effects that alienation, coercion, and ignorance can have on a division of agential labor, and the deliberator’s authority-based accountability for what the executor does. With respect to alienation, I argued that the deliberator’s alienation—not the executor’s—has the potential to affect authority-based accountability. Recall that one way to understand authority-based accountability is this: in a division of agential labor where the executor commits a wrong, ascertaining the executor’s purpose requires that we “look through” her conduct, to the deliberator who furnished the purpose of the executor’s conduct. But what happens when the deliberator doesn’t care whether the executor acts accordingly? Or when the executor doesn’t care whether she acts accordingly? I called these kinds of alienation “weak”. And what happens when the deliberator goes so far as to deny that she has any authority over the executor? Or when the executor denies that the deliberator has any authority over her? I called these kinds of alienation “strong”. I argued that as far as authority-based accountability is concerned, the deliberator’s alienation matters whereas the executor’s alienation does not. That is, the deliberator’s alienation—weak or strong—can mitigate her authority-based accountability, whereas the executor’s alienation—weak or strong—cannot. This should not come as a surprise. According to the argument for authority-based accountability, the practical reasons the deliberator takes there to be determine the purpose of the executor’s conduct, and this potentially serves as a wrong-making feature of that conduct. If the deliberator lacks those reasons, then they cannot serve as a wrong-making feature of what the executor does. And this is just what

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

127

happens when the deliberator is alienated from the protected reason she ostensibly confers upon the deliberator; she is weakly alienated if she does not care whether the executor acts in accordance with the protected reason in question, or she is strongly alienated in that the deliberator denies that the executor has a protected reason in the first place. As I argued, these kinds of alienation are not alone enough from which to conclude that the deliberator bears no authority-based accountability for what the executor does. Nonetheless, the deliberator’s alienation is at least of the sort capable of potentially mitigating her authority-based accountability since the reasons the deliberator takes there to be furnish the purpose of the executor’s conduct. I then turned to cases of coercion. Sometimes participation in a division of agential labor is less than fully voluntary. How does this effect authority-based accountability? Suppose an executor is coerced into making a promise or a deliberator is coerced into accepting a promise. The resulting promise is not binding qua promise. But I argued that this doesn’t vitiate the deliberator’s authority-based accountability. This is because it is possible to establish a division of agential labor by force—through a demand backed by a threat. The result is that the deliberator bears authority-based accountability after all. Things are not so simple, though, when a deliberator is coerced into adopting a role as such. I argued that though an agent can be forced to take on the role of a deliberator, she cannot be forced into making particular decisions. It is not conceptually possible to compel someone to decide for themselves while simultaneously telling them what to decide. The result is that in such cases, coercion, depending on its severity, can mitigate or eliminate the putative deliberator’s authority-based accountability. The upshot, in any case, is that the practical reasons a deliberator takes there to be will determine the purpose of the executor’s conduct only if the deliberator is relevantly free to decide accordingly. Otherwise, the individual who is actually calling the shots—that is, the individual coercively threatening the putative deliberator—is the individual whose motivating reasons determine the purpose of the executor’s conduct. I ended the discussion of coercion by considering authority-based accountability in the context of complex cooperatively committed harms involving several participants. Finally, I considered the role of ignorance in authority-based accountability. As is the case of alienation and coercion, it’s the deliberator’s rather than the executor’s ignorance that matters when it comes to authority-based accountability. The executor’s ignorance of what she is doing, why she is doing it, or for whom, does not itself affect the deliberator’s authority-based accountability for what the executor does. The deliberator might choose to leave the executor completely in the dark, and yet the deliberator can still bear full authority-based accountability for what the executor does. Things are different when the deliberator is ignorant in important respects of what she has directed the executor to do. A deliberator might have no idea how an

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

128

, ,  

executor will fulfill the protected reason she has. The executor might do so in ways that the deliberator does not anticipate or does not want—as is the case when an acquaintance who promised to repay a debt unexpectedly does so by committing a theft. Does the deliberator bear authority-based accountability in such cases? At first it might seem that the deliberator does not; after all, the reasons that the deliberator takes there to be are what determine the executor’s purpose. If the deliberator did not take there to be a practical reason for the executor to act in the way that she did, then the deliberator did not confer upon the executor the purpose of acting in that problematic way, in which case the deliberator bears no authority-based accountability for it. But I argued that if the deliberator suspected (or otherwise should have suspected) that the executor might fulfill her protected reason in that problematic way, and still gave the executor the go-ahead, then the reasons the deliberator took there to be evidently included reasons to risk the problematic outcome. The result is that the deliberator bears authority-based accountability after all, though perhaps not as much as she would if she had wanted the outcome in question. That being said, if the deliberator’s ignorance of the executor’s intentions is nonculpable, then the deliberator bears no authority-based accountability. I ended the discussion of coercion by considering more complex cases of cooperatively committed actions, in which multiple individuals working are ignorant of what they’re doing, or ignorant of how others will fulfill their respective tasks.

Summary of Part I I have completed a preliminary account of the argument for authority-based accountability. I hope to have presented a theoretical basis for the view that, if you confer upon me and I accept a protected practical reason to do as you say, it is possible for the practical reasons you take there to be to determine constitutively the purpose of my conduct. If the practical reasons you take there to be are morally problematic, the purpose you furnish for me is morally problematic as well. If you are accountable for that difference you make, you bear what I called “authority-based accountability” for what I do. The result is that promises, requests, agreements, shared action, and so on, can all ground authority-based accountability. The cases of discussed are for the most part pretty abstract. For this reason, I will cap the discussion so far by applying authority-based accountability to a real-life case.⁹ On February 15, 1947, a taxi driver—Thomas Watson Brown—was robbed and stabbed to death near Greenville, South Carolina. The next day, local authorities

⁹ In describing this case, I draw from (Gravely, 2019), (West, 1947), (Hoover, 2003).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

129

arrested Willie Earle, a 24-year-old Black man. He was taken from his mother’s house to the county jail where he was held based on highly circumstantial evidence. That night, Greenville taxi drivers organized a convoy of cabs; they drove to the Pickens County jail and demanded Willie Earle. Once in their hands, the cab drivers beat, stabbed, and shot Earle to death. Over the next four days, the US Marshal and the FBI charged thirty-one White men, twenty-eight of whom were taxi drivers. They all signed confessions. In an infamous court proceeding, an all-White jury acquitted all thirty-one suspects. In what follows, I describe the events on the night of Earle’s murder. I will then apply the argument for authoritybased accountability to those at all connected to the murder, whose names or descriptions I have indicated in bold. I do so with the purpose of demonstrating how the argument can shed light on their accountability. Roosevelt Carlos Hurd, a 45-year old taxi-driver, also working as a taxi dispatcher, allegedly organized the lynching. According to one defendant’s confession, Hurd was working on the switchboard at the Blue Bird Taxi Office when he made several telephone calls to other taxicab companies in Greenville. Each time he called, Hurd mentioned who he was. He asked representatives from the company to see “how many men it wanted to go to Pickens.” The context was clear: the purpose of going “to Pickens” was to confront Earle who was being held at the county jail there and to do him harm. Indeed, according to the signed confession, several individuals at the Blue Bird Cab Office said that “the nig–r ought to be taken out and lynched.” Not everyone Hurd called agreed though. In particular, one taxi-cab driver, U. G. Fowler, gave evidence that he had been asked to join the party but had refused. (He was later beaten and threatened into silence.) Hurd effectively put himself in charge; the other taxicab drivers seemed to defer to him. According to one tax-cab driver, Hurd asked him to take several men, including Hurd himself, to another cab company where they would gather more men. He responded by saying to Hurd that “he was the boss” and accordingly did as he was told. The confession continues: At the Yellow Cab Company, we met all the other cab drivers from the cab companies. After all got organized, the orders given me by R.C. Hurd were to go back and pick up my cab at the Blue Bird Office. I would like to say here that Hurd had already made arrangements for everybody to meet at the Yellow Cab Company.¹⁰

At the Yellow Cab Company, several decided against participating. The crowd jeered at them, labeling them “yellow” and “chicken.” From there, court

¹⁰ (West, 1947).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

130

, ,  

documents confirm, those who agreed to participate agreed to meet at Riverview Tavern just within Pickens County. This gave drivers a chance to fill their gas tanks, change their oil, and engage in other logistical operations necessary for the crime, which included picking up other participants. This included Hubert Carter, a 33-year-old driver and father of four, who explained in a statement to the police that another driver, Paul Griggs, asked him if he “wanted to go with the others to get the Negro being held for stabbing Mr. Brown.” Carter responded by saying that he’d “go along with the crowd.” One cab driver, Woodrow Wilson Clardy, was late to the rendezvous at the tavern; he stopped to pick up a shotgun and some shells. From here, Hurd once again took charge. He made clear that he would speak for the group at the county jail, since he knew the jailor. He also warned the group that he would not tolerate any attempt to defend Earle. Everyone—fifteen cabs in total—was to follow him and Clardy in the lead car. The convoy proceeded to Pickens County jail. Once at the county jail, John C. Joy and John Marchant used their cars’ headlights to illuminate the front and side entrance of the building. Hurd, Clardy, and Hendrix Rector knocked on the door, and roused the jailor, J. E. Gilstrap. Hurd carried Clardy’s shotgun. Meanwhile, several taxicab drivers milled outside, while others stayed in their cars with their doors opened either to hear what transpired or in preparation for a quick getaway. Duran O. Keenan, holding a shotgun, stood guard in front of the crowd of taxi drivers. Hurd addressed the jailor; he indicated that they had no intention of harming him, and that they had just come for Earle. The jailor asked for time to put on his shoes and coat, due to the night’s cold; Clardy interjected by denying the jailor’s request, in the interests of haste. The jailor went to retrieve the keys. In doing so, he told the men “I guess you boys know what you’re doing.” His only protest was that he asked the men to refrain from using profanity, since his wife and daughter were upstairs. Several more men had entered the building by now. Jesse Lee Sammons secured the phone lines, while others went to retrieve Earle who was passed from man to man, and eventually shoved into the lead car. The convoy then departed. Some returned home at this point. Most pressed on. During their procession, Joy shouted from his cab to Clardy’s that they should act before they reached Greenville. During the drive, one of the cabs broke down; its four occupants returned on foot to Riverview Tavern. They tried to rouse the tavern’s owner to secure a ride, but ultimately missed the ensuing murder. One of them later mentioned that “he sure wished he could have gotten to the party.” According to Hurd, one of the taxi-drivers in the Yellow Cab knelt on the front seat where he exhorted Earle: “You know we brought you out here to kill you. You don’t want to die with a lie in your heart and on your tongue.” According to one statement, when Earle denied killing Brown, Rector demanded that they “get it

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

131

over with.” Ernest Stokes, a former deputy sheriff, opened a pocketknife and said, “Before you kill him, I want to put the same scars on him that he put on Brown.” When Hurd pointed a gun at Earle, the cab’s driver, Clardy, stopped him and cried out, “Don’t kill that negro in my cab, that’s where I make my living.” He instead led the convoy, now consisting of seven cars, to a secluded spot near Judge Martin’s family property. Griggs, riding in Joy’s cab, announced his intention to take the lead since he had the other firearm. But his car ended up taking a wrong turn, perhaps confused about where exactly they were to meet. Arriving late, they missed Earle’s torture and subsequent murder. Finally halting near a slaughterhouse, several of the remaining men surrounded Earle. There, the questioning continued. One of the men urged Earle to confess, reminding him that “he didn’t have long to live.” Another man suggested that they take Earle to the hospital so that he could identify the man they believed he had killed. Others demanded that Earle divulge whether he had any accomplices. During a lull in the inquisition, Rector grabbed Clardy’s shotgun and with its butt he beat Earle savagely. Others joined in. Rector proceeded to knife Earle; Griggs and Stokes might have joined in. Earle cried out “Lord Mercy, you all done have killed me.” Clardy grew vexed at the damage that the beating was causing his shotgun. He yelled, “If you’re going to kill him with the gun, kill him; don’t get blood all over it.” Someone at that point heard Earle whisper “I’m dying now.” Hurd, grabbing the shotgun said, “Let’s get it over with,” after which he shot once from the singlebarrel gun. He subsequently requested a second shell, and then another. Griggs and Keenan allege that they refused, adding, “Come on, R.C. He’s dead now.” Part of Earle’s face and head had been blown away. Just as the shooting occurred, Joy’s cab pulled up. One of the occupants remarked, “Hell, the negro is dead. Let’s go.” The conspirators scattered. By this time, it was nearly morning. Some returned home, while others began putting cabs back into service, altering the mileage and gas records for the taxis used in the murder. The county medical examiner testified that Earle’s death resulted from gunshots to his head, though he suffered a vicious beating and five knife wounds. A cook at the Southern Cafe, who had apparently been aware of the lynching, asked if it had gone off as planned, as he served several of the conspirators breakfast. At some point one of them contacted the newspaper office and the Black mortuary with the same message: a “dead nig–r” could be found out on Bramlett Road. What does the argument for authority-based accountability say about this heinous affair? We can begin by dividing up the timeline according to several key events: a. Making calls at the Blue Bird Cab Office b. Meeting at the Yellow Cab Company c. Meeting at the Riverside Tavern

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

132

, ,  

d. Abducting Earle at the Pickens County Jail e. Murdering Earle on Bramlett Road It seems that Hurd was either the person in charge of the entire operation, or at least one of those in charge. He was a key figure in organizing the men at the Blue Bird and the Yellow Cab offices—and he explicitly put himself in charge at the Riverside Tavern. He directed the lead car in the convoy to the jail, he addressed the jailor holding Earle, he was among those who threatened Earle in the cab, and he was the one who ultimately shot Earle. Hurd’s culpability in the lynching is manifest. There were, of course, a host of acts that he didn’t personally commit, such as: driving the cars, securing the phone lines at the county jail, manhandling, threatening, and stabbing Earle, and so on. But Hurd was nonetheless an important cause of those events—intentionally so. We needn’t, then, advert to authoritybased accountability to put Hurd on the hook for all or nearly all that transpired. But what about the others? Recall that at the Yellow Cab Company office several of the men there declined to participate in the plan; they subsequently left. They are, at worst, accountable for failing to warn the authorities. But what about those who agreed to proceed to the Riverview Tavern? Upon meeting there, Hurd effectively indicated that he was in charge. The men, by acceding to his instructions, adopted the role of executors; and Hurd, by proffering those instructions, adopted the role of deliberator. But, importantly, the men under Hurd counted as deliberators as well insofar as they all tacitly agreed with one another to cooperate in furtherance of “getting Earle.” So, it is not just Hurd who bears authority-based accountability for what the other men did in foreseeable furtherance of “getting Earle”; each of the men also bears authority-based accountability for what the other men did in foreseeable furtherance of that goal. What exactly does it mean to “get Earle” though? The men might have a variety of intentions. Some might just want to question him. Some might want to beat him up—perhaps as a means to extracting a confession, or as a means of determining whether he had any accomplices. Some testified that they wanted to take him to the hospital to see if he can identify Brown. Others might want to kill him, but only if he confesses, whereas still others might want to kill him regardless. Some might have no particular intentions as to what happens to Earle following the kidnapping; instead, they might have wanted to help the others kidnap Earle after which they leave it up to them what’s done to him. And still others might not want any of this. Instead, they agree to go along solely to avoid being labeled a coward such as those who, at the Yellow Cab Company office, declined to participate. Some of these motivating reasons are clearly morally worse than others. On the argument for authority-based accountability, any given cab driver confers upon every other cab driver a protected reason to enact the practical reasons he takes there to be, and thereby has that concomitant purpose. Thus, any given cab driver

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

133

is accountable for a wrong-making feature of what every other cab driver does: the purpose for which he acts. But as we’ve seen, the purpose that the cab drivers furnish for one another can vary in morally important ways. The upshot is that each cab driver’s authority-based accountability varies with the immorality of the purpose he confers upon the others. So, for example, a cab driver who wants Earle to be killed outright furnishes such a purpose on all the others and is thus accountable for the fact that they have that purpose. A cab driver who, alternatively, agrees to participate in the kidnapping solely to avoid humiliation also furnishes a wrongful purpose on the others: the purpose of helping him avoid humiliation even at the cost of harming an innocent. He is thus accountable for the fact that they have that purpose. Since this purpose, though wrongful, is not as wrongful as furnishing outright murder as a purpose, he will bear less authority-based accountability than his more bloodthirsty counterparts. The upshot is that though all the cab drivers bear authoritybased accountability for Earle’s murder, the accountability they bear differs based on their motivating reasons for participating. But what about Carter who was not present at the Riverview Tavern? Recall that Griggs recruited him shortly thereafter, at which point Griggs agreed to “go along with the crowd.” At first, it might seem that Griggs does not count as a deliberator or an executor for anyone other than Carter, since Griggs wasn’t at the meeting in Riverview Tavern. Indeed, no one other than Griggs even knew that Carter was involved, at least at first. The statements from the drivers indicated, however, that they had more or less agreed to recruit as many other interested cab drivers as possible. They, in effect, authorized one another to do so. As such, Carter counts as a deliberator and an executor not just for Griggs, but for everyone else as well. At the Pickens County Jail, the men adopted different tasks. Joy and Marchant provided lighting; Keenan stood guard; Hurd and Clardy addressed the jailor, Sammons secured the phone lines; others manhandled Earle. But even those who waited outside in the cars doing nothing bear authority-based accountability for what the others did, in virtue of their status as deliberators in a division of agential labor. They had, after all, tasked one another with retrieving Earle, albeit for potentially different reasons. What about Gilstrap, the jailor? Hurd demanded of him that he release Earle. Perhaps he ought to have protested more. Facing at least two shotguns, any such protestations would have likely been futile. But it would at least serve to indicate that he did not condone their actions. His sycophantic suggestion that the men “knew what they were doing” suggests a degree of affinity with them. Against this, it might be thought that the men intimidated Gilstrap into such obsequiousness. But the fact that he had sufficient backbone to ask that they refrain from cursing suggests otherwise. Regardless of whether we think Gilstrap’s conduct reveals a faulty character, it is clear that at no point did he serve as a deliberator for any of the men. He conferred no protected reasons upon them. And though he enabled

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

134

, ,  

their plans by serving as an executor, he was forced to do so at gunpoint. The result is that he bears no authority-based accountability for what they subsequently did to Earle. Recall that after Earle was kidnapped, several of the cab drivers declined to participate further. This might very well be in keeping with the agreement they made to “get Earle” from the county jail. Having successfully done so, they might have decided to leave the rest of the crime in the capable hands of Hurd, Clardy, and the others. Another possibility is that some of the cab drivers had a change of heart; perhaps upon actually seeing Earle accosted, they decided to abandon their previous intention to assist in carrying out whatever it is that Hurd, Clardy, and the others had in mind. This difference might seem to matter as far as authoritybased accountability goes. Those who went home on the grounds that the others can effectively “take it from here” still function as deliberators; they have furnished for Hurd, Clardy, and the others a particular purpose: that of “finishing what they started.” Should Hurd, Clardy, and the others subsequently let Earle go, those who had decided to leave things in their hands would have a basis (albeit a non-moral one) for complaint. They were all presumably aware that though they might harbor different ideas as to what counts as “finishing what they started,” murdering Earle is a live possibility. In as much, each participant, then, furnishes the others with the purpose of harming Earle in a way that might include murder. So, even those whose participation ended at the county jail can bear authority-based accountability for Earle’s subsequent murder. An upshot is that a cab driver who causally contributed little or nothing to the criminal enterprise, and who possessed no unilateral authority over anyone, can end up bearing authority-based accountability for Earle’s murder in virtue of his status as a deliberator. The same goes for those who tried but failed to participate personally in the murder, such as those whose car broke down on the way to the murder site, as well as those who lost their way to the murder site. They too bear authority-based accountability for the murder. What about those who, after the kidnapping at the county jail, had a change of heart and for that reason ceased any further participation? For the sake of expository convenience, let’s invent a name for such a person—call him Neander. He is surely a better person, than, say, Griggs and Joy who tried and failed to participate personally in the ensuing murder. Neander might within the confines of his own conscience, release the other drivers from their putative obligation to continue the criminal enterprise. But so long as he does not communicate that intention to Hurd, Clardy, and the others, the promise to them remains “in force.” Despite his change of heart, Neander cannot ingenuously claim that he is no longer part of a division of agential labor with Hurd, Clardy, and the others. He accordingly retains authority-based accountability for what the others foreseeably do in accordance with the agreement he made.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

135

Still, it might seem procrustean to lump Neander with the likes of Hurd, Clyde, and the others. The argument for authority-based accountability, though, can accommodate that difference. Neander’s change in heart suggests he now believes there aren’t sufficiently good reasons to harm Earle any further. But, evidently, he also believes that there aren’t sufficiently good reasons to prevent Earle from being harmed. After all, Neander not only refrained from communicating his change in heart to the others but also refrained from contacting the authorities. In this limited respect, he is akin to those who declined to participate in the first place. But unlike them, Neander is now serving as a deliberator and executor in a division of agential labor and has not declined to change his status as such. The promise he accepted remains in force; whether he likes it or not, he has conferred upon them the purpose of doing harm to Earle. Neander can’t change that without indicating as much to Hurd, Clyde, and the others. But he does indeed determine, on his own, why they are to harm Earle: not to give him “what he deserves,” but because that’s necessary for Neander to withdraw himself from this whole affair free from recrimination and humiliation. And that purpose is wrongful, albeit not as wrongful as the purpose of “giving him what he deserves.” An upshot is that those who experience a change of heart and consequently decided to go home after the kidnapping at the county jail without indicating at all that their reasons for doing so, bear authority-based accountability for what the others do in pressing on. But they do not bear as much authority-based accountability as the likes of Griggs and Joy who purposed the others with retributively murdering Earle. Recall that at the scene of the murder, at least one man suggested that they take Earle to the hospital so that he might identify the murder victim. Call him “Leander”. The men presumably knew that there is room for reasonable disagreement as to what they should ultimately do to Earle. But, again, the men knew that murdering Earle was a live possibility; by furnishing for the others a purpose that included murder as a live possibility, Leander thereby inculpated himself in that subsequent murder, in that he bears authority-based accountability for it—even if that is not exactly the outcome he wanted, and even if he was not among those who stabbed or shot Earle. The next day, one of the drivers contacted the local newspaper and the Black mortuary. He also bears authority-based accountability for Earle’s murder provided he was “in on it” the night before—that is, provided he functioned as a deliberator who furnished for the others the purpose of giving Earle “what he deserves.” What about the cook? He apparently knew beforehand about the lynching and is accordingly culpable for failing to do anything to stop it. But he does not bear authority-based accountability unless he accepted a promise from the others that they would commit the murder. And what of Fowler, who refused to participate, but who also refused to identify those who did? He might be accountable for the resulting miscarriage of justice, but his refusal to make or

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

136

, ,  

accept any promises to lynch Earle means that he bears no authority-based accountability for the murder. Different taxi-cab drivers, then, played different roles in the lynching of Willie Earle. Some functioned as recruiters, some as drivers, some as guards, some provided lighting, some threatened the jailor, some secured telephone lines, some manhandled Earle, some interrogated him, some beat him, some stabbed him, and some shot him. Many did nothing in particular—they just accompanied those who played these different roles. Some left after their specific tasks were completed, well in advance of the actual murder. The argument for authoritybased accountability provides a way to ground the claim that each man is accountable for Earle’s murder, regardless of whether what he did actually contributed to that murder. To show that each cab driver is “on the hook” for the cooperatively committed murder, we needn’t “causally trace” what each cab driver did. Rather, each is severally accountable for a wrong-making property of the killing: the purpose for which it was done. And though the various men might have had different ideas as to what fate Earle should ultimately suffer, they all knew or were in a position to know that the agreement they accepted included the possibility of murdering Earle. By conferring this purpose upon one another, each ends up bearing authority-based accountability for Earle’s lynching. Though there is much more that can be said about the preliminary account I have developed, I believe it is more fruitful to turn to applications of authoritybased accountability. I do so in the second part of this book.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

PART II

APPLICATIONS

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

5 War Ethics Warfare is a cooperative activity—it is replete with cooperatively committed harms institutionally organized according to a hierarchical command structure.¹ For these reasons I apply the account of authority-based accountability to warfare. I begin, in section 5.1 by articulating a particular challenge that contemporary war-ethics faces. On some contemporary accounts, a combatant fighting in furtherance of an unjust aim is liable to be targeted only if she contributes to that aim. This view has difficulty accommodating the liability of those combatants who play highly subsidiary or peripheral roles in war, whose individual contributions are negligible. I will then explain in section 5.1.2 how the account I’ve developed can help address this challenge, after which I apply the account in section 5.1.3 to a particular example of a cooperatively committed harm in war. In section 5.1.4 I respond to the worry that my account illicitly inculpates civilians as well. I apply the account of authority-based accountability to warfare for another reason. War proves troublesome for my account since warfare is replete with three kinds of blame-mitigating circumstances. First, soldiers are often ignorant of the aims for which they are fighting. And even where they are cognizant of these aims, they remain ignorant of the empirical facts relevant to morally evaluating such aims. Sometimes this ignorance is incidental, in that it is challenging for anyone to determine exactly what is going on in the war being fought let alone the consequences of waging the war. Layers of bureaucracy make it difficult to ascertain what decisions those in the higher echelons are making. Other times their ignorance is the result of deception in that the military or the government either feeds the soldiers lies or otherwise conceals the truth about the aims for which they are fighting. Youth, military training, and cultural indoctrination leave soldiers all the more susceptible to deception. The resulting ignorance can affect authority-based accountability in war. I address this issue in section 5.2.1. Second, soldiers are often alienated from the larger aims of the war in which they are fighting. A combatant might not care whether the ultimate aims of the war are achieved. She might care only about getting paid, or earning the respect of her peers, or accomplishing her mission, or simply getting home in one piece. Or,

¹ Parts of Section 5.1 come from (Bazargan-Forward, 2020).

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0006

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

140

, ,  

a soldier might have a highly developed sense of duty toward those in his immediate unit, while treating others, at least in the confines of his own conscience, with contempt. How does this affect his authority-based accountability for what his comrades do—and the authority-based accountability of his comrades for what he does? I address this issue in section 5.2.2. Third, soldiers are often coerced into fighting wars. Governments coerce conscripts with threats of fines, imprisonment, or death. But circumstances, such as dire financial difficulties, can also “coerce” individuals into enlisting. And even among those who are not coerced into participating, the physical and psychological exigencies of combat can have a coercive effect on their decisionmaking in the heat of battle. If coercion can mitigate accountability, do coerced combatants bear authority-based accountability for what their comrades do? And do uncoerced combatants bear authority-based accountability for what coerced combatants do? I address this issue in section 5.2.3.

5.1 Revisionism and Authority-Based Accountability in War There are two assumptions that have guided theorists writing on the ethics of war, since the scholastics and jurists of the late Renaissance and Early Modern periods (from roughly the sixteenth through the nineteenth centuries). First, war is a relation between states (or between a state and a collective aspiring to statehood) rather than a relation between individuals. Second, the moral rules governing conduct in war do not depend on the morality of the war being fought. These doctrines form the basis of what has come to be known as Just War theory. Since the turn of the century, however, the doctrines grounding Just War theory have come under sustained attack. According to this new revisionist critique, war is a relation between individuals rather than states. Consequently, individual soldiers who participate in an unjust war cannot absolve themselves of accountability for what they do by adverting to their role as instruments of the state. Hence, the moral rules governing conduct in war (jus in bello) do in fact depend on the morality of the war being fought (jus ad bellum). This view undermines two mainstays of traditional Just War theory. According to revisionists, 1) the moral equality of combatants is mistaken, and 2) not all civilians enjoy blanket moral immunity from attack. Revisionism, though, has difficulty explaining why the combatants on the unjust side in a war, who are fighting in furtherance of an unjust cause, are morally liable to be attacked even when the harms they commit or cause are negligible. I address this difficulty by arguing that such combatants, as participants in shared action, are morally liable to be attacked on the grounds that they bear authority-based accountability for what their cohorts do.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

141

5.1.1 Revisionist Accounts of War Ethics What makes a combatant a morally permissible target in war? On the orthodox view, all active-duty combatants—even those whose actual contribution to the war is small or non-existent—are in principle fair game. The most influential defender of the orthodox view is Michael Walzer. He argues that combatants are morally permissible targets since they are all “currently engaged in the business of war”.² By having made themselves into combatants they have made themselves dangerous. The orthodox view, though, has been heavily criticized, most prolifically by Jeff McMahan.³ He argues that whether you pose a threat is neither a necessary nor a sufficient condition for losing your right not to be killed. A combatant does not lose her right to be killed when she poses a morally justified threat, no more than the victim of an attempted mugging loses her right not to be attacked when she engages in necessary and proportionate self-defense. Hence, combatants who are fighting in accordance with the rules of war and in furtherance of morally just aims have done nothing to lose their right against being killed. That is, such combatants are not morally liable to be killed. On this revisionist view, combatants who are acting in furtherance of a war’s unjust aims have indeed forfeited their right against being subjected to necessary and proportionate defensive violence. They are potentially liable to be killed, in that killing them does not wrong them provided that doing so is necessary to achieve a sufficiently important good. Such a combatant is morally liable because (and only if) she is accountable for her contribution to an end which violates the rights of others. So, a combatant who scrupulously targets solely military installations and personnel still thereby makes herself liable to be killed defensively if she does so in furtherance of an unjust aim, whereas her targets, by attacking her, are not morally liable to be killed if their attack is a necessary and proportionate act in furtherance of a just aim. So, on this revisionist view, only combatants accountable for contributing to an unjust cause (or against a just cause pursued by unjust means) are morally liable to be targeted. If this is correct, the orthodox view is mistaken in its claim that all combatants are morally entitled to kill enemy combatants. The revisionist view faces a challenge, though. On the revisionist view, a combatant is morally liable to be killed only if she contributes to an unjust cause (or to a just cause unjustly). Moreover, that contribution must be substantial—so substantial that she loses the right against being subjected to necessary and proportionate lethal violence. However, a significant portion of combatants, such as many of those who occupy subsidiary support roles,

² (Walzer, 2000, p. 43).

³ (McMahan, 2009). See also (Frowe, 2015).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

142

, ,  

contribute very little to the war being fought. Indeed, most combatants never see combat. If, on the one hand, these combatants are not morally liable to be killed, then how do we distinguish the liable combatants from the non-liable ones? Waging a morally just war would seem to be pragmatically impossible. Anticipating this problem, McMahan argues that the threshold for liability is quite low; as a result, virtually all combatants acting in furtherance of an unjust aim will likely be morally liable to be killed.⁴ Other responses are possible, though. Bradley Strawser, in defense of the revisionist view, argues that we should drop the bivalent distinction between combatant and noncombatant in favor of a more complex scheme in which we track levels of liability.⁵ Still others maintain that appealing to the liability of combatants cannot justify intentional killing in war. For example, Janina Dill and Henry Shue argue that most combatants participating in furtherance of an unjust aim do not contribute enough to make lethal attack necessary and proportionate.⁶ Recently, a different strategy for defending the revisionist view has emerged. I have argued elsewhere that most combatants—including those whose contributions to an unjust aim are insignificant—are typically morally liable to be attacked because they are engaged in cooperatively committed harms.⁷ Likewise, Adil Ahmad Haque argues that each participant is morally accountable not just for her own contribution to that cooperatively committed harm, but for the shared action itself. He says, “[i]n general, persons are morally liable to defensive killing only if they pose unjust threats directly, jointly with others, or indirectly through others they control.”⁸ So when combatants cooperate in furtherance of a mission, each is typically accountable for what they together do, since “members of a military unit performing coordinated roles in a combat operation are paradigmatic joint perpetrators.”⁹ Haque elaborates further: . . . several people jointly kill another by together performing a joint action that causes the other’s death, without subsequent intervening agency by others and, typically, without which the victim would not have died. [ . . . ] For example, if A restrains V while B stabs V to death, each performing coordinated roles in a common plan, then A and B together kill V. Importantly, joint actions are the actions of each participant, not solely the actions of the last participant to perform her role or only the actions of those participants without whose participation the plan would not have succeeded. For example, in the previous case, A and B together kill V even if B’s participation was counterfactually unnecessary because V would not have resisted. What we do, I do. When we kill, I kill.¹⁰

⁴ (McMahan, 2009). But see (Lazar, 2010). ⁵ (Strawser, 2011). ⁷ (Bazargan, 2012). ⁸ (Haque, 2017, p. 57). Emphasis is mine. ⁹ (Haque, 2017, p. 59). ¹⁰ (Haque, 2017, p. 65).

⁶ (Dill & Shue, 2012).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

143

This view has an intuitive appeal; indeed, Haque isn’t the only one to have proffered it.¹¹ Yet we need to provide grounds for this view. What we need is an account explaining how and why cooperation dramatically expands the scope of accountability.¹² The argument for authority-based accountability steps in at this point to fill the gap.

5.1.2 Cooperatively Committed Harms in War I argued in the first half of this book that a deliberator in a division of agential labor bears authority-based accountability for what the executors do qua executors. In a cooperatively committed harm where all the participants serve as both deliberators and executors, each participant bears authority-based accountability for what the others foreseeably do qua executor. If this account of authority-based accountability is correct, causal overdetermination (whether concurrent or preemptive) in the context of cooperative activity is no impediment to accountability. Suppose many individuals together agree to assist one another in furtherance of some wrongful cooperatively committed harm ϕ which each of them hopes to promote. Though each participant’s contribution to ϕ is negligible, together they achieve ϕ, which, we can suppose, constitutes a substantial harm. On a standard account of accountability, any given participant is accountable at most for her contribution to what she causes (or what she could have prevented), which by hypothesis is negligible. But given authority-based accountability, any given participant will be potentially accountable for the fact that what the others do is wrongful. We have an argument, then, for the view that an agent participating in a cooperatively committed harm with others can be accountable for more than her contribution to that harm. For example, Haque writes: . . . when individuals indirectly or jointly threaten to kill they are directly liable [ . . . ]. In these cases, we do not kill one person to prevent another from killing; instead, we kill one person to prevent her from killing through or with others. It follows that most direct participation in hostilities on behalf of an unjust cause falls within the paradigm case of moral liability to defensive killing.¹³

¹¹ See (Kutz, 2000), (Lepora & Goodin, 2013, p. 8), (Miller, 2006, pp. 177–178, 181), who adopt versions of this view. ¹² One might try to causally ground the claim that each soldier is responsible for what they together do in the following way. Any given soldier participates because other soldiers participate. In this respect, each soldier exerts a small, aggregable causal influence on every other soldier. (This is assuming we can make sense of degrees of causation. For a recent overview of this issue, see (Kaiserman, 2018).) But even then, each soldier will end up responsible for only a small fraction of what they together do, which will likely fall short of what’s necessary for liability to be killed. ¹³ (Haque, 2017, p. 66).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

144

, ,  

The argument for authority-based accountability can ground this view. Each soldier engaged in a cooperatively committed harm will end up at least partly accountable for the fact that what the other soldiers do is wrongful, in virtue of the protected reason that each soldier confers upon every other soldier that she do her part (as discussed in chapters 1 and 2). This means any given soldier bears authority-based accountability for what other soldiers do in conformity with the protected reasons she confers upon them; likewise, every other soldier bears authority-based accountability for what she does in conformity with the protected reasons they confer upon her.¹⁴ The argument for authority-based accountability, though, might seem to prove too much when applied to the context of war. It seems to have as a consequence that each soldier bears authority-based accountability not only for what she does but for what every other soldier on her side does as well. This is because each soldier might be described as participating in the cooperatively committed harm in which the war itself consists. By virtue of joining the armed forces, soldiers implicitly (if not explicitly, via an oath) agree to protect one another, qua soldiers, and to assist one another in furtherance of their shared cause. This is tantamount to multiple, pairwise, agreements between all the soldiers in the armed forces. There is, then, a sense in which the war tout court is itself a cooperative act in which all the combatants participate. Given the argument for authority-based accountability, it might seem that we are left with absurd results. For example, a private manning a traffic-control checkpoint, who personally kills no one, might end up bearing full authority-based accountability for all the deaths that her side causes in that war. Surely this is implausible. The argument for authority-based accountability, however, does not have the absurd consequence that each soldier bears authority-based accountability for the entirety of what her side does. This is because protected reasons can vary in their strength. For example, soldiers in a squad on a reconnaissance mission have claims against one another in virtue of the shared activity in which they are participating. They also have claims against other soldiers in other squads which together compose the platoon of which they are a part, where that platoon in combination with other platoons compose a company, which together with other companies form a battalion, then a brigade, a division, and finally an army. And a soldier in an army has claims against other combatants in other branches of the military.

¹⁴ The authority here has important features in common with a Razian service-conception of authority: both yield protected reasons. See (Raz, 1979). But there are important differences. According to the service conception, soldiers are morally required to defer to their superiors only if by doing so they are more likely to do what’s right than they would be if they acted on their own judgment. The deference characteristic of agentive functions, on the other hand, are established by the decision to participate in a cooperative action. See chapters 1 and 2.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

145

But as we telescope outward in this way, the claims the combatants have against others will typically weaken, in that the protected reasons they yield will weaken.¹⁵ Recall that a protected reason to do ϕ comprises both a first-order reason to do ϕ, and a second order reason to exclude from deliberation certain reasons against doing ϕ. The constitutive first-order reasons obviously admit of degrees; but the exclusionary reasons do so as well. They admit of degrees by way of the size of the exclusionary zone the second-order reasons set. Depending on how we specify the scalarity of the reasons the authoritative claims yield, there is room for the view that a typical soldier bears little accountability for what the vast majority of the other combatants fighting in the war do—and yet is accountable, e.g., for what her platoon does on a mission in which she is participating. Recall the private manning the traffic-control checkpoint; it’s true that given the solution I presented here, she is not completely off the hook for what her side does in the war. But I take this to be a strength of the account. There is another way in which the argument for authority-based accountability avoids the consequence that each combatant bears full authority-based accountability for what every other combatant does. In a hierarchically organized command structure, subordinates have the function of following the orders issued by their superiors, but not vice versa. That is to say, the orders superior officers issue to their subordinates are protected, whereas subordinates lack the standing to issue such orders to their superiors. Given the account I have developed, superiors will be accountable for what their subordinates do in conformity with the orders they issue. This isn’t to say, though, that subordinates will bear no authority-based accountability for what their superiors do. According to the argument for authority-based accountability, a superior bears authority-based accountability for a wrong-making feature of a subordinate’s conduct: the wrongful purpose for which she acts. But if there is reason for the subordinate to think that the superior will furnish such a purpose, and the subordinate nonetheless chooses to continue functioning as such, then she too is accountable for the wrongful purpose her conduct serves. After all, the superior cannot furnish a purpose for a subordinate if the subordinate refuses to function as an executor in a division of agential labor. So, there is a straightforward sense in which the superior relies on the subordinate to furnish such a purpose; this might accordingly inculpate the subordinate.¹⁶ This is by no means to say that the subordinate is as accountable as her superior. Indeed, the subordinate might not be accountable all, if she

¹⁵ This depends on much more than how the military is organized; for example, a particular battalion in the army might happen to work more closely with a particular air force wing than they do with other battalions—the strength of the authoritative claims that the involved combatants have against one another will reflect that fact. For more analysis see (Bazargan, 2012). ¹⁶ See section 2.1.3, chapter 2, for further discussion.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

146

, ,  

lacks reasons for thinking that the furnished purpose is wrongful or if she has little choice but to follow orders. I address these mitigating conditions in section 5.2. The upshot, though, is that authority-based accountability will (as we might expect) accrue at the higher echelons of command. At the same time, though, soldiers who are working together as equals (say, on a mission ordered by a superior officer) will bear authority-based accountability for what they together do. In what follows, I consider an example of shared action in war, and the implications it has for authority-based accountability.

5.1.3 Example of a Cooperatively Committed Harm in War Consider the follow example. Suppose a squad of soldiers fighting in an unjust war volunteers to rescue a wounded POW. There is every reason for them to believe that the enemy will treat the POW scrupulously by tending to his wounds and by holding him in decent conditions until the war’s end (which the unjust side could hasten by ceasing their aggression). Nonetheless, the squad chooses to fight—and kill—in furtherance of releasing the POW now. Though rescuing a POW is not intrinsically unjust, it is morally wrong to kill in furtherance of doing so when a) the POW is held treated fairly, and b) the soldiers guarding the POW are fighting in furtherance of a just cause. The squad leader—call her “A”—orders two soldiers in the squad—B and C— to retrieve the wounded POW and carry him to safety. The squad leader also orders five other soldiers in the squad—D through H to cover B and C by laying down suppressive fire, thereby enabling them to carry the wounded POW to safety. Clearly, as far as they are concerned, A has authority over B and C that they carry the wounded POW. Likewise, as far as they are concerned, A has authority over D through H that they lay down suppressive fire. But the relations of authority among the eight members of this squad are more complicated than this. Each of B and C confers a protected reason upon each other that he is to do his part in carrying the POW. After all, each would have a legitimate, directed complaint against the other should he abandon the attempt to rescue the POW on a whim. Likewise, each of D through H confers protected reasons upon each other to do his part in laying down suppressive fire. Again, each would have a legitimate, directed complaint against any other should he cease firing on a whim. Neither B nor C has authority over D through H that they carry the wounded POW; after all, only B and C have been assigned that task in the established division of labor. Likewise, none of D through H have authority over B or C that he lay down suppressive fire since, again, neither has been assigned that task. However, each of B and C does indeed confer a protected reason upon D through H for each of them to lay down suppressive fire, just as each of them

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

147

confers upon B and C a protected reason to carry the POW to safety. These protected reasons can be adumbrated as follows: A confers a protected reason upon each of B and C for him to assist in carrying the POW. • A confers a protected reason upon each of D through H for him to lay down suppressive fire; B and C each confers a protected reason against each other for him to assist in carrying the POW. • Each of D through H confers a protected reason against one another for him to lay down suppressive fire. • Each of B and C confers a protected reason upon each of D through H for him to lay down suppressive fire. • Each of D through H confers a protected reason upon each of B and C for him to assist in carrying the POW. • Each of B through H confers a protected reason upon A for her to function as a commanding officer. These relationships can be visualized in the following way, where the vertical dotted arcs indicate a unilateral authority-relation, and the horizontal solid arcs indicate a bilateral authority relation (see Fig. 5.1). A

B

C

SUPPRESSIVE FIRE

D

E

F

G

H

CARRY WOUNDED

The grounds for the claims in 1 and 2 are obvious: the soldiers involved have established an agential division of labor in which the squad leader, A, counts as the deliberator and the rest of the soldiers count as executors. The grounds for claims 3 through 6 might seem less obvious. Indeed, it might seem that the rest of the soldiers cannot have authority over one another given that they rank equally in the chain of command. But soldiers agree, either explicitly or implicitly, to cooperate in carrying out the missions they have been assigned. This is not merely some contingent feature of soldiering. Superior officers, in effect, order soldiers to cooperate in carrying out the missions they have been assigned. The explicit or implicit agreements soldiers make to assist one another in furtherance of the mission to which they have been assigned derive from following those orders. Inasmuch, in any given mission each soldier takes herself to have a practical claim against any other soldier in that mission that she do her part. This is what grounds claims 3 through 6. The content of the claims, though, differ depending on the role

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

148

, ,  

the soldier plays in a mission. So, whereas B has a practical claim against C that she assists in carrying the POW, B does not have that claim against D. Instead, B has a practical claim against D that D lay down suppressive fire (mutatis mutandis for the claim that D has against B). A similar point can be made in terms of shared action. The five soldiers laying down suppressive fire are participating in one shared act, and the two carrying the wounded soldier are participating in another. But all the soldiers in the squad are engaged in the broader shared act of rescuing the wounded POW. That is, the narrower shared act of laying down suppressive fire, and the narrower shared act of carrying the wounded POW, are both nested within the broader cooperative act of rescuing the POW. This shared act is itself nested within the even broader shared act the members of which compose a platoon. As we move outward concentrically, generalizing the shared action along the way, we can characterize each soldier in the squad as part of a shared act encompassing the nation’s army, and eventually the armed forces as a whole. Recall, though, that the protected reason that any given squad member confers upon any other given soldier outside the squad, diminishes in strength as we move outward concentrically. The protected reason—and thus the purpose—that each of the two soldiers carrying the POW confers upon one other, and upon the solders laying down suppressive fire, are much stronger than the protected reason—and thus the purpose—that he confers upon, say, a naval technician working aboard an aircraft carrier, engaged in a different theater of the same war (mutatis mutandis for the claim that the naval technician has against the squad member). This, in turn, means that the soldier will bear little authority-based accountability for what the naval technician does, and vice versa. This example, suitably generalized, shows how authority-based accountability helps vindicate the revisionist turn in war ethics. Recall the challenge that revisionism faced. According to revisionism, combatants are morally liable to be targeted only if they are accountable for contributing substantially to an unjust cause, or to a just cause pursued unjustly. The problem is that the typical combatant does little to advance the aims of a war; indeed, the typical combatant never participates in combat. This suggests that many if not most combatants would be morally immune from intentional attack. Targeting them would be tantamount to targeting civilians. But if what I have argued is correct, a country’s armed forces comprise a vast division of agential labor in which combatants confer protected reasons upon one another that they do their part in furtherance of the war they are fighting. These combatants, then, bear at least some authoritybased accountability for what their comrades do, which suggests they can be morally liable to be targeted after all, even if such combatants do not contribute substantially to that war’s unjust aims. In what follows I consider two challenges to this account.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

149

5.1.4 Challenges So far, I have been arguing as if all wars have either only just aims or only unjust aims. But in many wars each side will have at least some unjust aims (even if they are subsidiary to the achievement of a larger, just aim). Call this a “morally heterogeneous war”.¹⁷ If combatants bear authority-based accountability for the unjust aims pursued by their comrades, then it seems the presence of any unjust aim is enough to make all the combatants on that side morally liable to be killed. But this is not so. If killing a combatant helps forestall the achievement of both unjust and just aims, then we bear some accountability for having indirectly contributed to forestalling the war’s just aims. If these just aims are significantly more numerous or morally important than the unjust aims, then killing the combatant can be impermissible. Still, one might argue that if the permissibility of killing a combatant depends on comparing the effects that killing the combatant will have on forestalling the just aims versus the unjust aims of the war in which she fights, then we are again left with an account in which the basis of permissible killing is highly contingent. The worry is that, on this account, discriminating between permissible and impermissible targets in morally heterogeneous wars is pragmatically impossible. But this worry is unfounded. On the account I have outlined, discriminating between permissible and impermissible targets is not pragmatically unrealistic. After all, one way to discriminate between permissible and impermissible targets in a morally heterogeneous war is to abandon or eliminate the war’s unjust aims. Consider the following example. Suppose our adversary is waging a war with both just and unjust aims. We are, in turn, resisting our adversary’s aims. Thus, we too are waging a morally heterogeneous war: resisting our adversary’s unjust aims is just, and resisting our adversary’s just aims is unjust. Because our adversary is pursuing both just and unjust aims, it is pragmatically very difficult for us to discriminate between enemy combatants who can and cannot be permissibly killed. But suppose that we cease resisting our adversary’s just aims—which is, presumably, what we morally should do. That is, we allow them to achieve their just aims. Since our adversary is now pursuing only unjust aims by means of war, we no longer have the problem of discriminating between enemy combatants who can and cannot be permissibly killed; neither we nor our adversary is waging a morally heterogeneous war. By ceasing our pursuit of the unjust aims, we affect the moral status of the war our adversary is waging, by shifting it from a morally heterogeneous war to a purely unjust war, thus dissolving the problem of discrimination.

¹⁷ See (Bazargan, 2013).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

150

, ,  

The upshot is that we do have a way to discriminate between permissible and impermissible targets in morally heterogeneous wars—namely, by ceasing pursuit of the war’s unjust aims. This comports with a pre-existing obligation: in general, we already have a very strong reason to eliminate or abandon the pursuit of unjust aims in morally heterogeneous wars. Now we have even greater reason: in waging a war with an unjust aim, we will have difficulty discriminating between permissible and impermissible targets. Still, one might argue that even the best wars will inevitably include some unjust subsidiary aims (i.e., aims that serve as a means to the achievement of larger aims). They crop up, despite the best efforts of military and civilian leaders. But if we are fighting in an overwhelmingly just war with only a few minor unjust aims, it is very unlikely that a substantial portion of our attacks in that war will forestall the enemy’s just aims. Hence, it is very unlikely that a substantial portion of the enemy’s attacks are made just by the fact that the war we are waging includes minor unjust aims. I now turn to a second challenge: authority-based accountability might overgeneralize by inculpating civilians, thereby violating the principle of civilian immunity in war. In a well-functioning democracy, citizens of a country indirectly authorize the armed forces, via civilian leaders, to undertake military operations. In such an arrangement, it might seem that citizens enjoy a kind of authority over the combatants in that combatants function at the behest of the citizens on whose behalf they fight at home. On this view, the armed forces act on protected reasons proffered by the civilian leadership, who in turn act on protected reason proffered by the citizenry which elects the leadership. Given such an arrangement, it would seem that citizens bear authority-based accountability for what the combatants do on their behalf, by virtue of furnishing a purpose for them. This would, implausibly, inculpate citizens in a war, making them potentially liable to necessary and proportionate attack in furtherance of just aims.¹⁸ The typical civilian citizen will cast a vote for a civilian leader, who also functions as the commander-in-chief of the armed forces. The commander-inchief has direct authority over members of the armed services—what she says, goes. In that division of agential labor, she is the ultimate deliberator. The commander-in-chief will, accordingly, bear authority-based accountability for what the armed services foreseeably do under her command. Suppose, though, that as a single civilian citizen, I complain (given the opportunity to do so) to the commander-in-chief by voicing my opposition to the military decisions she has made. It is unclear that this complaint confers a protected reason. By electing her, the citizenry divests itself of authority over its armed services, by

¹⁸ Jeff McMahan, raising this sort of worry, writes that if combatants bear complicitous liability, then civilians will as well, since “the combatant is not more a part of the collective than the noncombatant” (McMahan, 1994, p. 212).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

151

conferring it upon a third party—the commander-in-chief. It is for this reason that the commander-in-chief is under no legal obligation to do as the citizenry says or wants when it comes to military decision-making. The people of a country do not stand atop the hierarchical chain of command in which the armed forces consist. Rather, the people determine who stands atop that chain of command. This difference is crucial; determining who is the commander-in-chief does not give us authority over the commander-in-chief. The result is that we do not bear authority-based accountability for what the armed forces do, after all. We needn’t worry, then, that the account I have developed over-generalizes by inculpating civilian citizens in a democracy. I hope to have shown that authority-based accountability can provide warethicists operating within a revisionist account of the morality of war with the resources for showing that an individual soldier in a war can be accountable for more than what she causes or the difference she makes. The result is that even those soldiers who contribute little or nothing to an unjust aim of which she is a part can be morally liable to be targeted. This helps dissolve the pragmatic difficulty with which we began: that of determining who is and isn’t morally liable given the revisionist account of war.

5.2 Mitigating Conditions and Authority-Based Accountability in War I’ve argued that a combatant participating in an unjust war can end up liable for more than her individual contributions to that war. This is because she bears authority-based accountability for what her comrades do; the stronger the putative claim she has against them that they do their part, the greater the degree of authority-based accountability she bears. But very often in war typical soldiers are ignorant of or otherwise deceived about the missions in which their comrades are fighting, not to mention the ultimate aims of the war itself. In addition, combatants are often alienated from the aims of the war in which they are participating if not the aims of the mission in which they’re engaged. And combatants are often coerced into participating, either by being conscripted or by exigent financial circumstances. In what follows I address how all of these factors affect a soldier’s authority-based accountability in war.

5.2.1 Ignorance in War Soldiers agree to assist one another in furtherance of aims they might know little or nothing about. Recall from section 4.3 (chapter 4) that where Deliberator accepts Executor’s promise, Deliberator’s ignorance of what promise

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

152

, ,  

was made, who made the promise, or what the effects of the promise will be, can indeed undercut her authority-based accountability for what Executor does—but only if Deliberator’s ignorance is non-culpable. This rarely describes the kind of ignorance from which soldiers suffer, since almost all combatants are in a position to reasonably suspect that their government might be fighting in furtherance of unjust aims. There are, of course, exceptions. The ignorance from which child-soldiers suffer is wholly non-culpable, as is, perhaps, the ignorance of indoctrinated soldiers who come from closed societies in which the regime controls and encompasses the totality of the soldier’s social world. But most soldiers are in a position to reasonably suspect that they have accepted a promise to fight in furtherance of unscrupulous aims. Ironically, it is the very fact that combatants are so often deceived about the war’s aims that serves to render their ignorance culpable. Recall also from section 4.3 that your ignorance of the promise you make does not itself do much to affect whether the person to whom you made the promise bears authority-based accountability for what you do in accordance with it. Suppose a soldier—call her Jane—has agreed to participate in a war she believes to have aims that are quite different from what they actually are. The soldier is in fact acting in furtherance of unjust aims. Those in the know are still fully inculpated in what Jane does, in that they bear authority-based accountability for her actions. They determine the purpose of what Jane does even if she is wholly unaware of what she is doing or why she is doing it. After all, those in the know still count as deliberators even if Jane is left largely in the dark about what’s going on. The point, then, is that ignorant promise-making does little to affect authoritybased accountability. Consider the following case, though. Suppose a soldier—call her Kara—is participating in a just war. Her platoon is deployed on a just mission to neutralize terrorists believed to be hiding out among civilians in a village. The soldiers in the platoon have authority over each other that they do their part in furtherance of the mission that they have been tasked with accomplishing. Thus, each soldier will bear authority-based accountability for what the other soldiers do in foreseeable furtherance of that task. During the mission, one of the squads composing the platoon accomplishes their task by committing a war-crime; in the course of engaging the enemy, they target and kill innocent civilians in the village. Does Kara bear authority-based accountability for what these war criminals have done? Of course, the war criminals never agreed to commit the massacre. This alone, however, does not get Kara off the hook. Suppose the squad-members committed the war-crime as a means to fulfilling their mission. And suppose Kara suspected that they might do something like this. Alternatively, suppose members of the squad committed the war-crime even though it does not help fulfil the aims of their mission; instead, they commit the atrocity “on the side”, gratuitously. Again, suppose Kara suspected that they might do something like this. Regardless of the

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

153

officially stated policy prohibiting war-crimes, and regardless of whether Kara would prefer that they accomplish their task in some other way, she bears authority-based accountability for what they do. Why? In keeping with the interpersonal division of agential labor they’ve established, Kara is among those who provide a purpose behind the squad’s conduct. Even if she does not want members of the squad to commit a war-crime, she is evidently willing to accept such an event as a cost of fulfilling the mission. To see why this is so, suppose the surviving victims of the war-crime approach one of the soldiers who actually committed the atrocity and demands an explanation from him. In addition to whatever private reasons he might have had, he can licitly refer the victims to the other members of the squad whose practical reasons furnished part of the purpose of what he did, since he was engaged in a cooperative action. And that includes Kara who accordingly furnished the following purpose for the soldiers who committed the war-crime: that they eliminate the terrorist threat, even if it requires committing an atrocity or even if they commit an atrocity “on the side”. Such a purpose is altogether too morally permissive. She is thereby inculpated at the bar of authority-based accountability as a result of having furnishing that morally problematic purpose.¹⁹ That being said, Kara is inculpated in the manner of a negligent or reckless wrongdoer, because she did not take there to be reasons to commit the war-crime per se; instead, she took there to be reasons for fulfilling the mission even at the cost of a massacre. Hence, the degree of authority-based accountability she bears will be less than it would be if she intended that outcome. But what if Kara has no good reason to think that the squad in question will commit a war-crime? In this case, Kara is not party to an agreement in which the squad will fulfill the mission at the risk of committing a war-crime, since it is by hypothesis unreasonable for her to interpret the task that they take themselves to have in that way. Suppose again that the surviving victims approach one of the soldiers, demanding an explanation from him. In addition to whatever private reasons he took there to be in favor of his actions, he can licitly refer the victims to the other members of the squad whose practical reasons furnished part of the purpose of what he did, since he was engaged in a cooperative action. This includes Kara, whose practical reasons furnished part of the purpose of what they did. But by hypothesis she can ingenuously deny that she thought that the mission warranted risking a war-crime. The result is that, in this case, Kara bears no authority-based accountability for the war-crime since her ignorance is nonculpable. An upshot is that accountability for war-crimes is not limited to those who committed them or to those who commanded the perpetrators. Materially

¹⁹ For further discussion of cases of this sort, see sections 4.1 and 4.3, chapter 4.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

154

, ,  

innocent combatants who did not themselves commit or otherwise facilitate the war-crimes can share in accountability, provided that they could reasonably foresee that the perpetrators would commit such an act, and provided that the materially innocent combatants had a claim against the perpetrators yielding a protected reason for them to do their part in the war-effort. The degree of foreseeability, and strength of the claim, determines how much authority-based accountability a materially innocent combatant will bear for what the warcriminals do.

5.2.2 Alienation in War The endless tedium, the inscrutable bureaucracy, and the hellish trauma of war easily alienate combatants from the larger aims for which they are fighting. A combatant might not care whether the ultimate aims of the war in which she is participating are achieved—she might care only about getting paid, or earning the respect of her peers, or accomplishing her mission, or simply ending her tour of duty. A member of a squad rescuing a POW might care deeply about her immediate mission, while failing to care about the broader aims of her regiment, the army, or the country’s armed forces in general. How does this affect her authority-based accountability for what her comrades do—and the authoritybased accountability of her comrades for what she does? Suppose a particular combatant—call him “Kurt”—is a fuel specialist; he helps manage the process of refueling aircraft on the flight-line of an aircraft carrier. In addition to handling jet fuel, he is responsible for operating the vehicles, equipment, and storage facilities critical for refueling operations. Kurt’s aircraft carrier is currently engaged in combat operations aimed at neutralizing an enemy forward-operating base. He is tasked with assisting the flight sorties by inspecting and refueling the aircraft. But Kurt does not particularly care whether the mission succeeds in destroying the enemy base, or whether his efforts end up successfully contributing to that mission. He takes himself to have a protected reason to conduct inspections and refuel the aircraft, but he does not take himself to have a protected reason to aid in the destruction of the enemy base, even though refueling the aircraft does just that. So, though his superiors take themselves to have a claim against him that he assists in destroying the enemy base, he only takes them to have a claim against him that he refuels the aircraft. Recall that on the argument for authority-based accountability, those who have authority over you that you do a wrongful act are accountable for a wrong-making feature of your action—viz., the purpose for which it is done. If Kurt denies that his superiors have a claim over him that he assist in destroying the enemy base—that is, if Kurt denies that he has a protected reason to assist in destroying the base—then on what grounds do we defer to them in evaluating what Kurt has done?

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

155

The answer is this. Kurt and his commanding officer take him to have a protected reason to inspect and refuel the aircraft. They agree on that point. His commander is therefore in a position to furnish a purpose for Kurt’s conduct, without Kurt’s consent. So, the fact that Kurt doesn’t care whether the mission succeeds does not vitiate his commanding officer’s power to furnish for him the purpose of assisting in the destruction of the enemy base. Kurt, in effect, gave her that power by consenting to her authority. The result is that his commanding officer (among others) is accountable for a wrong-making feature of Kurt’s conduct, insofar as she has furnished for Kurt the purpose of assisting in the base’s destruction. Thus, Kurt’s attitudes toward the larger mission does not affect whether his superiors bear authority-based accountability for what he does. What if, though, Kurt is not fully committed to his actual duties? Suppose he is disposed to surreptitiously skip elements of the pre-flight inspections he is tasked with completing if his workload becomes burdensome. In this case, Kurt is what I called (in section 4.1.1, chapter 4) “weakly alienated” from the task assigned to him. Recall that you are weakly alienated from a protected reason when you fail to adopt the dispositions that the protected reason requires. The conditions under which Kurt is disposed to perform do not include all the conditions under which he has an obligation to perform those tasks. Does this undermine Kurt’s status as an executor in a division of agential labor? If so, then it short-circuits his commanding officer’s authority-based accountability for what Kurt does; she would be freed of accountability for what Kurt does precisely because Kurt is violating his duties. This would certainly be an odd result, to say the least. But Kurt’s weak alienation does not undermine his role as an executor in a division of agential labor. This is because what grounds his role is the protected status of the reason he has; the reasons retain that status regardless of whether he adopts the requisite dispositions. Because Kurt’s weak alienation does not affect the protected status of the reason he has, it does not affect his role as an executor. If Kurt ends up contributing to the destruction of the enemy base, his commanding officer will bear authority-based accountability for that contribution; if Kurt hindered that outcome as a result of failing to properly conduct the pre-flight inspections, and if it was foreseeable that Kurt might act in such a way, then his commanding officer will bear authority-based accountability for that too. The upshot is that Kurt’s weak alienation does not undercut the purpose the commanding officer furnishes—the result is that she retains authority-based accountability. What if Kurt denies that his commanding officer has any claim over him? Suppose he still does his job because he wants to get paid. But he denies that anyone has the practical authority to demand that he do so. In such a case, Kurt is strongly alienated from his role as an executor. It might seem that his commanding officer’s motivating reasons do not constitutively determine Kurt’s purpose,

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

156

, ,  

since Kurt does not construe himself as having the function of enacting his commanding officer’s motivating reasons. But (as discussed in section 5.1), Kurt put himself under the authority of his superior officers by agreeing to participate as a member of the armed services. An insincerely made promise does not vitiate the normative force of that promise. Kurt cannot ingenuously deny that he has a protected reason to do as he ostensibly promised to do. But what if Kurt invokes the unjustness of the war as a basis for denying that his commanding officer has the authority to compel him to do as he agreed? Recall from section 3.1.4 (chapter 3) that even immoral promises can be binding in the sense that they confer practical reasons. As a result, Kurt cannot ingenuously deny that his commanding officer continues to have authority over him. If he chooses to conduct the preflight inspection and to refuel the aircraft, that warrants evaluating his conduct by reference to the purpose with which he is furnished. The upshot is that Kurt’s strong alienation from his task as an executor in a division of agential labor does not vitiate the authority-based accountability that his commanding officer bears for what Kurt does. Suppose, though, that Kurt’s commanding officer—call her “Victoria”—is alienated from the tasks she assigns qua deliberator. In particular, suppose she does not care whether Kurt follows her order to complete the pre-flight inspection for the aircraft participating in the sortie against the enemy base. In this case, Victoria is weakly alienated from the order she issues (see section 4.1.2, chapter 4). Her indifference as to whether Kurt follows her orders does not vitiate their respective roles in the division of agential labor. Victoria is still the deliberator and Kurt is still the executor. But Victoria’s attitude can affect the purpose she furnishes for Kurt’s conduct, which, in turn, can affect her authority-based accountability. Suppose Victoria believes that inclement weather, which her superiors have discounted, will certainly force them to abort the attack. She nonetheless orders the pre-flight inspection because it is expected of her. Because she believes the sortie will be aborted, she does not care whether Kurt actually conducts the inspection. Alternatively, suppose she believes that the attack will succeed, but that the pre-flight inspection is itself useless. Again, she orders the inspection anyway because it is expected of her, though she does not care whether Kurt actually conducts it. In both cases, Victoria regards the orders she issues as wholly pro forma. But this is not because she wants the mission to fail. She takes there to be a strong reason to act in ways that assist in the mission’s success. The result is that, even though Victoria is weakly alienated from the orders she issues to Kurt, she bears undiminished authority-based accountability for Kurt’s conduct should he actually end up contributing to the mission’s success. Alternatively, suppose Victoria is weakly alienated from the orders she issues in that she does not want Kurt to contribute to what she regards as a morally unjustifiable mission. We can imagine that the war has already been won—an

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

157

attack at this point on the enemy forward-operating base would be gratuitous, in that it would result in a massive loss of life without any discernable strategic benefit. Victoria nonetheless orders Kurt to conduct the pre-flight inspection since that is what she is expected to do. She does not care whether Kurt actually follows the order; this is because she does not want the mission to succeed. Indeed, she hopes that as a result of neglecting the inspection, the aircraft will be forced to return to the carrier due to some technical anomaly. But she is still willing, for her own benefit, to order Kurt to conduct the inspection since doing otherwise (we can suppose) jeopardizes her chances of promotion. Thus, as far as Victoria is concerned, the purpose of Kurt’s conduct is not to help destroy the enemy base, but to act in a way that does not jeopardize her career. This means Victoria is willing to furnish for Kurt a purpose that risks contributing to the unjust destruction of the base. This is not quite as bad as furnishing for Kurt a purpose aiming at the unjust destruction of the base. Her authority-based accountability varies accordingly, depending on whether, and if so, why she wants the base destroyed. In these cases, where Victoria is weakly alienated from the orders she issues, she still believes that she has the authority to issue those orders even though she doesn’t care whether Kurt follows them. She still believes that, as a result of issuing the orders, Kurt comes to have a protected reason to act accordingly, since she takes herself to be a deliberator and Kurt to be an executor in a division of agential labor. But suppose Victoria denies even that much. Imagine she believes that because the war they are waging is unjust, no one actually has any authority to issue orders over anyone else, other than the order to stand down, to retreat, or to otherwise quit the war-effort. She nonetheless continues to ostensibly issue orders because she does not want to get into trouble. She does this despite hoping that Kurt fails to conduct the inspection. In this case, Victoria is strongly alienated from the orders she issues to Kurt (see section 4.1.2), in that she denies, within the confines of her own thoughts, that her orders confer a protected reason to act accordingly. Put differently, she denies, in the confines of her own thoughts, that she qualifies as a deliberator and that Kurt qualifies as an executor in a division of agential labor. Victoria’s personal and private attitudes, however, do not undercut her authority over Kurt. Victoria retains authority over Kurt regardless of her attitudes because that authority is grounded in a public practice: she ostensibly accepted the role of deliberator and Kurt ostensibly accepted the role of executor. They thus retain those roles quite aside from whether they think they do. Public rules rather than private attitudes ground the authority characterizing the chain of command in a military institution.²⁰ And because the normativity that this authority confers

²⁰ See (Benbaji & Statman, 2019, p. 126). That being said, I disagree with their reasons for thinking that the moral equality of combatants is sound.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

158

, ,  

is practical,²¹ Victoria retains her authority over Kurt despite her correct belief that continuing the war is immoral. The truth of this belief does nothing to vitiate the practical authority she enjoys over Kurt. Alienation, weak or strong, will rarely vitiate relations of authority in war. For that to occur, a soldier needs to divest herself, explicitly and publicly, of the authority she has over others and of the authority they have over her, by resigning or by refusing to participate. Though Victoria’s authority over Kurt remains intact, her strong alienation from the orders she issues affects her authority-based accountability. Recall that a deliberator’s motivating reasons determine the purpose she furnishes for the executor. Despite issuing orders for Kurt to conduct the pre-flight inspection, she hopes that he doesn’t act accordingly since she believes that waging the war is unjust. She just lacks the moral courage to put her career at risk in furtherance of that belief. Thus, the purpose Victoria furnishes for Kurt is not to contribute to the aim of destroying the enemy base, but to act in a way that does not jeopardize her career. Victoria is, then, willing to furnish for Kurt a purpose that risks contributing to the unjust destruction of the base. As before, this is not quite as bad as furnishing for Kurt a purpose aiming at the unjust destruction of the base. Again, this means that, qua executor, Victoria’s attitudes affect what she bears authoritybased accountability for, even though those attitudes do not undercut her status as a deliberator and Kurt’s status as an executor. The upshot is that a subordinate’s weak or strong alienation from the orders he follows does not affect the authority-based accountability of the superior who issued the order to him. But if the superior is weakly or strongly alienated from the orders she issues, that alienation can potentially affect the purpose that she furnishes for the subordinate. And this, in turn, can affect the superior’s authority-based accountability for what the subordinate does.

5.2.3 Coercion in War Many combatants are coerced into participating in war. They might be conscripts coerced outright on pain of fines, imprisonment, or death. Or they might be “coerced” by their circumstances; for example, financial difficulties might pressure them into enlisting. As discussed in section 4.2.1, coercion can mitigate accountability; the worse the coercive threat, the greater the excuse, all things being equal. Suppose a combatant—call her “Sara”—is coerced into participating in an unjust war. In particular, she is credibly threatened by other soldiers with death should she refuse service. She accedes to this threat; she has no independent desire to participate in the war. They demand that Sara promise to follow orders and to

²¹ See section 3.1.4, chapter 3, for further discussion.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

159

function as a member of the military. Sara ostensibly makes this promise. Clearly, the promise is not binding—it fails qua promise. Though this threat fails to elicit a binding promise, it nonetheless succeeds as a demand. Recall (from section 4.2.2, chapter 4) that a demand can establish a division of agential labor, much in the same way that requests can. Recall also (from section 3.2.1, chapter 3) that when I make a request of you, I thereby provide you with a protected reason to comply—albeit one that you can refuse without acting immorally or irrationally. Should you choose to comply with my request, you thereby undertake the function of enacting my motivating reasons which have the concomitant function of normatively guiding your conduct. In virtue of soliciting your assistance and in virtue of complying with my request, we establish a division of agential labor in which I count as the deliberator and you count as the executor. Sara is subjected to a demand rather than a request, of course. But a demand can function like a request. When a soldier demands of Sara that she follow his orders on pain of death, he insists that Sara treat that very demand as a protected reason to comply. What makes it a demand rather than a request in this case is that the officer will refuse to take “no” for an answer. Either way, if Sara complies, she thereby treats the soldier’s demand as a protected reason to act. In effect, the soldier coerces Sara into establishing a division of agential labor with him, in which he counts as the deliberator and Sara counts as the executor. The upshot is that the soldier will bear authority-based accountability for what Sara foreseeably does in the war, even though her actions are, by hypothesis, coerced. Moreover, the soldier will bear authority-based accountability for what Sara does even if she is not accountable for her own actions. Sara, then, is coerced into adopting the role of executor in a division of agential labor. Can she likewise be coerced into adopting the role of deliberator? Can she be coerced into adopting a position of authority in war? Recall from section 4.2.2 that forcing someone into the role of a deliberator is problematic, on the grounds that undertaking such a role just means exercising discretion over the individual for whom you are deciding—which includes the discretion of divesting one’s self of authority over that agent. So, it’s unclear that Sara can coherently be forced into the role of a deliberator. When acting qua deliberator, the decisions Sara ostensibly makes do not count as relevantly hers if it is prohibitively difficult for her to make the decision she wants to make. If the coercive threat she faces is sufficiently severe, the decisions she makes qua deliberator do not reflect her attitudes about what others should do. In such a case, Sara is not accountable for furnishing the purpose behind her comrades’ conduct. This, in turn, suggests that she bears no authority-based accountability for what they do. The result is that if Sara accepts an authoritative role only because she was coercively threatened into doing so, then, provided that the threat is sufficiently

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

160

, ,  

severe, she will bear no authority-based accountability, not only for what her immediate subordinates do, but for what their subordinates do as well. Such cases, though, will be comparatively rare, since conscripts rarely join on pain of death. Take, instead, a more typical case. Suppose the coercive threat to which Sara is subjected is not so severe as to make it prohibitively difficult to refuse. Instead, it is only severe enough to diminish—without eliminating—her accountability. In such cases, the coercive threat makes it significantly harder for Sara to do otherwise, but not so hard that it is wholly unreasonable to expect her to be able to do so. In such cases, the decisions she makes qua deliberator somewhat reflect her attitudes about what others should do. Sara is accordingly somewhat accountable for the purpose she thereby furnishes. The lesson here is that conscripts who are coerced into participating in an unjust war inculpate others in that those who participate freely bear authoritybased accountability for what the conscripts do. But the conscripts who are coerced into participating aren’t necessarily themselves fully inculpated in that they do not necessarily bear authority-based accountability for what their comrades do. The extent to which they do, depends on the severity of the threat imposed upon them and the wrongfulness of the acts for which they stand to be inculpated.

5.3 Conclusion Revisionists argue that, as a general moral matter, whether an individual is liable to defensive harm depends on whether she is contributing to (or failing to prevent) a wrongful harm. This applies to warfare as well. The fundamental conditions for the defensive liability of combatants are no different from the fundamental conditions for the defensive liability of ordinary civilians going about their lives. In this respect, war ethics and ethics in general, are “of a piece”. I believe that revisionism is fundamentally correct in this regard. The problem, though, with revisionism is not that the conditions for liability to defensive violence in ordinary life fail to generalize to warfare. Rather, the problem with revisionism is that it mischaracterizes the conditions for moral liability to defensive violence in ordinary life. Contrary to what revisionists tend to maintain, an agent can indeed be morally liable for a wrongful harm even if she didn’t contribute to it and couldn’t have prevented it. The argument for authoritybased accountability demonstrates as much. With respect to warfare, a combatant in a war can bear authority-based accountability for the wrongful harms her cohorts commit in furtherance of a cooperatively committed harm of which she is a part. This account helps explain why combatants who contribute negligibly to the unjust aim in an unjust war are still morally liable to be targeted in furtherance of a just cause. Absent authority-based accountability, revisionists are forced to

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

161

conclude that targeting such combatants is tantamount to targeting innocent civilians. Focusing on a particular example, I argued that the authority-based accountability a combatant bears will depend on a) the organizational characteristics of the cooperatively committed harm in which the combatant is participating, b) her specific role in that cooperatively committed harm, c) how that cooperatively committed harm is nested organizationally within the armed services as an institution. These factors determine the division of agential labor in which the combatant is participating, and with it, the authority-based accountability the combatant bears. That being said, ignorance, alienation, and coercion—factors prevalent in war—can mitigate a combatant’s authority-based accountability. In discussing the mitigatory effects of ignorance, I focused on war-crimes. I argued that those who committed war-crimes and those who commanded the perpetrators are not the only ones potentially accountable. Materially innocent combatants who did not themselves commit or otherwise facilitate the warcrimes, and who were not superior in rank to the perpetrators, can nonetheless bear authority-based accountability for what the perpetrators do. With respect to alienation, I argued that a combatant still bears authority-based accountability for the conduct of those who follow her orders, even if they a) do not care whether they achieve their objectives, or b) secretly deny that she has any such authority over them. In theory, though, a combatant who a) does not care whether her orders are followed, or b) secretly denies that she has authority over her subordinates, might bear less authority-based accountability for what her subordinates do, under the right conditions. Finally, in discussing coercion, I argued that those who freely participate in an unjust war bear authority-based accountability for what conscripts do who are coerced into participating. But the coerced conscripts do not necessarily themselves bear full authority-based accountability for what their comrades do, depending on the circumstances. In general, this is because the deliberator in a division of agential labor furnishes the purpose of the executor’s conduct; where an agent is coerced into adopting the role of a deliberator, the purpose she furnishes isn’t her own, but that of the individual who is coercing her. I hope to have shown the various ways that the argument for authority-based accountability might shed light on the ethics of war. The armed forces of technologically advanced countries are immensely complicated institutions; that complexity is imported into how warfare is organized, administrated, and executed. I have here, in this chapter, only scratched the surface of how authority-based moral accountability might apply to accountability in the context of warfare.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

6 Accomplice Liability In jurisprudence, ‘complicity’ marks out a particular way that one person can be criminally liable to sanctions for the wrongful conduct of another.¹ On this view, complicity is a form of derivative liability. An agent is derivatively liable when her criminally liability is parasitic off someone else’s criminal liability. In such a case the derivatively liable agent, by culpably contributing to someone else’s wrongful conduct, shares in the wrongdoer’s liability. After describing in section 6.1 this approach to complicity in the law, I will argue in section 6.2 that much of the motivation for presenting complicity as a separate basis of criminal liability is misplaced. Paradigmatic cases of complicity can be assimilated into standard causation-based accounts of criminal liability. This view also has the advantage of accommodating the possibility of reckless complicity. But unlike others who favor this approach to complicity,² I will argue that there is still room for genuine complicity as a separate basis for liability, in the law and in morality. I argue in section 6.3 that on both the orthodox account of complicity and the revised account which dispenses with derivative liability, an agent is complicit in the wrongdoing of another only if she causally contributes to that wrongdoing or causally influences the wrongdoer. The problem, though, is that such accounts have difficulty explaining what grounds complicity in cases where an agent participates in a wrongdoing without causally contributing to that wrongdoing. Imagine this case, for example: Favored Son Several criminals are organizing a bank robbery. A mafioso’s son, Blunderbuss, wants to be involved, not because he wants a portion of the loot (which he’ll decline) but because he wants to burnish his credentials as a criminal. If the others do not include him, his father will be incensed. But they know that Blunderbuss has a penchant for mucking up delicate operations. So, they assign him a peripheral and nearly useless task—he is to serve as a lookout in a blind alley. The robbery succeeds, though Blunderbuss contributes nothing materially to it. Presumably, Blunderbuss serves as an accomplice. But both the accounts of complicity on offer have difficulty explaining why. In light of this, I will argue ¹ Parts of this chapter come from (Bazargan-Forward, 2017). ² See for example (Moore, 2007).

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0007

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

163

that complicity is best understood as a form of vicarious criminal liability, which I will ground by reference to the argument for authority-based accountability.

6.1 Standard Account of Accomplice Liability In Anglo-American criminal law an agent standardly becomes liable for a crime by committing it herself. Doing so requires both performing the prohibited act (the actus reus) and harboring a culpable state of mind (the mens rea). However, the doctrine of complicity (also known as the law of “aiding and abetting”, or “accessorial liability”) states that an agent (known as the “secondary actor”, the “accomplice”, or the “accessory”) can be liable for the crimes of someone else: the individual (known as the “primary actor”, or the “principal”) who actually committed the crime. In such cases the secondary becomes complicit by intentionally aiding or encouraging the primary to perform the prohibited act.³ Complicity, then, is not a crime in its own right. One cannot be guilty of complicity (unlike the crime of conspiracy). Rather, complicity is a way to become criminally liable for a crime committed by another. When the complicity of the secondary is proved, she is treated as if she had fulfilled the actus reus and mens rea conditions of the crime itself (unless she is an accessory after the fact). In that respect she is treated like the primary. Under the law, however, the secondary’s liability is derivative rather than vicarious vis-à-vis the primary. In cases of vicarious liability, the defendant has committed no wrong; she is still (vicariously) liable for what the primary does in virtue of the defendant’s formal relationship with the primary (such as the relationship between a parent and child or between a commanding officer and her subordinates). This relationship permits attributing to the defendant liability for the wrongful actions the primary commits. When, alternatively, an agent is derivatively liable, her liability is parasitic off the primary’s liability in virtue of her own intentional actions—specifically actions aimed at contributing to the primary’s wrongful conduct. It is in virtue of so culpably contributing that she shares in the primary’s criminal liability. I will argue that complicity, to the extent that it functions as an independent doctrine, is better understood as a form of vicarious liability, which I will ground in the argument for authority-based accountability. Under the law, for an agent to be complicit in the criminal wrongdoing of another, the secondary must not only contribute to the wrongful conduct of the ³ In the UK, this doctrine is stated in the “Accessories and Abettors Act of 1861” (amended by the Criminal Law Act of 1977): ‘Whosoever shall aid, abet, counsel, or procure the commission of any indictable offence . . . shall be liable to be tried, indicted and punished as a principal off ender’. In the US the doctrine is stated in the Model Penal Code §2.06, and in the federal aiding and abetting statute: ‘[w]hoever commits an offense against the United States or aids, abets, counsels, commands, induces or procures its commission, is punishable as a principal’ 18 U.S.C. § 2(a) (1982).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

164

, ,  

primary but do so with some mens rea directed toward the primary’s crime. It is generally acknowledged that accomplices need not have an intention with the same content as the primary. That would rule out the possibility of complicity altogether since a secondary intentionally performs acts of assistance or encouragement, not the prohibited act itself. But there is no further consensus over what the requisite mental state is. Some have argued that the secondary must provide aid with the purpose that the crime in question succeed.⁴ This is despite the fact that many crimes do not require the mental state of purpose on the part of the primary. Others argue that mere knowledge of the fact that acting will enable the primary to commit the crime in question suffices for satisfying the mens rea.⁵ Still others abandon the notion that there is a single mens rea that the secondary must possess, in favor of the more flexible view that the requisite mens rea for complicity varies with the mens rea required for the crime in question.⁶ It might seem useful to repair to the literature on joint action since it tends to meticulously detail the mental states that cooperators must have in order to qualify as joint actors. Since cooperative action in furtherance of wrongdoing seems to imply moral complicity, the mental states partly definitive of joint action might help resolve the requisite mens rea for criminal complicity in the law. But this strategy cannot ground the existing law of complicity. This is because in the law, the secondary can be complicitously liable for assistance in furtherance of the primary’s criminal act, even if the primary was unaware that she had been aided, as in this case: Guardian Devil J witnesses, without being noticed, a bank robbery in progress. J sees a patrol car approaching the scene. She wants the robbery to succeed (since she has a grudge against the bank). So, J pretends to require assistance, thereby preventing the police from discovering the robbery. The robbery consequently succeeds. Under the doctrine of liability in Anglo-American law, the conditions for complicitous liability are satisfied so long as influence succeeds as intended in contributing to the decision of the principal to commit the crime. Accordingly, J is complicit in the robbers’ wrongdoing in that she can be charged as an accessory. But even on the most minimalist account of joint action⁷ the robber does not count as acting jointly with J since the robber has no intentions directed toward an action by a group that includes himself and J. Complicity in the law overflows joint action, in that it is possible to be complicit for a crime without partaking in a jointly intentional conduct. We cannot, then, simply turn to the literature on joint ⁴ Judge Learned Hand in United States v. Peoni, 100 F.2d 401, 402 (2d Cir. 1938). ⁵ See (Weiss, 2002, pp. 1396–1409). ⁶ See (Weiss, 2002, pp. 1410–1414). ⁷ See for example (Ludwig, 2007).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

165

action in an effort to ground criminal complicity. I will ultimately argue that J should indeed be held liable for the robbery, but not at the bar of complicity. In what follows I focus on an issue that will inform the remainder of the discussion. If satisfying the actus reus of complicity requires a causal contribution to the wrong in question, what need is there for an account of complicity at all?

6.2 Superfluity of Complicity In what follows I will argue that we can largely dispense with complicity as a form of derivative liability. On this view, an “accomplice” is criminally liable simply for having culpably served as a cause of what the “primary” wrongfully does. Accounts that reject the possibility of interpersonal causation yield unintuitive consequences—or so I will argue. A skeptic might maintain that any attempt to dispense with complicity as a form of derivative liability will leave us unable to properly account for “non-proxyable” wrongs—i.e., wrongs that we cannot commit by causing someone else to commit them. But I will argue that this worry is misplaced. I will then argue that an account of complicity that dispenses with derivate liability can also accommodate the possibility of reckless complicity. Suppose you want to commit a burglary but lack the appropriate tools. You ask to borrow mine. I strongly suspect that you might use them to commit a robbery. I let you borrow them anyway. If we analyze criminal complicity in terms of culpably risking wrongdoing rather than culpably intending to aid wrongdoing, then you end up “recklessly” complicit in the burglary. I take this to be a welcomed consequence of rejecting the orthodox account of criminal complicity.

6.2.1 ‘Complicity’ as Non-Derivative Liability In standard cases of legal liability where the wrongdoer acts on her own and without assistance from others, she is legally liable in virtue of culpably causing the wrongful event in question (liability for omissions and inchoate crimes notwithstanding). In such cases, the liability is “direct” or “non-derivative” in that we need only appeal to the fact that the wrongdoer has culpably caused the wrongful event in order to explain her liability. But it is alleged—famously by Sanford Kadish,⁸ following Hart and Honoré⁹—that we cannot appeal to direct liability in cases where one party contributes to a wrong via another’s voluntary acts. Where the primary’s actions are fully voluntary, the secondary cannot be characterized as

⁸ (Kadish S., 1985).

⁹ (Hart & Honoré, 1958).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

166

, ,  

having caused the primary’s actions. On this agent-causal view of human agency, the voluntary actions of agents are literally uncaused events. Accordingly, the secondary cannot be held liable for the crime on the basis that she caused it. The role of complicity, then, is to provide separate grounds for liability in such cases. A doctrine ruling out interpersonal causation seems to render otiose contributory action—a requisite for the actus reus in complicity. But Kadish, again following Hart and Honoré, distinguishes between causation properly construed and mere causal influence. A “causal influence” simply raises the probability of an event’s occurrence, whereas a “cause” necessitates that event’s occurrence. Thus, Kadish writes that voluntary human actions can only be causally influenced, rather than outright caused: “Since an individual could always have chosen to act without the influence, it is always possible that he might have”.¹⁰ This disjunctive structure of liability—derivative versus direct—is supposedly necessary, then, to accommodate the liability of aiders and abettors given an agent-causal view of autonomous agency. But agent-causation is an untenable view of metaphysics, at odds with a naturalistic view of the world. It claims that the will when operating freely does so in a realm distinct from that of ordinary natural events and laws, thereby insulating autonomous agents from causal effects. It is hard to square this with a view that identifies willing with mental states that supervene on the neurological structure of our brains. While it might be true that we ineluctably tend to regard the voluntary actions of others as uncaused events, any account of legal and moral accountability ought to be based not on our debunked pretheoretic intuitions about metaphysics, but on the most plausibly developed account of how things actually are.¹¹ Once we jettison the metaphysical baggage of agent-causation, we are in a position to see that the criminal law is not as dependent on a distinct doctrine of complicity as it might otherwise seem. Consider the following pair of examples. Uncertain Murder 1 I want my innocent enemy dead. I have a friend—you—who might aid me in my goal. I ask you to kill him. I calculate that there is a 60% chance that you’ll do so. You subsequently commit the murder. Uncertain Murder 2 I want my innocent enemy dead. I construct a machine which, if it functions correctly, will kill my enemy. The machine is not perfectly reliable though; it works only 60% of the time. The machine nonetheless successfully kills my enemy.

¹⁰ (Kadish S., 1985, p. 360). ¹¹ For a compelling series of arguments in support of this view, see (Moore, 2000).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

167

The basis of criminal liability in the two versions of the example should be exactly the same. The fact that my contribution to the innocent’s death in Uncertain Murder 1 is mediated by an agent, whereas in Uncertain Murder 2 it is not, makes no difference to my liability. In both cases, the basis of my criminal liability is direct (or, alternatively put, non-derivative) in that I culpably act in a way that causes the innocent’s death. It just so happens that in the first case my causal relation to the death is mediated by an autonomous agent whereas in the second case it is not. The basis of an agent’s criminal—and moral—liability for a harm is the same regardless of whether she causes the harm by a) committing it, or b) aiding and abetting a voluntary accomplice. In both cases the individual is directly/nonderivatively liable in that it is her causal relation to the wrong (rather than merely to the agent who commits it) that grounds her liability. The notion of complicity does no work in underwriting her liability. One might raise the following objection. For some acts, what the agent brings about precludes certain kinds of mediating agency. That is, some acts cannot “go through” the agency of others. Sanford Kadish famously called these actions “nonproxyable.”¹² For example, Kadish says, “[a] defendant may cause a married person to marry another by falsely leading the married person to believe his prior marriage was legally terminated. But the defendant could hardly be held liable for the crime of bigamy, since one does not marry simply by causing another person to marry.”¹³ The deceiver, in this case, is not guilty of bigamy regardless of how integral the contribution was. The upshot is that it seems any attempt to subsume complicity under causation by recasting accomplices as non-derivatively liable for what the principal does must fail—at least for non-proxyable wrongs. Such an attempt would mean that the deceiver in Kadish’s example would, bizarrely, be guilty of bigamy. For this reason, Gardner concludes that “whoever acts through a principal must be an accomplice.”¹⁴ But as Michael Moore points out,¹⁵ we define bigamy in such a way that the only person who can be guilty of it is the person who gets married twice. That is, the crime of bigamy is necessarily first-personal. It is thus impossible for a third party to become guilty of a necessarily first-personal crime. The result is that even if we subsume complicity under causation, the deceiver would still not be guilty of bigamy since he is not the one who got married twice. This is all the more reason to ensure that we refrain from defining most crimes as non-proxyable. For example, battery, homicide, and rape, should all be defined so as to be proxyable, since the wrong, presumably, is not of the sort that consists in causing one’s own body to come into contact with another person’s body, to kill another person, or to sexually penetrate another person’s body. And, as Moore ¹² (Kadish S., 1985, pp. 372–385). ¹³ (Kadish S., 1985, p. 373). ¹⁴ (Gardner, 2007, p. 135). ¹⁵ (Moore, 2007, p. 419).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

168

, ,  

points out, for those crimes that are properly defined as non-proxyable, it is impossible for a third party to be liable, either as an accomplice or directly. So, for proxyable crimes, there is no need for accomplice liability since those who cause the prohibited wrong will count as principals. And for non-proxyable crimes, there is no need for liability of any kind for “accomplices” since it is definitionally impossible for them to be guilty of the crime in question. As Moore puts it, “the existence of nonproxyable crimes goes no distance in showing a need for accomplice liability.”¹⁶ Still, intuitively, the person who actually commits the wrong seems typically to have done something morally and legally worse than the person who intentionally facilitates that wrong. Suppose P1 slips an incapacitating drug in Victim’s drink with the intention that P2 subsequently rape her. As wrongful as P1’s actions are, presumably we want to say that what P2 does is even worse—legally and morally. Can we accommodate this on a view that subsumes complicity under causation? The answer is “yes”. And there are three arguments for this. The first argument is that, all things being equal, someone who imposes a greater risk of wrongful harm has acted in a more morally objectionable way than someone who imposes a lesser risk of wrongful harm. When P1 commits his act— i.e., when he slips the drug into Victim’s drink—the probability that this will result in Victim’s rape is less than when P2 commits actually commits the rape. (Kadish seems to note similarly that intervening agency is relevant to probability when he writes that “the intervention of a second actor, whose action is required for the harm to occur, reduces the probabilities that the harm would ultimately happen.”)¹⁷ If the probability that an act will result in a wrongdoing is relevant to the assessment of the wrongfulness of that act, then ceteris paribus P1 acts less wrongfully than P2. The second argument is that many moral wrongs are morally wrongful in more than one way. In particular, the wrong of rape includes, but is not limited to, the agent-relative wrong of being the one who commits the rape. In general, the moral injunction against causing wrongful harm applies more stringently to the individual who actually commits that harm than it does to individuals who merely

¹⁶ (Moore, 2007, p. 420). ¹⁷ (Kadish S. H., 1997, p. 381). This is not always so, however. Consider the following case. If P1 presses a button, he will get $25,000. But he knows that if he presses the button, it will reveal one of two buttons—one for P2, and the other for P3. If P2 presses his button, he will get $25,000; but as a sideeffect, pressing the button has a 95% chance of killing J. If P3 presses her button, she will get $25,000; but as a side-effect, pressing the button has a 5% chance of killing J. P1 knows something about the personalities of P2 and P3. P1 knows that because P2 is unscrupulous, there is a 95% chance that P2 will press her button should he get the opportunity. But P1 also knows that because P3 is less unscrupulous, there is only a 5% chance that she will press her button, given the opportunity. Should P1 press the button, there is a 90.25% chance that P3 will end up killing J, and a 0.0025% chance P2 will end up killing J. But if P2 presses the button, there is a 5% chance that she will end up killing J. So P1’s act is much riskier than P2’s act. Here is a case, then, where mediating agency does not diminish the probability that a harm will occur.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

169

enable it. This is because an individual’s agency is “wrapped up” – to a greater degree—in the harm she commits, relative to a harm she enables. We might put this by saying that there is an agent-relative reason not to be the one who commits a wrongful harm. (It’s not mere squeamishness that explains why members in a group of would-be murderers try to avoid being the one who actually pulls the trigger—it actually is morally worse to be the one who does the deed.) Since that reason is agent-relative, it doesn’t apply to third parties who “merely” enable the wrong in question. This is not to say, of course, that third parties do no wrong by contributing to the wrong of rape (far from it). Rather, the point is that one aspect of rape’s moral wrongfulness—an agent-relative aspect—constitutes a nonproxyable wrong. The third argument is that for particularly gruesome or heinous wrongs, such as rape, there is reason to think that the proximate wrongdoer is a morally worse person than an upstream contributor. This is because performing these heinous crimes while actually facing the victim requires vicious character traits which the proximate wrongdoer, by virtue of committing the act, reveals himself to possess, whereas we cannot safely attribute these traits to the upstream contributor. The upshot is this: even if there is a moral difference between intentionally committing a non-proxyable wrong and intentionally causing that wrong to be brought about, it is a mistake to think that the only way to accommodate this difference is by introducing derivative forms of liability.¹⁸ So again, in the sorts of cases described so far, the contributor’s causal relation to the wrong (rather than merely to the agent who commits it) grounds her moral—and with it, her criminal—liability. We do not need to invoke the notion of complicity as a form of derivative liability to ground a causal contributor’s liability.

6.2.2 Non-Derivative Liability and Reckless ‘Complicity’ A non-derivative account of complicity can also accommodate the possibility of reckless complicity. Suppose you undertake conduct that is wrong only in virtue of the risk that it will enable my wrongdoing. And suppose your conduct does indeed enable my wrongdoing. The wrong you committed counts as an instance of reckless complicity in that it wasn’t your aim to assist me in my wrongdoing. But you were in a position to recognize the risk that your actions would end up assisting me in my wrongdoing. Here are two examples.

¹⁸ For a far more thoroughgoing analysis of nonproxyability friendly to the sort of suggestion I’m making here, see (Moore, 2007), (Moore, 2009). But see (Yaffe, 2012).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

170

, ,  

Race P2 would like to participate in a reckless and illegal drag-race, but her car is in the shop. P1 agrees to allow P2 to borrow P1’s car. P1 is aware that P2 plans on using it to race. During the race, P2 strikes an uninvolved pedestrian, causing serious injuries to her. If it weren’t for P1’s assistance, P2 would not have been able to participate, and consequently the pedestrian would not have been injured. In this example P1 plays an indispensable role in causing the pedestrian’s injuries. But the injuries were committed by P2—not by P1. We can say that the harm which P1 brought about was “agentially mediated” by P2, and that P2 was a “mediating agent”. Here, P1 is recklessly complicit in the wrongful harm P2 commits. Reckless complicity can be described more generally as follows. P1 is recklessly complicit in a harm φ that P2 commits, with respect to act A1, act A2, for times t1 and t2, if: • P1 commits A1 at t1, which significantly increases the probability that P2 will voluntarily commit A2 at t2, which causes a harm φ. • P2 would not have committed A2, and φ would not have occurred, if P1 had not committed A1. • P1 knows (or is in a position to know) both of the above. The fact that P1’s act is a sine qua non of φ provides a prima facie moral reason for P1 to curb her conduct accordingly. And it provides a prima facie basis for regarding P1 as at least somewhat complicit in what P2 does (without thereby diminishing P2’s accountability for her own conduct). To avoid confounding variables, it is important to keep morally relevant features constant between P1 and P2. I will assume that though P1 intends A1 and P2 intends A2, neither intends φ. I will also assume that P1 is in a position to recognize that committing A1 imposes a risk of φ’s occurrence via P2’s agency; likewise, P2 is in a position to recognize that committing A2 imposes a risk of φ’s occurrence. A final preliminary point: there are at least two kinds of reckless complicity. P1 enables P2 by providing P2 with otherwise absent means of committing A2, which causes φ. But φ also counts as recklessly complicit if P1 motivates P2 by drawing her attention to motivating reasons to commit A2. I suspect that whether P2 is motivated or enabled is morally relevant to the evaluation of P1’s and P2’s conduct. But here I am concerned with reckless complicity simpliciter—I do not distinguish between enabling and motivating.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

171

Given this account of reckless complicity, does P2’s status as a mediating agent diminishes P1’s accountability for ϕ?¹⁹ A prominent pre-theoretic view is that when the upshot of one’s action “passes through” the agency of another, one’s accountability for the upshot is diminished or even eliminated—as if the other person were like a resistor in an electrical circuit. As Sanford Kadish put it, we have “an ingrained view of the world” according to which “the causal route through which an upshot occurs is a central feature in assessing blame.” Consequently, “recklessness with respect to a natural happening is not seen to be commensurable with recklessness with respect to another person acting in a certain way.” He confesses, though, that “[w]here this view comes from is a bit mysterious.”²⁰ The “Voluntary Intervention Principle” (VIP), versions of which are accepted in Anglo-American criminal and tort law, reflect that intuition insofar as the VIP states, in the words of Hart and Honoré, that “[t]he free, deliberate and informed intervention of a second person, not acting in concert with the first, and intending to bring about the harm which in fact occurs or recklessly courting it, is normally held to relieve the first actor of criminal responsibility.”²¹ Hart and Honoré attempt to defend this view by appealing to an agent-causal view of human action. Again, I will not comment on the (im)plausibility of this defense, as it has been sufficiently criticized by others.²² (One might point out that all things being equal what P2 does seems worse than what P1 does—but I addressed this point in the previous section.) If what I have argued is correct, many paradigmatic instances of complicity, as well as all cases of reckless complicity, will fall under the rubric of non-derivative/ direct liability. Suppose P2 wants to commit an armed robbery, but she lacks a firearm. P2 asks P1 for a weapon; she agrees to give him a quarter of the loot in compensation. P1 agrees. On the orthodox account of complicity, P1 is criminally liable for armed robbery even though she didn’t commit it; her liability in this case is derivative, where derivative liability is grounded in an attempt to enable the wrongdoer to commit a wrong. On the heterodox account I have presented, P1 is criminally liable for armed robbery even though she didn’t commit it—but her liability on this view is non-derivative. Like the protagonist in Uncertain Murder 1 and 2, she bears it wholly in virtue of the fact that she culpably acted in a way that risked causing an armed robbery. The fact that it was mediated by the agency of another does not itself change the basis of her liability.

¹⁹ Others who address this include: (Hart & Honoré, 1958), (Kadish S., 1985), (Zimmerman, 1985), (Hurd, 2001), (Moore, 2009). ²⁰ (Kadish S. H., 1997, p. 393). ²¹ (Hart & Honoré, 1958, pp. 42–43). Kadish similarly argues that in criminal law “voluntary actions cannot be said to be caused” (Kadish S., 1985, p. 371). ²² See especially (Moore, 2000).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

172

, ,  

If criminally liability in such cases is non-derivative, is there any room for complicity construed as a form of genuinely derivative liability? I will argue that there is, and that in such cases we should analyze complicity in terms of authoritybased accountability.

6.3 Complicity and Authority-Based Accountability Recall the example Favored Son in which Blunderbuss insisted on participating in a bank robbery. He ended up fulfilling a useless role in the scheme, as a result of which he contributed nothing to the bank robbery. Yet presumably he is still morally and legally complicit in the robbery. But both the orthodox account, which analyzes complicity as a form of derivative liability, and the revised account, which analyzes complicity as a form of non-directive (i.e., direct) liability, have difficulty explaining why Blunderbuss is complicit. The same goes for cases in which many agents together intentionally contribute to an overdetermined harm. Again, both the orthodox account and the revised account have difficulty explaining why each of the intentional contributors is complicit given that no single contributor’s contribution makes a morally relevant difference to the resulting harm. After describing this problem, I will argue that we can accommodate these cases by analyzing complicity as a form of vicarious liability, where you are vicariously liable (morally and thus potentially legally) for a wrongdoer’s actions simply if you bear the right sort of relationship to the wrongdoer. The requisite relation, I will argue, is one in which the complicit party serves as a deliberator in a division of agential labor and where the primary serves as an executor. Hence, on the view I defend, complicity reemerges as a form of authority-based accountability.

6.3.1 The Re-Emergence of Complicity On both the orthodox account of complicity and the revised account, which dispenses with derivative liability, an agent is complicit in the wrongdoing of another only if she causally contributes to that wrongdoing or causally influences the wrongdoer. The problem, though, is that it is possible to be complicit in a wrong without causally contributing to it. To see why, it’s necessary to look closer at what it means to causally contribute to an event. The two dominant theories of causation are regularity and counterfactual accounts. On regularity accounts, one event is a cause of another event when the first event is an insufficient but necessary element of a set of conditions

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

173

actually sufficient but not necessary for the occurrence of the second event.²³ On counterfactual accounts, the first event is a cause of the second if the two can be related to each other, either directly or by a chain of mediating events, such that if the first had not occurred, the second would not have occurred.²⁴ There are, of course, numerous problems plaguing both accounts. But the problem of preemptive overdetermination in particular reveals the role that complicity plays in morality.²⁵ Here is a typical example. Assassination Fund A villain wishes to assassinate a political figure meddlesome to local criminal elements. With the expressed purpose of doing so, the villain solicits financial donations from various criminals in furtherance of hiring a hitman. Hundreds of small donations pour in. The donations she receives are far more than what is necessary to hire a hitman, which she subsequently does. No one donation was necessary or sufficient for hiring the hitman. This sort of case is troublesome for any account grounding an accomplice’s moral or legal liability in her contribution to the wrongful act. Though each donor contributed to the fund, there were many contributions that were not part of a necessary set of funds sufficient for hiring the assassin. We have difficulty, then, explaining why those donors are liable for a murder when those donors did not cause that murder. It might be argued that such a donor slightly raised the antecedent probability of the murder’s occurrence by donating, but it is unclear why this should make her liable for the murder when it turns out that her donation made no actual difference to the murder’s occurrence. We can imagine her donation lying at the bottom of the barrel, unused—yet she is liable for murder. An appeal to the supposedly “derivative” character of complicity is of no help since we still need to explain why a contributor can be derivatively liable for a wrong that she did not enable the principal to commit. Some have attempted to circumvent this problem by foregoing causal accounts of non-derivative legal and moral liability altogether. For example, Christopher Kutz denies that a causal contribution is necessary for complicity; the “participatory intentions” of the accomplice are what ground liability.²⁶ Moore defends a singularist account of causation, which purports to solve problematic cases of overdetermination.²⁷ (He also argues that subjective chance-raising can be a determiner of culpability.) Sanford Kadish suggests that by “extending our wills” to the actions of others we come to be complicit in what they do.²⁸ Daniel Yeager,

²³ ²⁴ ²⁵ ²⁷

For variations of this account, see (Hart & Honoré, 1958), (Wright, 2001). The modern progenitor of such accounts is David Lewis. See (Lewis D., 1973). For a detailed analysis, see (Moore, 2009). ²⁶ (Kutz, 2000). (Moore, 2009, pp. 496–512). ²⁸ (Kadish S., 1985, p. 355).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

174

, ,  

taking a different approach, argues that there is a morally relevant distinction between helping and doing; complicitous individuals, by merely helping, incur a risk-based rather than a harm-based form of liability.²⁹ I believe Kutz and Kadish are on the right track. When we intentionally undertake a shared action, each participant can become morally and thus potentially legally liable for what other participants do, in virtue of the division of agential labor they establish. That is, when someone agrees to act at your behest in furtherance of some goal you specify, you can become liable for what that person foreseeably does in furtherance of that goal independent of whether you caused (or even causally influenced) your agent to so act. This is, of course, just the doctrine of authority-based accountability that I developed in Part I of this book. We have, then, a way to ground complicity in cases where the accomplice does not causally contribute to what the primary wrongdoer does. Recall that in Assassination Fund each donator contributes a small amount to the fund necessary for the villain to hire an assassin. Each donator, by donating, effectively “hires” the villain to procure an assassin. Inasmuch, the donor confers upon the villain a protected reason to hire an assassin, and thus a purpose for doing so. The result is that each donator bears authority-based accountability—and thus legal liability—for what the villain does. To better understand the nature of the agency-relation between the donors and the villain, it is helpful to strip the example of its inessential elements. Suppose the villain simply announces to the criminal underworld that if she receives enough donations, she will hire an assassin. That is the extent of the relationship between the villain and the donors. It seems, then, that the villain is not acting at the behest of the donors in that she does not seem to be acting as an executor in a division of agential labor where the donors count as the deliberators. Indeed, to the extent that we are inclined to think that there is a division of agential labor here, the villain seems to be the one making the decisions—not the donors. But this is illusory. If the villain presented her conditional intention to hire an assassin as a promise to the donors—a promise to hire an assassin should she receive enough funds—and, if the donors donated to the fund on the understanding that doing so is tantamount to accepting that promise, then the villain thereby establishes a division of agential labor with the donors. On this view, simply offering and accepting a promise results in a division of agential labor in which the promisee has authority over the promisor that she act in accordance with the terms of the mutually agreed-upon promise. (See section 3.1, chapter 3) This authority is asymmetrical: the promisee but not the promisor can at any time unilaterally free the promisor of the obligation the latter bears to the former.

²⁹ (Yeager, 1996).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

175

None of this is to say that a given donor is as accountable as the villain or as the assassin. This is partly because, recall, that the authority-based accountability is scalar in accordance with the strength of the protected reason it confers. Part of what determines the strength of a protected reason is the strength of the putative obligation it confers. Some agreements are stronger than others, where the strength determines among other things the stringency of the presumptive duty to follow through with the agreement. In some cases, the strength of the agreement is determined in part by its relative importance for the parties involved. So, for example, my promise to go shopping with you this weekend is likely to be less stringent than my promise to serve as godfather for your child. Another factor determining the stringency of an agreement is how the agreement is made. An agreement put in writing, assessed by a lawyer, and signed in the presence of a notary, will ceteris paribus be stronger than an agreement sealed by a wink and a nudge. We are now in a better position to evaluate the grounds for legal complicity in cases like Assassination Fund. At first, it might seem that the donors made no agreement with the villain. But in soliciting financial assistance the villain is understood as having agreed to use those funds in furtherance of hiring a hitman. Indeed, if he used it for completely different purposes the donors would have a claim against the villain precisely because the villain acted in a way tantamount to violating a promise. This shows we can interpret donating money under those circumstances as establishing an authority-relation between the villain and each donor. That the content of the agreement is largely implicit and informal might weaken the protected reason the relation confers, in which case a given donor might be only weakly inculpated at the bar of authority-based accountability. Then again, both parties might recognize that it is quite important for the villain to hire the assassin, as a result of which the donor might be more strongly inculpated at the bar of authority-based accountability. The devil, here, is in the details. In any case, the argument for authority-based accountability has the resources to ground any given donor’s criminal liability for what the villain does, even though no such donor’s conduct makes a difference to what the villain does.

6.3.2 Complicity as Vicarious Liability Legal scholars will likely balk at the notion of analyzing complicity in terms of authority-based accountability.³⁰ This is because authority-based accountability is ³⁰ “It is a long-standing principle in Anglo-American law,” Larry May points out, “that the state of mind of a given person does not transfer to another due to an authorization or any other basis of vicarious agency” (May, 1987, p. 99).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

176

, ,  

a form of vicarious liability. To be vicariously liable, you needn’t cause, nor try to cause, nor risk causing the offense in question. Rather, to be liable for the offense, it is enough if you bear the right kind of relation to the actual wrongdoer. There are various forms of vicarious agency—i.e., relations of the sort resulting in vicarious liability. I will briefly distinguish three kinds. In doing so, I will distinguish them according to their domains of application, rather than their justificatory bases.³¹ I will then respond to skeptics who deny that complicity can fall under the rubric of vicarious liability. In a proxyship, one party—the principal—authorizes another party—the proxy—to act on his behalf within a circumscribed domain of conduct. A proxy can be granted varying degrees of discretion to act on the principal’s behalf. At one extreme, the proxy is merely the principal’s interlocutor. In such a case, the proxy has no normative discretion over the content of what she says; she is instead solely a means by which the principal communicates her intentions to others. At the other extreme, a proxy is granted significant discretion to make substantive decisions the consequences of which, whether beneficial or detrimental, are binding on the principal. Trustees exemplify this kind of proxyship. A maximally authorized proxy is the Hobbesian sovereign, all of whose “actions and judgments” each citizen has antecedently authorized “in the same manner, as if they were his own.”³² However, regardless of how much discretion a proxy is granted, when she acts on the behalf of the principal it is not literally true that the proxy acted, even though the normative upshot is attributable to the principal. Rather, we have been permitted to treat the principal as if she has herself done what her proxy did. It is on these grounds that the principal bears vicarious liability for what her proxy does.³³ In a supervisorship, a “master”, such as an employer, is liable for tortious misconduct of her “servant”, such as an employee. Such misconduct most often takes the form of negligent wrongdoing, but can also include intentional wrongdoing committed without, or contrary to, the explicit orders of the employer. In such a case, compensatory liability is imposed on the employer even if she committed no negligence in hiring, supervising, or instructing the employee. So, if I run a food-delivery business, and one of my employees injures you in a car crash while he is on the job, I might be held liable to compensatory deprivations even though I didn’t injure you.³⁴ Supervisorships, though, are not limited to employer-employee relations. They can obtain in any hierarchical institutions, the paradigmatic examples of which are military organizations. A superior officer must answer for the failure of her subordinates, even if that failure is not her fault.

³¹ See also (Moore, 2007, pp. 56–61), who divides the space of vicarious liability differently. ³² (Hobbes, 2008, p. 266). ³³ See (Copp, David, 1979). ³⁴ I discuss such cases in the next chapter.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

177

In a suretyship, one party guarantees that she will perform an obligation belonging to another party. There are, then, three parties to the agreement. The principal is the party that agrees to undertake the obligation. The surety guarantees that the principal will act accordingly. And the obligee is the party who receives the benefit. For example, an employer can insure herself, via a bonding company, against the financial misconduct of a new employee, for a fee. If the employee commits embezzlement and escapes with the money, the employee bears the guilt. But compensatory liability falls on the surety, even though the surety is innocent of any wrongdoing. In this respect, the surety bears vicarious liability for the compensatory obligations that the wrongdoer incurs. Michael Moore writes that we should resist the “natural temptation” to assimilate accomplice liability to any of the above forms of vicarious liability. He writes: The often-expressed idea is that complicity [ . . . ] is an agency form of vicarious liability. For this form of liability, one need not have contributed in any way toward the occurrence of some legally prohibited result; one only need be a member of some group or combination, the other member(s) of which do cause some legally prohibited result to occur. On this view, an accomplice stands as the principal liable for the action of his ‘agents’, i.e., the perpetrator.³⁵

This is what happened in the infamous Pinkerton case. The brothers Walter and Daniel Pinkerton transported and sold bootlegged whiskey from a shared farm. Like the more famous bootlegger, Al Capone, the government charged the Pinkerton brothers with tax evasion rather than bootlegging. It was Walter, however, who committed the lion’s share of the substantive crimes, while Daniel was in prison. Both were charged, though, with tax evasion and one count of conspiracy to commit tax evasion. The majority in Pinkerton held that the agreement between the brothers to cooperate in furtherance of the crime created an agency-relationship between them, akin to a legal partnership. And thus, “so long as the partnership in crime continues, the partners act for each other in carrying it forward.”³⁶ But Moore argues that accomplice liability is not a sui generis form of vicarious liability; instead, if it is to serve as a form of vicarious liability, it must be identified with an existing kind, such as a proxyship, supervisorship, or suretyship, none of which ground criminal liability. Even if we think that an agreement to embark in criminal activity together, in the form of a conspiracy, establishes vicarious agency, the fact remains that accomplice liability applies in cases absent any agreement between the accomplice and the primary, as exemplified in Guardian Devil. Moreover, Moore points out, in many states mere agreement is not enough;

³⁵ (Moore, 2009, p. 283).

³⁶ 328 US at 646.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

178

, ,  

one has to actually assist in committing the crime to be an accomplice. “[I]n such states,” he writes, “the aiding required refuses to be reduced to mere group membership or general agreement.”³⁷ Accordingly, the Model Penal Code as well as most American jurisdictions, rejects the Pinkerton doctrine. A 1943 opinion following a Minnesota case demonstrates just how reluctant legal scholars are to apply the doctrine of vicarious responsibility—or, as it is known here, as the doctrine of respondeat superior—to criminal cases: Ordinarily the doctrine of respondeat superior has no application in criminal cases. Criminal liability, except for certain statutory offenses and others not here material, is based upon personal guilt. Responsibility for the crimes of others rests upon causation. A defendant is held criminally liable for having counseled, procured, commanded, incited, authorized, or encouraged another to commit a particular crime. At the very time the rule of respondeat superior was being developed as a basis for vicarious civil liability it was rejected as a ground for similar criminal liability.³⁸

In civil law, the employer or master needn’t be culpably involved in an offense to be liable. But this opinion makes clear that in criminal law, unless the master or employer is personal involved in the crime, the employer or master is not liable. But why think vicarious criminal liability is problematic? Moore suggests an answer. Vicarious liability does not actually extend or transfer guilt, Moore says. Rather, vicarious criminal liability amounts to treating an innocent person as if she were guilty—a practice, which might be justified in extreme cases, but which is generally verboten. If the primary, the surety, or the supervisor were literally guilty of what her agent does in virtue of the relation she establishes with her agent, then vicarious criminal liability would not be problematic. But, Moore says, “. . . it obviously cannot be literally true that the guilt transfers as well. For guilt to transfer literally, the action and intention too must transfer literally.”³⁹ But I’ve argued that something quite like this is indeed what happens when I have authority over you of the sort that confers upon you what we take to be a protected reason for you to ϕ. When we establish this division of agential labor, I partly determine the purpose of ϕ. Now, this isn’t the literal transfer of intentions. But by formulating particular intentions pertaining to ϕ, I thereby constitutively determine the purpose of what you do, which means I am accountable for a potentially wrong-making feature of what you do.

³⁷ (Moore, 2009, p. 283). ³⁸ State v. Burns, 215 Minn. 182, 187, 9 N.W. 2d 518, 520–1 (1943). For further discussion see (May, 1987, p. 99). ³⁹ (Moore, 2009, p. 61). See also the quote by H. D. Lewis at the outset of the introduction to this book.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

179

None of this is to say that Pinkerton was decided correctly. It is unclear that the brothers took Daniel to have a claim over Walter that he continue bootlegging while Daniel was in prison. And even if they did take there to be such a claim, the resulting authority-based accountability does not necessarily fully inculpate Daniel, in that it does not make Daniel as guilty as Walter for the bootlegging. The point, rather, is that we can make sense of the claim that Daniel is potentially inculpated as an accomplice by adverting to authority-based accountability. The moral here, then, is that we can indeed subsume legal complicity under the rubric of vicarious liability, provided that we analyze the type of vicarious liability in question not as a proxyship, a supervisorship, or a suretyship, but as a division of agential labor. The deliberator in this division of agential labor, by furnishing a morally problematic purpose for the executor, is morally and thus potentially legally accountable for a wrong-making feature of what the deliberator does. If the executor commits a crime qua executor, and if the deliberator is not directly liable for that crime, then the deliberator nonetheless qualifies as an accomplice, in virtue of her authority-based accountability for what the executor does.

6.4 Conclusion Accomplice liability in the law is typically characterized as a form of derivative liability, where an agent—a secondary—is derivatively liable if she has done something as a result of which she shares in the liability of another agent—the primary. On this account, a secondary is complicit in a crime just in case she has intentionally aided the primary. I’ve argued in this chapter that this orthodox account of complicity is superfluous in the law. An ordinary account of criminal liability based on culpably causing a crime can adequately ground the liability of accomplices insofar as they serve as a cause of wrongdoing. On this account, accomplices are directly liable, rather than derivatively liable for criminal wrongdoing. A result is that we can make sense of ‘reckless complicity’—a category of criminal wrongdoing in which an agent is liable for unintentionally but culpably contributing to a crime. This attempt to analyze complicity in terms of direct liability faces two challenges, both of which can be addressed—or so I argued. First, a skeptic might deny that the requisite form of interpersonal causation is possible. But skepticism toward the requisite form of interpersonal causation rests on a dubious metaphysics. Second, a skeptic might argue that analyzing complicity as a form of direct liability leaves us unable to make sense of so-called “non-proxyable” crimes. But, following Moore, I argued that for non-proxyable crimes, there is no need for liability of any kind for “accomplices” since it is definitionally impossible for them to be guilty of the crime in question.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

180

, ,  

Most putative cases of complicity, then, are actually straightforward instances of direct liability. Such an account, though, does not eliminate the category of legal complicity; there is a recalcitrant class of cases which direct liability cannot accommodate. These are cases of concurrent or preemptive overdetermination in which multiple individuals work together in furtherance of a cooperatively committed harm to which at least some of the individuals contribute only a small amount—an amount insufficiently substantial to ground direct liability. Here, I argued, is where complicity re-emerges, construed as vicarious liability. To be vicariously liable, it is enough if you bear the right kind of formal relationship to the actual wrongdoer. The requisite relationship, I argued, is the relationship that a deliberator bears to an executor in a division of agential labor. According to the argument for authority-based accountability, a deliberator can be accountable for furnishing a wrong-making feature of what the executor does qua executor: viz., the wrongful purpose for which the executor acts. As a result, a deliberator can bear authority-based accountability for what the executor does even if the deliberator does not causally contribute to what the executor does. In this way, we can accommodate legal complicity in cases where the accomplice fails to contribute in any meaningful way to the crime in question. Legal theorists have tended to roundly reject the possibility of vicarious criminal liability; I ended by briefly defending such a possibility. Authority-based accountability, then, helps us make sense of legal complicity by filling in a lacuna left by accounts analyzing complicity as direct liability. Still, no account of complicity in terms of vicarious liability will result in an account that can do the work that complicity does in Anglo-American criminal or tort law; this is because in the law an agent can be complicit in a wrongdoing by contributing to it without the wrongdoer’s knowledge. Consider again Guardian Devil, where J aids robbers unbeknownst to them. Since J assists without their knowledge, clearly there is no authority-relation between J and them. Yet in AngloAmerican criminal law, J is legally complicit in their wrongdoing. J can be charged as an accessory to robbery. But I argued that this is a straightforward case of direct liability—it is J’s causal relation to the wrongdoing (in combination with satisfying the requisite culpable mental states) that grounds her liability. So, although this sort of case will not be covered by an account of complicity analyzed in terms of vicarious liability, it does not need to do so. It is no limitation of that account that it does not impute complicity as such in cases like Guardian Devil. There are many issues pertaining to derivative and non-derivative complicity which need much more discussion, such as: whether we can bear non-derivative liability for the unsuccessful attempts of wrongdoers we enable; whether an unsuccessful attempt to enable a wrongdoer who nonetheless succeeds can make the attempted enabler liable; whether we should treat non-derivative liability for the harm we enable another to commit differently from the non-derivative liability of innocent agency (i.e., cases where we use unwitting or cognitively

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

181

impaired individuals in furtherance of wrongful ends), and so on. My goal here is more modest: to show that insofar as there is room for genuinely non-causal derivative liability—which seems necessary to implicate wrongdoers in cases of concurrent and preemptive overdetermination—we should take more seriously the possibility that vicarious agency in general and authority-based accountability specifically can serve as grounds for complicity in such cases.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

7 Respondeat Superior and Enterprise Liability Suppose you own a pizza delivery establishment. You hire your own drivers to deliver pizza. One night, one of your drivers delivers his pizza while intoxicated, as a result of which he drives his car into a pedestrian, severely injuring her. She consequently sues you for damages. Even though you did not know and had no reason to believe that your driver would deliver pizzas while drunk, you might be financially liable for the damages your driver caused, as a result of respondeat superior. The doctrine of respondeat superior is a claim about the relationship between a person in a position of formal authority—the “master”—and the person over whom she has authority—the “servant” (see the brief discussion of supervisorships in section 6.3.2, chapter 6). According to the doctrine, the master can bear vicarious legal liability for the damages the servant wrongfully causes when he is acting under the aegis of the master’s authority. In Anglo-American tort law, the doctrine holds employers financially liable for the tortious wrongs of their employees. In the US, the test for respondeat superior liability typically derives from section 219(1) of the Restatement (Second) of Agency: “A master is subject to liability for the torts of his servants committed while acting in the scope of their employment.”¹ To recover from the employer under this doctrine, the plaintiff must prove that the defendant and the tortfeasor are in a “master-servant” relationship and that the tortfeasor committed the tort qua employee. The master-servant relationship is based on the former’s “right to control the physical conduct” of the latter.² Traditionally, then, respondeat superior applies when the employee is acting within the scope of her employment. As I note in section 7.1, the problem with this formulation is that it tends to exclude respondeat superior from those cases in which the employee committed an intentional tort—most notably, sexual abuse. The courts have attempted to fill this gap by grounding respondeat superior in enterprise liability, according to which the owners of an enterprise should bear the

¹ (Restatement (Second) of Agency §219(1) (1958)). ² (Restatement (Second) of Agency §2(l) (1958)). For illustrative cases applying section 220, see Santiago v. Phoenix Newspapers, Inc., 794 P.2d 138 (Ariz- 1990) and Eden v. Spaulding, 359 N.W.2d 758 (Neb. 1984).

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0008

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

183

costs of addressing the foreseeable negative externalities of running that enterprise. It seems, then, that respondeat superior can be analyzed wholly in terms of enterprise liability. However, I argue in section 7.2 that respondeat superior applies in some cases which enterprise liability has difficulty accommodating. These include cases in which the employers do not help cause, even indirectly, the wrong that the employee commits. They also include cases in which the customers or clients are fully aware of the risks pertaining to doing business with the enterprise, yet reasonably choose to do so anyway. How, then, do we accommodate respondeat superior in such cases? I argue in section 7.3 that respondeat superior is grounded partly in special duties of care that the owners of an enterprise have toward their patrons. The owners discharge these duties of care by establishing a division of agential labor with their employees thereby conferring upon them those duties of care. The result is that should their employees violate duties of care toward the patrons, the owners ipso facto do so as well, which serves as grounds for compensatory liability. The upshot is that to properly analyze respondeat superior we need the argument for authority-based accountability.

7.1 Enterprise Liability as Grounds for Respondeat Superior According to the doctrine of respondeat superior as it is traditionally understood, the employer ought to bear the legal reparative burden of the foreseeable harms that the employee commits within the scope of employment.³ I will begin by considering what falls within the scope of employment. In what follows I will lend some initial support to the view that we ought to answer this question by repairing to the doctrine of enterprise liability. According to this view, those who benefit from the operation of the enterprise are those who ought to bear the costs of the negative externalities that the enterprise imposes. I will then argue that the best version of this view is one which denies that the employee must be acting with the intention of fulfilling her role as an employee at the time she is committing the wrong. It is enough if the employee misuses the opportunities that her employment affords. I will ultimately argue, though (in the next section) that such attempts to grounds respondeat superior in enterprise liability are foredoomed to failure. This is because respondeat superior is a “formal” doctrine in which one agent is regarded, for the purposes of the law, as another. Enterprise liability, on the other hand, is a doctrine grounded in causal responsibility. Because they operate in

³ See (Honoré, 1999, p. 81).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

184

, ,  

radically different ways, one cannot be grounded in the other. Yet attempting to do so remains popular and is thus a possibility worth exploring; doing so will help ultimately reveal the ways in which the argument for authority-based accountability can help in an analysis of respondeat superior.

7.1.1 The Scope of Employment On the traditional reading of respondeat superior, an employer is financially accountable for what her employee does within the scope of her employment. The doctrine is typically understood as a species of strict liability.⁴ But first: what counts as the “scope of employment”? It is tempting to say that conduct the employer authorized is what falls under the scope of the employment. But this is clearly too narrow. The owner of a courier service can be legally liable for the accident a drunken employee causes while driving even if the owner did not authorize the employee to drink. Accordingly, the American Law Institute as well as the courts in the UK and Canada have adopted a more expansive account. According to the Restatement (Second) of Agency, conduct falling within the scope of employment “must be of the same general nature as that authorized, or incidental to the conduct authorized.”⁵ In UK courts (until recently) the “Salmond Test” limited liability to conduct that the employer “has not authorized, provided they are so connected with acts which he has authorized that they may rightly be regarded as mode—although improper modes—of doing them.”⁶ The problem with these formulations is that it makes it very difficult to recover in cases where the tortfeasor commits an intentional wrong “which could be portrayed as the very antithesis of what the employee was employed to do and, therefore, unconnected with his employment.”⁷ Thus, in a UK Court of Appeal decision pertaining to a teacher accused of abuse, a judge stated that “in the field of

⁴ See, e.g., Konradi v United States, 919 F.2d 1207, 1210 (7th Cir. 1990) “The liability an employer for torts committed by its employees—without any fault on his part—when they are acting within the scope of their employment, the liability that the law calls “respondeat superior,” is a form of strict liability.” ⁵ (Restatement (Second) of Agency §229(1) (1958)). According to the Restatement, the conduct of a servant is within the scope of employment if and only if: a) b) c) d)

it is of the kind he is employed to perform; it occurs substantially within the authorized time and space limits; it is actuated, at part, by a purpose serve the master. if force is intentionally used by the servant against another, the use of force is not unexpectable by the master.

To determine whether conduct meets this condition, the Restatement (§229(2)) adumbrates a host of relevant criteria, which we need not consider here. ⁶ (Heuston & Buckley, 1996, p. 443). ⁷ (Brodie, 2007, p. 494).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

185

serious sexual misconduct, I find it difficult to visualize circumstances in which an act of the teacher can be an unauthorized mode of carrying out an authorized act.”⁸ In light of this difficulty, the Canadian Supreme Court⁹ modified the Salmond Test by allowing liability given a “close connection” between the nature of the employment and the employee’s tort. Here, the court stated that vicarious liability is justified given that “the employer puts in the community an enterprise which carries with it certain risks. When those risks materialize and cause injury to a member of the public despite the employer’s reasonable efforts, it is fair that the person or organization that creates the enterprise and hence the risk should bear the loss.”¹⁰ The UK accepted the Canadian modification of the Salmond test in a case concerning child abuse. In justifying vicarious liability, Lord Millett stated that “a person who employs others to advance his own economic interest should in fairness be placed under a corresponding liability for losses incurred in the course of the enterprise.”¹¹ In writing for a majority judgment in a different case, Lord Nicholls stated that carrying on a business enterprise necessarily “. . . involves the risk that others will be harmed by wrongful acts committed by the agents through whom the business is carried on. When those risks ripen into loss, it is just that the business should be accountable for compensating the person who has been wronged.”¹² Courts in the US—primarily in California—have made a similar move. The California Supreme Court adverted to enterprise liability as a justification for respondeat superior in the statement that “California cases have long recognized that the employer’s responsibility for the torts of his employee extends beyond his actual or possible control of the servant to injuries which are ‘risks of the enterprise.’ ”¹³ According to the doctrine of enterprise liability, the employer should “be held liable for those things which are fairly to be regarded as risks of his business” even if he is innocent of any wrongdoing. This reasoning is best articulated in Prosser’s tort law handbook: What has emerged as the modern justification for vicarious liability is a rule of policy, a deliberate allocation of a risk. The losses caused by the torts of employees, which as a practical matter are sure to occur in the conduct of the employer’s enterprise, are placed upon that enterprise itself, as a required cost of doing business. They are placed upon the employer because, having engaged in an enterprise, which will on the basis of all past experience involve harm to others ⁸ ⁹ ¹⁰ ¹² ¹³

In Trotman v North Yorkshire County Council, [1999] LGR 584. In Bazley v Curry (1999) DLR 174 (4th) 45. In Bazley v Curry (1999) DLR 174 (4th), at 60. ¹¹ In Lister v Hesley Hall at 243. In Dubai Aluminium v Salaam [2003] 2 AC 366 at 377. In Hinman v. Westinghouse Electric Co. at 990.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

186

, ,  

through the torts of employees, and sought to profit by it, it is just that he, rather than the innocent injured plaintiff, should bear them . . . .¹⁴

Thus, in 1947 Justice Traynor stated that “[t]he principal justification for the application of the doctrine of respondeat superior in any case is the fact that the employer may spread the risk through insurance and carry the cost thereof as part of his costs of doing business.”¹⁵ As a result of such precedents, Justice Mosk concluded in 1995 that “it must be deemed settled in California that in accordance with the principal justification for the doctrine, the employer’s liability extends to risks inherent in or created by the enterprise.”¹⁶ The doctrine of enterprise liability, then, is grounded in the fairness of allocating the outcomes of option luck. It is unfair to impose the costs resulting from a risk on those who did not undertake and did not stand to benefit from that risk when those costs can be shifted to those who willingly chose to undertake that risk, and who did so for their own benefit. As Gregory Keatings puts it, “Victims who are strangers to the enterprise derive no benefit from it; therefore, it is unfair to ask them to bear a substantial loss when that loss might be dispersed across those who participate in the enterprise and, therefore, do benefit from it.” And as far the riskimposers go, they have no complaint if we shift the legal reparative burden onto them since they imposed the risk “for their own advantage, fully expecting to reap the benefits that accrue from imposing those risks.”¹⁷ The issue, then, is whether a given harm is a result of a risk “inherent in or created by the enterprise.” One way to answer this question is by ascertaining whether the type of harm at issue is a foreseeable consequence of the activity from which it resulted.¹⁸ This foreseeability-test is not utterly alien to the Restatement. Indeed, it uses something quite like it in cases of employer liability for intentional torts that the employee commits within the scope of her employment. The doctrine of respondeat superior applies to unauthorized intentional torts (typically assault and battery) only if the “foreseeable use of force” is “not unexpectable in view of the duties of the servant.”¹⁹ This amounts to the “hindsight foreseeability” ¹⁴ (Prosser, Keeton, Dobbs, Keeton, & Owen, 1984, pp. 500–501). ¹⁵ In Johnston v. Long, 81 P.2d 645 (Cal. 1947). ¹⁶ Farmers Ins. Group v. County of Santa Clara, 906 P.2d 440, 463–464 (Cal. 1995) (Mosk, J., dissenting) (emphasis removed). ¹⁷ (Keatings, 2014, pp. 307–308). Enterprise liability specifically, and strict liability more broadly, is primarily concerned with what Keatings calls “commutative justice”: they prescribe that “those who inflict harm reasonably on others must repair the harm that they reasonably inflict in order to align burden and benefit fairly.” Jeremy Waldron argues for a fairness-based justification for enterprise liability beyond torts in (Waldron, 1995). ¹⁸ This is the strategy California has adopted. See the appellate case, Rodgers v. Kemper Construction Co 124 Cal. Rptr. 143 (Cal. Ct. App. 1975). ¹⁹ Restatement (Second) of Agency §245 (1958). The exemplar of this is the 1973 Minnesota Supreme Court ruling in Lange v. National Biscuit Co. 211 N.W.2d 783 (Minn. 1973). In this case, a salesman for the defendant—the National Biscuit Company—attacked the plaintiff—a grocery store manager—following a business-related argument between the two.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

187

test adopted in California for cases of employee negligence. For example, in LeBrane v. Lewis,²⁰ a manager stabbed an employee at work who had just been fired. Supreme Court of Louisiana found the employer liable on the following grounds: The dispute which erupted into violence was primarily employment rooted. The fight was reasonably incidental to the performance of the supervisor’s duties in connection with firing the recalcitrant employee and causing him to leave the place of employment. It occurred on the employment premises and during the hours of employment. In short, the tortious conduct of the supervisor was so closely connected in time, place, and causation to his employment-duties as to be regarded a risk of harm fairly attributable to the employer’s business, as compared with conduct motivated by purely personal considerations entirely extraneous to the employer’s interests. It can thus be regarded as within the scope of the supervisor’s employment, so that his employer is liable in tort to third persons injured thereby.²¹

This suggests that employee violence counts as “not unexpectable” when the violence is employment-related. The assault must occur “within work-related limits of time and place”²² not because this serves as an independent requirement, but rather because it serves as circumstantial evidence for the employment-related nature of the dispute. As others point out,²³ “[e]mployer liability should not turn on the fact that the employee waited until after working hours to exact revenge.” The upshot is that respondeat superior applies where the employee commits a tort within the scope of employment, where torts committed within the scope of employment are those resulting from a risk “inherent in or created by the enterprise,” which is in turn analyzed in terms of foreseeability.

7.1.2 Occasioning vs. Aiding a Tort As noted, California has largely adopted the enterprise liability model of respondeat superior. It is all the more surprising to learn, then, that the California Supreme Court has declined to hold employers liable in cases where: a middleschool teacher allegedly molested a student at the teacher’s home during an officially sanctioned extracurricular program,²⁴ a deputy-sheriff sexually harassed ²⁰ 292 So. 2d 216 (La. 1974). ²¹ Id. at 218 (footnote omitted). Texas adopted a similar approach in Texas & Pac. Ry. v. Hagenloh, 247 S.W.2d 236, 239–240 (Tex. 1952). ²² Lange v. National Biscuit Co., 211 N.W.2d 783, 786 (Minn. 1973). ²³ (Roszkowski & Roszkowski, 2005). ²⁴ John R. v. Oakland Unified School District 769 P.2d 948 (Cal. 1989).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

188

, ,  

other deputies working at a jail,²⁵ a medical technician performing an ultrasound sexually assaulted a patient,²⁶ a physician misused a donor’s eggs,²⁷ a scoutmaster committed sexual molestation,²⁸ a security guard committed sexual assault after a traffic stop,²⁹ and a counselor sexual assaulted a resident at a facility for the emotionally disturbed.³⁰ California’s decision to deny respondeat superior in such cases is perplexing given that the state analyzes employer liability in terms of whether the tort of that kind was foreseeable. In each of these cases, a tort of that kind is, unfortunately, eminently foreseeable. It is hardly unusual for people in positions of power and trust to abuse these privileges. Yet California has insulated employers against liability in such cases because the stated basis for analyzing respondeat superior in terms of foreseeability lies in the doctrine of enterprise liability: A risk arises out of the employment when ‘in the context of the particular enterprise an employee’s conduct is not so unusual or startling that it would seem unfair to include the loss resulting from it among other costs of the employer’s business’. In other words, where the question is one of vicarious liability, the inquiry should be whether the risk was one ‘that may fairly be regarded as typical of or broadly incidental’ to the enterprise undertaken by the employer.³¹

In the cases of sexual assault described above, employment merely occasions the tort. This is important because enterprise liability—and ipso facto respondeat superior—are grounded at least partly in considerations of fairness. It is only fair for the enterprise that created a harm to absorb the costs of that harm. But, in cases of sexual assault, the enterprise does not create a de novo risk—rather, it merely occasions a pre-existing risk. Peter Cane makes this point through a hypothetical example: [A]n engineering company would not be vicariously liable where, without more, an employee sexually assaulted another employee on the employer’s premises; carrying on an engineering business does not (as far as we know) materially increase the risk of assault on the employer’s premises.³²

²⁵ Farmers Ins. Group v. County of Santa Clara, 906 P.2d 440 (Cal. 1995). ²⁶ Lisa M. v. Henry Mayo Newhall Memorial Hosp., 907 P.2d 358 (Cal. 1995). ²⁷ Stone v. Regents of the Univ. of Cal.9,2 Cal. Rptr. 2d 94 (Cal. Ct. App. 1999). ²⁸ Juarez v. Boy Scouts of Am., Inc., 97 Cal. Rptr. 2d 12 (Cal. Ct. App. 2000). ²⁹ Maria D. v. Westec Residential Sec., Inc., 102 Cal. Rptr. 2d 326 (Cal. Ct, App. 2000). ³⁰ John Y., Jr. v. Chaparral Treatment Ctr., Inc., 124 Cal. Rptr. 2d 330 (Cal. Ct. App. 2002). ³¹ 814 P.2d 1341 (Cal. 1991). (quoting Perez v. Van Groninger & Sons, Inc., 719 P.2d 676, 679 (Cal. 1986)) (citations omitted). ³² (Cane, 2002).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

189

Thus, as the California Court of Appeals puts it: Our Supreme Court has stated that sexual assaults are not per se beyond the scope of every employment. However, as with other assaults, ‘a sexual tort will not be considered engendered by the employment unless its motivating emotions were fairly attributable to work-related events or conditions. If the injury is inflicted out of personal malice or compunction, not engendered by or connected to the employment, in other words if the tort is personal in nature, there is no vicarious liability.³³

That being said, the Restatement (Second) of Agency section 219(2) provides exceptions to the “scope of employment requirement.” Notably, respondeat superior does not apply to employers whose employees are acting outside the scope of their employment, unless the employee “was aided in accomplishing the tort by the existence of the agency relation.” Under this exception, the employer is vicariously liable because the employment provides the means by which the employer is able to do harm. The problem, though, is that at least in principle, “every case where vicarious liability is at issue, the agent will have been aided in some way in committing the tort by the position that he holds” if only due to the proximity and access to the victim.³⁴ Thus, courts have narrowed the principle so that respondeat superior applies only if the tort “was accomplished by an instrumentality, or through associated conduct with the agency status.”³⁵ The United States Supreme Court adopted this line of reasoning in fixing employer liability for cases in which supervisors sexually harass employees.³⁶ In Faragher v. City of Boca Raton, the Supreme Court held that: . . . it makes sense to hold an employer vicariously liable for some tortious conduct of a supervisor made possible by abuse of his supervisory authority, and that the aided-by-agency-relation principle embodied in §219(2)(d) of the Restatement provides an appropriate starting point for determining liability for the kind of harassment presented here.³⁷

As others have pointed out,³⁸ an enterprise that grants authority over vulnerable members of the population does indeed, presumably, increase the risk of sexual

³³ John Y., Jr. v. Chaparral Treatment Ctr., Inc., 124 Cal. Rptr. 2d 330 (Cal. Ct. App. 2002) at 337–338 (footnotes omitted). ³⁴ Barnes v. Costle, 561 F.2d 983, 996 (D.C. Cir. 1977) (MacKinnon, J., concurring). ³⁵ Gary v. Long, 59 F.3d 1391, 1397 (D.C. Cir. 1995) (quoting Barnes, 561 F.2d at 996 (MacKinnon, J-, concurring) (citing Restatement (Second) of Agency §219 cmt. e (1958)). ³⁶ See Faragher v. City of Boca Raton, 524 U.S. 775 (1998). ³⁷ Faragher v. City of Boca Raton, 524 U.S. 775 (1998), at 802. ³⁸ (Brodie, 2007).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

190

, ,  

assault, especially if the authority-figures are situated closely to the vulnerable individuals and placed in a position of trust, as is the case in schools, prisons, nursing homes, residential homes for the young, and so on. That being said, the aided-by-agency relation is not limited to cases where the tortfeasor has authority over the victim. In Costos v. Coconut Island Corp,³⁹ the defendant owned an inn the manager of which illicitly entered the plaintiff ’s room, who was a guest at the inn, and raped her. The Court of Appeals for the First Circuit held: By virtue of his agency relationship with the defendants, as manager of the inn, Bonney was entrusted with the keys to the rooms, including Costos’ room, at the Bernard House. Because he was the manager of the inn, Bonney knew exactly where to find Costos. The jury could find that Bonney had responsibilities to be at the inn or to have others there late at night. In short, because he was the defendants’ agent, Bonney knew that Costos was staying at the Bernard House, he was able to find Costos’ room late at night, he had the key to the room and used the key to unlock the door, slip into bed the beside her as she slept, and rape her.

Given California’s commitment to analyzing respondeat superior in terms of enterprise liability, the Restatement’s principle makes sense in that it applies respondeat superior to cases where the enterprise enabled the employee to commit a wrong. This also means that not every instance of harassment occurring at work will result in vicarious liability. As one commentator puts it, “. . . quarrelling neighbours, who extend their feud into the workplace, do not render the employer liable thereby,” because the employer has done nothing to substantially enable or enhance the risk of harassment.⁴⁰ Where does this leave us? There is a basis for thinking that in reasonably attempting to limit the scope of vicarious liability, the California Court of Appeals goes too far in declaring that the employer must have “motivating emotions [ . . . ] fairly attributable to work-related events or conditions.” Instead, the Restatement (Second) of Agency provides a more plausible principle. For the employer to be liable, it is enough if the employee availed himself of “the instrumentalities, access, authority or knowledge” he possesses as an employee, which he then “misused to commit the injury.”⁴¹ The role of the law is to identify particular aspects of the agency relationship and determine whether the employee’s abuse of them was a substantial factor in causing the injury.

³⁹ 137 F.3d 46 (lst Cir. 1998). ⁴⁰ (Brodie, 2007, p. 500). ⁴¹ (Roszkowski & Roszkowski, 2005, p. 239).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

191

7.2 Inadequacy of Attempts to Ground Respondeat Superior in Enterprise Liability Recall the problem with which we began. On the traditional reading of respondeat superior, the employer is not liable for the wrongs her employee commits if the conduct does not qualify as a “mode” of fulfilling her authorized duties. The result is that we have difficulty applying respondeat superior to a host of cases, such as those in which a school counselor sexually abuses a student, or in which a security contractor commits a theft against a client. But it seems we can fill this lacuna by analyzing respondeat superior as a form of enterprise liability. When an enterprise engages in activities from which it benefits but which also imposes foreseeable negative externalities on the community, the enterprise ought to bear the costs of those harms. These include the torts that participants in the enterprise commit using the “instrumentalities” that the enterprise lends to its participants (as I suggested in the previous section), thereby enabling them to commit the wrong in question. The victims of these torts can recover from the enterprise in question. We thus have, seemingly, a basis for respondeat superior capable of capturing the sorts of cases that fell between the cracks on the traditional construal. In what follows I will argue that the attempt to analyze respondeat superior as a form of enterprise liability is only partly successful. Such an analysis fails to accommodate two sorts of cases. The first kind of case is one in which the enterprise inflicts a harm on clients or customers who knew that doing business with the enterprise carries such a risk. The second kind of case is one in which the enterprise inflicts a harm the cause of which cannot be traced back to the conduct of those who own or operate the enterprise. As indicated in the previous section, it should come as no surprise that attempts to ground respondeat superior in enterprise liability fails, considering how radically different these two doctrines operate.

7.2.1 Assumption of Risk and Enterprise Liability Recall the justification for shifting the costs of negative externalities to the owners of the enterprise. A successful enterprise will typically benefit the owners and will confer some benefits upon its patrons as well. But an enterprise also risks inflicting harms, not only upon its patrons, but also upon others who had nothing to do with the enterprise. We can imagine that these individuals had no interest at all in the enterprise, in that the enterprise did not increase the expected welfare of those individuals. To expect them to shoulder the burdens of the harms that the enterprise causes is unfair. Rather, the burden should be shifted to those who inflicted the risk in the first place and who aimed to benefit from imposing that

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

192

, ,  

risk—i.e., the owners of the enterprise itself. Put simply, it is unfair to adopt the policy of retaining the benefits of foreseeable good option luck while simultaneously shifting the burden of foreseeable bad option luck to others. Consider, then, this case: Camping Brushfire A private business, Summer Camp Inc., offers summer camping experiences for youths throughout the region. The owners of the corporation do everything feasible to ensure that the camp workers are properly vetted and trained. Counselor, an employee of Summer Camp Inc., starts a campfire that she culpably fails to monitor, resulting in a brushfire that destroys nearby private property, burning down Rancher’s home. Do the owners of Summer Camp Inc. legally owe compensation to Rancher for the harm Counselor caused? That is, should respondeat superior apply in this case? There is a basis for thinking so, if we ground respondeat superior in enterprise liability. The owners of Summer Camp Inc., by doing business, imposed some risk on the community, including the foreseeable risk of wildfires. Though, by hypothesis, they did everything they could feasibly do to ensure that their employees minimize that risk, one such employee committed a tort all the same. The party that ought to shoulder the burden of the bad option luck is the party that stood to benefit from imposing that risk, i.e., the owners of the Summer Camp Inc. Thus, we can and should allow Rancher to shift the costs of the harm resulting from Counselor’s culpable actions to those who own Summer Camp Inc. That would constitute a fairer allocation of the costs of the risk that the enterprise undertook. Hence, grounding respondeat superior in enterprise liability yields the conclusion that those who own Summer Camp Inc. are legally liable for what Counselor does. The problem with this strategy, though, is that enterprise liability has difficulty accommodating other sorts of cases in which respondeat superior clearly applies. Suppose the individual who suffers the harm resulting from the risk the enterprise undertakes is not someone who had no interest in the enterprise, but is instead one of its clients or customers, as in this case: Security Abuse The homeowners of a gated community unanimously agree to purchase the services of Private Security Inc. to patrol and monitor shared grounds. The homeowners all know that hiring such a service introduces its own risks—the security guards might themselves abuse the position with which they have been entrusted. But the owners of the corporation do everything feasible to ensure that its employees are properly vetted and trained. As a result, there is nothing they can

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

193

feasibly do, within their remit, to further increase the expected welfare of the homeowners. Nonetheless, one night while patrolling the grounds, Patrolman breaks into Homeowner’s residence and attacks her. The doctrine of enterprise liability has surprising difficulty explaining why the owners of Private Security Inc. are legally liable for what Patrolman does. It is a risk of such a business that its employees, no matter how carefully vetted and no matter how scrupulously trained, might abuse their position. And by hypothesis, the patrons knew the risk of employee misconduct. On what basis can we adopt the view that the legal burden of compensation should fall solely on the owners of Private Security Inc.? Recall the basis for enterprise liability: the party that ought to shoulder the burden of the bad option luck is the party that stood to benefit by imposing that risk. This explains why the owners of Summer Camp Inc. ought to bear the costs of the harms its employee negligibly inflicts on Neighbor. But in Security Abuse, the victim’s expected welfare increases by doing business with the security contractor, and the victim willingly decides to purchase their services despite knowing that there is a risk an employee might abuse his position in precisely the way that happened to occur. The result is that the enterprise has not adopted the policy of shifting to its clients the foreseeable harms of the risks it undertakes; rather, the client counts as among those who willingly undertook the risk in the first place, precisely because it is rational for them to do so; that decision maximizes their expected welfare. One might point out that enterprise liability is a form of strict liability, and it is on this basis that the owners of the Private Security Inc. should be held liable for the tort its employee commits. I am not gainsaying this point, though. Rather, the issue is whether there is a basis for holding the enterprise strictly liable. The doctrine of strict liability, after all, does not impute liability absent any moral or legal reason for doing so; rather, it imputes liability absent fault. In cases where the enterprise’s employee commits a tort against a party who did not choose to undertake the risk of doing business with the enterprise, there is indeed a principled reason for holding the enterprise strictly liable: it is unfair to shift the consequences of bad option luck to parties who would not have benefited from good option luck, or on parties who did not choose to undertake that risk. But in Private Security, the bad option luck falls on a party who would indeed have benefited from good option luck, and who indeed choose to undertake that risk. So, if the owners of the enterprise are strictly liable in this case—that is, if the owners of Private Security Inc. bear enterprise liability—it has to be on some other basis.⁴² ⁴² One might argue that enterprise liability also has difficulty explaining why those who run public entities (such as public schools and state-run healthcare institutions) can be liable for the torts their

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

194

, ,  

At this point, we might repair to policy-based reasons for enterprise liability. Perhaps there are economic reasons to adopt enterprise liability even when there are no fairness-based considerations in its favor. For example, according to Robert Morris:⁴³ Enterprise theorists observe that enterprise liability causes little business dislocation while, at the same time, it does important service by repairing injured limbs, replacing lost breadwinners, and making livable the invalid lives of industry’s victims. The entrepreneur can avoid the shock effects of individual judgments by building a fund to pay them or by purchasing insurance. Since his competitors will have to do likewise no one suffers a competitive disadvantage. In this view the cost of insurance or funding is one of the regular costs of businesses and, as such, can be passed on to customers. In short, a victim’s loss is shifted to one better able to bear it, and thence spread on to the customers who benefit from the productive activity. It is proper that they take the burden with the benefits. And, furthermore, since each customer’s proportionate share is quite small, they can sustain the burden easily, whereas the original victim could not.

In addition, enterprise liability yields socially useful incentives for businessowners. As William Prosser puts it: “. . . an employer who is held strictly liable is under the greatest incentive to be careful in the selection, instruction and supervision of his servants, and to take every precaution to see that the enterprise is conducted safely.⁴⁴ So, these are reasons for thinking that enterprise liability holds in cases like Security Abuse after all. But we want to know not merely whether it is economically expedient to adopt a policy in which victims are allowed to recover from the enterprise in such cases. Nor do we want to know merely whether such a policy yields an incentive for businesses to select, train, and monitor their employees more carefully. In addition, we want to know whether the owners of the enterprise are morally accountable for the mistreatment the victim suffers. In such a case, the victim would have a directed claim against the owners, where that claim serves as its own basis for recovery. If we ground the doctrine of respondeat superior in enterprise liability, then the answer seems to be “no”. In such a case, the owners of Private Security, by failing to compensate Homeowner, have wronged the employees commit. If what grounds such liability is that the owners of an industry must bear its costs, then there seems to be no one liable where the industry is in the public sector. It is true that there is a sense in which public institutions (in a well-functioning democracy) ultimately operate for the benefit, and at the behest, of the public. But those who occupy positions of authority vis-à-vis these institutions derive personal benefits—in the form of salaries—that might serve as a basis for applying enterprise liability, thereby holding them liable for the torts their employees commit. ⁴³ (Morris, 1961). Footnotes omitted. For a critique of economic-based analyses of tort law in general, see (Coleman, 1992, p. 380). ⁴⁴ (Prosser, Keeton, Dobbs, Keeton, & Owen, 1984, pp. 500–501).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

195

community. The victim has only a derivative complaint against the owners insofar as compensating her is the prescribed means by which the enterprise discharges what is owed to the community. This strikes me as implausible.⁴⁵ To be clear, I do not take myself to have decisively argued that enterprise liability cannot ground respondeat superior. There are two reasons for this. First, I presented what I took to be the most compelling argument in favor of enterprise liability: when we stand to benefit by undertaking a gamble, fairness requires that we shoulder the costs should that gamble fail. I argued, though, that this principle fails to ground respondeat superior in cases like Private Security. Still, there might be other accounts of enterprise liability that can accommodate such cases. I only claim that the most compelling account does not. Second, some might accept the conclusion that policy-based considerations exhaust the legal duties that the owners of enterprises have in cases like Private Security. Those inclined to think as much need not repair to an alternative basis for respondeat superior. But those who believe that the owners of the enterprise are directly accountable to the victim and not just to the community need some other basis for respondeat superior.

7.2.2 Causation and Enterprise Liability There is, though, another reason to doubt that we can ground respondeat superior solely in enterprise liability. The attempt to ground respondeat superior in enterprise liability relies on the claim that if you non-consensually impose risks on others who do not stand to benefit from that risk-imposition, it is legally incumbent upon you to compensate them if that risk results in a harm. That is, you cannot keep all the good option luck to yourself while shifting all the bad option luck to others. But consider this case: Rehab Clinic Rehab Inc. is a privately owned inpatient drug rehabilitation center. The owners sell the center to another private party, who re-christen the business New Rehab Clinic, Inc. The transition happens quickly and clandestinely at noon on a Wednesday; the new owners do not immediately make any changes in personnel or policy, all of whom remain unaware of the change in ownership. At 2:00 pm on that day, one particular staff member, Abusive, assaults Inpatient. Are the owners of New Rehab Inc. a cause of the harm that Abusive committed at 2:00 pm? The answer is “no”. Though, by signing the agreement at noon, the owners caused Abusive to be employed at New Rehab Inc. rather than the ⁴⁵ Others, though, have averred that enterprise liability lacks a pre-legal basis. See (Zipursky, 2005, p. 132).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

196

, ,  

organization’s predecessor, the owners in no way caused the employee to be where he is or to do what he was doing. Now, according to enterprise liability, the owners of an enterprise are legally liable for the harms resulting from the foreseeable risks the enterprise imposes on others. But the harm that Abusive commits against Inpatient does not result from any risks the owners of New Rehab impose. This is because the owners of New Rehab are not among what causes Abusive to commit that harm. The owners neither constitute a but-for cause of the harm, nor do they satisfy a NESS/INUS test of what causes the harm. It is hard to see, then, how the doctrine of enterprise liability can accommodate this case. The upshot is that if we ground respondeat superior in enterprise liability, it turns out that the owners are not legally liable as a result of respondeat superior. This strikes me as implausible. One might argue that New Rehab clinic might bear successor liability for what Abusive does. In such cases, torts that predate new ownership are still cognizable against the new company. But such a move is unlikely to work in this case. After all, when New Rehab clinic takes over, no tort has been committed and no tort is in progress. Abusive commits the assault after the change in ownership. For this reason, successor liability is unlikely to succeed in explaining why New Rehab Inc. is liable for the assault Abusive commits. It might be argued, in response, that because New Rehab Inc. has purchased the business “lock, stock, and barrel,” they are on the hook for the torts that their new employees commit. But this is not in dispute; the issue, rather, is: why does ownership confer liability? We might, of course, say that new owners allow Abusive to commit the harm, by refraining from making any changes to personnel or staff. But presumably our legal responsibility for what we allow to happen is substantially less than our legal responsibility for what we make happen. It is tempting to respond by saying that the owners bear special responsibility for what their employees do, of the sort that eliminates the relevance of the distinction between doing and allowing. But the claim that owners bear special legal responsibility of this sort is precisely the claim we are trying to ground—i.e., respondeat superior—and we thus cannot appeal to it in doing so. Indeed, the very fact that respondeat superior ignores the doing/ allowing distinction is itself evidence that we cannot ground it by appealing to what owners of an enterprise risk causing. Rehab Clinic is, of course, a recherché case. We might, accordingly, question whether it demonstrates that enterprise liability cannot ground respondeat superior. But the example has no such ambitions. Rather, the purpose of raising it is to help distinguish two kinds of relationships that typically but do not always overlap: the causal versus the formal relationship between owners and their employees. The doctrine of enterprise liability is a claim about the causal relationship between owners and their employees; the doctrine is grounded in a claim about how we ought to distribute the benefits and burdens that our gambles cause. But the doctrine of respondeat superior, on the other hand, might be best

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

197

construed at least in part as a claim about the formal relationship between owners and their employees. We should expect difficulty in grounding a claim about the formal relationship between owners and their employees in a claim about the causal relationship between them. To recap, I have presented two arguments against the view that we can ground respondeat superior in enterprise liability. The first argument against this view was that it is unclear why fairness requires shouldering the compensatory burden when the risk increased the victims expected welfare and when the victim consented to that risk. The second argument against the view was that an appeal to our liability for the harms we cause by imposing risks cannot accommodate the fact that respondeat superior ignores the relevance of doing and allowing. The upshot is that if we are to ground respondeat superior in enterprise liability, we need to consider alternative methods.

7.3 Authority-Based Accountability as a Solution In what follows I develop a wholly novel account of respondeat superior. I argue that service-providing vendors will typically have a defeasible special duty of care toward their patrons in virtue of an “asymmetry of vulnerability” in their relationship. That is, the patron puts herself in a vulnerable position by entrusting the vendor with the patron’s body, personal property, or general wellbeing. Though the specifics of the resulting special duty of care will vary depending on the service the vendor is providing, it morally and potentially legally enjoins the duty-bearer to do no harm to the individual to whom the duty is owed. This duty is what grounds the vendor’s strict financial liability in cases where the vendor nonculpably causes the victim harm. But why believe that those who own or run an enterprise are morally liable in cases where their employees violate the special duty of care? Those who own or run the enterprise confer the special duties of care on their employees through a division of agential of labor. The result is that those who own or run the enterprise can bear authority-based accountability for what their employees do when they violate that special duty of care. This account, I will argue, can accommodate the sorts of cases with which enterprise liability had difficulty in the previous section.

7.3.1 Duties of Care and Asymmetric Vulnerability In the first half of this book, I developed a picture of how authority-based accountability inculpates participants in a cooperatively committed harm. In a division of agential labor in which the supervisor counts as a deliberator and the subordinate counts as the executor, the supervisor can bear authority-based

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

198

, ,  

accountability for what the subordinate does when the supervisor’s motivating reasons are morally problematic. The deliberator’s own morally problematic motivating reasons are part of the mechanism inculpating her in what the executor does: those motivating reasons constitute the purpose that the deliberator furnishes for the executor. Where those motivating reasons are problematic, the deliberator thereby furnishes for the executor a morally problematic purpose. The result is that the deliberator can be accountable for the fact that what the executor does is wrongful. That is, the deliberator can bear authority-based accountability for what the executor does. An upshot of this view, though, is that a deliberator is inculpated in the actions of an executor only if the deliberator is in some way already accountable. As I put it earlier,⁴⁶ the deliberator’s practical reasoning “infects” the actions of the executor, and the deliberator is thereby accountable for that “infection”. But in the examples of respondeat superior discussed so far, those in charge (the owners, the employers, the superior officers, etc.) are not themselves accountable for anything. In each example, those in charge did everything that could reasonably be asked of them when it comes to selecting, training, and instructing their subordinates. Again, put metaphorically, the deliberator is not “infected” with morally problematic motivating reasons. There is thus nothing with which to “infect” the executors, and thus no “infection” to pin on the deliberator. It might seem, then, that the account of authority-based accountability can have nothing to say about respondeat superior. But this is not so. To see why, I will begin with a discussion of special duties of care. The decision to provide a service to others—whether ultimately for personal gain or for more altruistic reasons—will typically yield a special duty of care toward those to whom we are providing that service. This special duty of care that service-providing vendors have toward their patrons has several characteristics. When we inflict harm against someone who has a claim against being harmed, that treatment is wrongful, of course. That wrong, though, is typically made morally worse if we have a special duty of care toward the victim. Holding fixed the severity of the harm in question, it is markedly worse for a parent to harm his child wrongfully and culpably than it is for a stranger to do so.⁴⁷ The same goes (albeit to varying degrees) when taxicab drivers assault or endanger passengers, when school instructors or counselors abuse students, when police-officers use unnecessary force on civilians, when healthcare personnel engage in malpractice, when plumbers steal from their customers, and so on. This is not to say that the duties are the same in strength or breadth; the moral duties that a nurse-practitioner has toward a patient is stronger than the moral duty that a stylist at a hair salon has toward a client. The degree to which such a

⁴⁶ See section 4.1 (chapter 4).

⁴⁷ See (Bazargan-Forward, 2018).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

199

violation exacerbates the wrongness of the harm inflicted will depend partly on the duty of care owed—its strength and its type. But, in general, when we entrust someone with our bodies, our homes, our personal belongings, or our wellbeing in general, and they betray that trust, the wrong that they thereby commit is made morally worse not only by virtue of violating the standing claim we have against gratuitous harm in general, but also by violating the special duty of care that they have toward us specifically. But what is the source of this special moral duty of care? It is “special” in that the patrons have an agent-centered claim to that care not merely in virtue of their status as persons, but also in virtue of the relationship they have established with the service-providing vendor: a relationship in which it is mutually agreed that the patrons will entrust the vendor with their bodies, their personal belongings, or their wellbeing in general. There is, then, a relevant moral asymmetry between the patron and the serviceproviding vendor: the service requested of the vendor often requires providing them with access to the patrons’ bodies, private information, personal property, and so on. This often leaves the patron especially vulnerable to the vendor’s actions. Sometimes this vulnerability is quite severe, as is the case when a surgeon is performing surgery on a patient, or when armed guards provide a security escort to clients transferring valuables to a banking establishment. The vulnerability is less so in other cases, as when a hairstylist is cutting a client’s hair, or when a taxicab driver drives a client to a destination. It is because the patron is made vulnerable in this way by entrusting the vendor that the vendor comes to have a special duty of care toward the patron. None of this is to say that service-providing vendors are never vulnerable to decisions and actions of their patrons. A woman taxicab driver might be especially vulnerable to drunken men in her car. An immigrant self-employed as a handyman might be especially vulnerable to the bigoted accusations of a wealthy client. But in cases like these, the vendor is not vulnerable qua vendor, but instead vulnerable qua woman, or qua immigrant. I am limiting myself to the claim that there is an asymmetry between vendors and patrons, not in their vulnerability sans phrase, but in their vulnerability qua vendor and qua patron. There is, put differently, an asymmetry in role-based vulnerability. It is due to this asymmetry in role-based vulnerability that that the vendor comes to have a special duty of care toward the patron qua patron. Other morally relevant vulnerabilities, such as gender-based or race-based vulnerabilities—might, of course, interact with—and potentially defeat—that special duty. Of course, sometimes vendors qua vendors are financially vulnerable to their patrons qua patrons. A small law firm might financially depend on the business of a single, wealthy client. But again, that vulnerability is not peculiar to the role that the vendor or the patron has adopted. It is, instead, a highly contingent feature of the situation that they are in. And again, this contingent feature might very well be

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

200

, ,  

morally relevant to the all-things-considered duties that the parties have to one another. So far, then, the view can be summarized as follows: service-providing vendors typically have a defeasible, special, moral duty of care toward the patron in virtue of the role-based vulnerability that the patron qua patron has toward the vendor qua vendor, where that role-based vulnerability results from entrusting the vendor with the patron’s body, personal property, or general wellbeing. The content of the special duty of care depends on the service the vendor is providing. At the very least, special duties of care enjoin the duty-bearer to do no harm to the individual to whom the duty is owed. If the vendor ends up irreparably harming the patron, culpably or not, the vendor might still have a chance to compensate the patron for that harm. Doing so reduces the amount of harm the vendor caused, which more effectively satisfies the special duty she had not to cause undue harm. The special duty of care, then, when violated, morally (and potentially legally) grounds a compensatory duty toward the victim. I will call this “the argument from vulnerability”. This argument also helps provide a ground for strict product liability. From the 1940s through the 1960s, American law professors Fleming James Jr. and William Prosser adopted competing accounts of product liability, a concept still in its infancy at the time.⁴⁸ Both agreed that the then-current negligence and warranty law failed to address the legal problems that defective products presented in the marketplace. But whereas James argued that warranty law could be tailored to meet the needs of product liability, Prosser, on the other hand, argued otherwise; his view ultimately came to dominate the legal analysis of product liability.⁴⁹ In a landmark case, Greenman v. Yuba Power Products, Inc. (1963), the Supreme Court of California explicitly formulated and adopted the doctrine of strict liability in tort for defective products.⁵⁰ This marked a shift in the theory of product liability toward the rationale grounding enterprise liability. On such a view, the manufacturer’s liability is grounded not in fault, but rather in an obligation to internalize the costs of any harms that their defective products might cause. Associate Justice Roger J. Traynor wrote the majority opinion, in which he cited his own earlier concurring opinion in Escola v. Coca-Cola Bottling Co. (1944), which helped establish a new regime for product liability: Even if there is no negligence, however, public policy demands that responsibility be fixed wherever it will most effectively reduce the hazards to life and health inherent in defective products that reach the market. [ . . . ] It is to the public interest to discourage the marketing of products having defects that are a menace to the public. If such products nevertheless find their way into the market it is to ⁴⁸ See (White, 2003, pp. 169–170), (Kiely & Ottley, 2006, pp. 2–21). ⁴⁹ See (Kiely & Ottley, 2006, pp. 2–21). ⁵⁰ (Owen, 2015, pp. 14–22).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

201

the public interest to place the responsibility for whatever injury they may cause upon the manufacturer, who, even if he is not negligent in the manufacture of the product, is responsible for its reaching the market. However intermittently such injuries may occur and however haphazardly they may strike, the risk of their occurrence is a constant risk and a general one. Against such a risk there should be general and constant protection and the manufacturer is best situated to afford such protection.⁵¹

Soon after Greenman, the Supreme Court of California extended strict liability to all parties involved in the manufacturing, distribution, and sale of defective products (including retailers).⁵² William Possner, the official reporter for the Second Restatement of Torts, and who had earlier written that strict product liability should be “declared outright,”⁵³ codified the Greenman doctrine in Section 402A.⁵⁴ The highest courts in nearly all US states followed suit.⁵⁵ This occurred in tandem with America’s rapid embrace of consumer protection laws at the state and federal level.⁵⁶ As reflected in Traynor’s opinion, the argument for strict product liability is largely policy-based. The burden of a defective product must fall on either one of two non-negligent parties: the manufacturer or the consumer. The conventional wisdom is that it is preferable to impose the economic burden on the manufacturer because it can better absorb such costs, which it then passes on to other consumers. The manufacturer thus becomes a de facto insurer against its defective products.⁵⁷ But the argument from vulnerability helps demonstrate that there is an inprinciple basis—not just a policy-driven one—for strict product liability. Recall that according to the argument from vulnerability, when we entrust a vendor with our body or our personal property the vendor has a special duty of care toward us, which requires the vendor to compensate for harm that they cause—including ones for which they are not at fault. In such cases, we typically put our body or our personal property in the hands of the vendor; it is by so doing that we end up vulnerable. When we purchase a product, we typically bring that product into our lives in some way, by bring it into, on, or near our bodies or our property. We are thereby made potentially vulnerable by that product should it malfunction. ⁵¹ Escola v. Coca Cola Bottling Co., 24 Cal. 2d 453; 150 P.2d 436; 1944. ⁵² (Nolan & Ursin, 1995, p. 117). ⁵³ W. Prosser, The Assault upon the Citadel (Strict Liability to the Consumer), 69 Yale L.J. 1099, 1133–1134 (1960). ⁵⁴ (Kiely & Ottley, 2006, pp. 2–21). Prosser imposed in Section 402A a requirement that a product defect must be “unreasonably dangerous.” It is unclear why Prosser included such a caveat given that it implies “fault” and is thus inconsistent with strict liability for defective products as articulated in Greenman. The caveat was subsequently removed by Alaska, California, Georgia, New Jersey, New York, Puerto Rico and West Virginia. See (Heafey & Kennedy, 2006, pp. 2–9). ⁵⁵ (Owen, 2015, pp. 14–22). ⁵⁶ (Cohen, 2008, p. 360). ⁵⁷ See (Heafey & Kennedy, 2006, pp. 2–10).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

202

, ,  

Products are no less effective than services in rendering us vulnerable, and thus in grounding a duty of care on the part of those who sell the product. So, if the argument from vulnerability provides an in-principle basis for strict liability of vendors providing services, then it should do the same for vendors providing goods. But since my focus here is on respondeat superior, I will continue to focus on service-providing vendors.

7.3.2 Reparative Duties and Conferring Duties of Care Suppose, then, a self-employed dentist is working on a patient when the drill she is using suddenly and unexpectedly malfunctions thereby causing a painful injury to the patient. Morally, the dentist has a reparative duty to the patient, even given that malfunction was not the dentist’s fault. This is because the dentist has a duty to ensure that she provides a certain standard of care pertaining to her patient’s oral health; should she fall below that standard, it is incumbent upon her to make up the difference. We might think of this in terms of local distributive justice. Supposing that the harm in question has to fall either on the patient (in terms of an injury) or on the dentist (in terms of remuneration), and supposing that the dentist has a special duty of care toward the patient (and not vice versa), it is fairer—as a matter of morality—for dentist rather than the patient to absorb the cost of that harm. The patient’s legal claim falls out of that moral claim. Likewise, suppose a self-employed hairdresser is cutting hair when a sudden thunderclap startles her, causing her to snip off a handful of her client’s hair. Morally, the hairdresser has a reparative duty, even though she is not accountable for being startled. Or suppose a self-employed taxicab driver offers an alternative, quicker route which happens to be blocked off, as a result of which the passenger fails to catch her train on time. Morally, the taxicab driver has a reparative duty toward the passenger even if it was entirely reasonable to think that the route he chose would indeed be quicker. In all these cases, the service-providing vendors have reparative moral duties even though the harms they non-culpably committed are the type of harms the patrons knew might occur. These reparative duties are partly constitutive of the moral duties of care they owe. The kind of reparative duty owed will differ depending on the details of the case. Perhaps in each of the three discussed the self-employed service-providers should offer their patrons free access to their services, up to the point necessary to repair damage incurred. Where the patron might understandably wish to refrain from doing business with them again, the service-providers might offer monetary compensation. A constraint of proportionality, though, limits the compensatory deprivations to which the service-providers are liable. According to this constraint, the morally weighted harm that the compensatory deprivation inflicts on the liable party cannot be too great relative to the morally weighted good it

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

203

confers on the beneficiary.⁵⁸ Since, by hypothesis, the service-provider’s duty-ofcare violation was non-culpable, the proportionality constraint will be quite stringent, in that the compensatory deprivation will be limited to those that do not impose significant hardship on the service-provider. To be clear, this is not to say service-providers always have reparative moral duties whenever they harm their patrons. Where a third party is a cause of the violation, or where the duty-bearer is herself a direct victim of the circumstances resulting in the violation, that might attenuate or eliminate the reparative duty. Neither is any of this to say that service-providing vendors cannot antecedently divest themselves of reparative liability for non-culpable wrongdoing, by forming an agreement with the patron to that effect. Rather, the point is that there is a strong presumption in favor of holding the vendor morally and potentially legally liable for the non-culpable transgressions she commits, even if the patron was aware of the risks involving in doing business with that service provider. The vendor is presumptively liable because she had a special (though defeasible) duty of care toward the patron. She has that duty of care in virtue of the asymmetry in role-based vulnerability between the two. That asymmetry is an effect of entrusting the vendor with the patron’s body, personal property, or wellbeing. So, the owners of an enterprise, by virtue of offering particular kinds of services, can come to have concomitant duties of care toward their patrons. But in examples like Security Abuse, the owners are not the ones committing the violation. And in cases like Rehab Clinic, the owners are not even a cause of the violation. So how do we justify the view that the owners are among those who nonculpably violated a duty of care toward their patrons? It might be tempting to think that if it is because of the employers that the employees come to possess a duty of care, then the employers are liable should the employees violate that duty. But this cannot be right. Suppose a doctor has a newborn as his patient, to whom he consequently has a duty of care. Suppose further that this doctor served in a fertility clinic where he facilitated the in vitro fertilization that resulted in that same newborn. Hence, it is because of the doctor that newborn’s mother has a duty of care toward her child. But clearly, should the mother violate that duty of care, the doctor does not thereby violate his duty of care toward the infant, even though it is because of the doctor that the mother has the duty of care in the first place. This is just to show that we do not “transfer” or otherwise “share” duties of care toward an agent simply by causing others to have duties of care toward that agent.⁵⁹ Though the exact picture will depend largely on how the enterprise is bureaucratically organized, we might say that in general the owners act vicariously through their employees, as a result of which the employer is morally and legally ⁵⁸ See discussion of ‘narrow’ proportionality in (McMahan, 2009, pp. 20–21). ⁵⁹ For any overview of these issues, see (Brake & Millum, 2016).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

204

, ,  

liable for the violations the employees commit. But this claim is, of course, just the explicandum, and as such cannot serve as the explicans. How is it, then, that the owner comes to be liable should the employee violate a duty of care, when the owner is not a cause of that conduct? The argument for authority-based accountability helps answer this question. The owners establish a division of agential labor with the employees, in which the owners confer upon the employees a protected reason to discharge certain duties of care toward their patrons. By fixing the content of the employee’s protected reasons, the owners thereby determine the function of what the employee does qua employee.⁶⁰ The owners thereby confer upon the employee the purpose of acting on the employer’s duty of care. So, when the employee violates that duty, the owners eo ipso fail in that duty as well. This is not to say that the employer is necessarily culpable for that failure. But recall that, ceteris paribus, should we fail in a duty of care, we are morally and potentially legally liable for reparative costs, even if that failure is non-culpable. The result is that the non-culpable employer is liable to proportionalityconstrained deprivations instrumental to compensating the patron whom the employee culpably mistreated. The employer is liable even if the employer is not a cause and could not have prevented that violation. This is because the employer purposed the employee with discharging that duty of care; where the employee fails, the employer thereby fails. We can, then, make sense of respondeat superior by grounding it in special duties of care that are themselves conferred through a division of agential labor. Note that this strategy can accommodate cases like Rehab Clinic. In a division of agential labor in which the employers count as the deliberators and the employees count as the executors, the practical reasons the employers take there to be in favor of providing the relevant services anaphorically fix part of the employer’s purpose in providing those services. That is, in providing a comprehensive account of the employee’s protected reason, she would have to refer her patron to the employer. In Rehab Clinic, the change in ownership means Abusive might not know who his employers are, specifically. But a de dicto reference to his employers can suffice in an explanation of the basis for the duty he has toward his patients. Put simply, the new owners, by virtue of taking ownership, confer a moral and legal duty of care on Abusive, even if he does not know that they, specifically, have done so. So, though the employers do not help cause Abusive to do anything, Abusive has a duty toward the inpatients partly in virtue of the formal agreement he has with his employers. This strategy can also accommodate cases like Private Security as well. In that example, Homeowner decided, reasonably, to procure the services of Private Security Inc., knowing that there is always a chance that their employees might

⁶⁰ See chapter 1.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

    

205

abuse the power with which they have been entrusted. The suggestion was that by accepting this risk, it becomes morally and legally unfair to require Private Security Inc. to shoulder the burden of bad option luck in its entirety. But I argued that vendors have a defeasible special duty of care toward their patrons as a result of which the vendors are morally and legally liable to the proportionality-satisfying deprivations necessary to compensate their patrons should the vendors harm their patrons, culpably or not. The result is that the owners of Private Security Inc. are liable after all. None of this is to say that enterprise liability plays no role in grounding respondeat superior. Rather, the point is that enterprise liability cannot do all the work we want it to do. Though we might be legally liable for the harms resulting from the risks we undertake, respondeat superior is not limited to cases in which the owners cause the risk of harm. And though there are reasons to allocate the good and bad outcomes of option luck so that it falls on those who stood to benefit from undertaking the risk in question, respondeat superior applies in cases where the patrons also stood to benefit by undertaking that risk. A more complete analysis of respondeat superior is one grounding it partly in the special duties of care that owners discharge via their employees by conferring those moral and legal duties of care upon them in a division of agential labor they establish. The result is a more plausible account of respondeat superior, not because it accommodates recherché cases, but because it locates the employer’s legal liability in the moral risks of formally authorizing others, in a division of agential labor, to discharge the special duties of care that she herself owes. Where they fail, she fails.

7.4 Conclusion In this chapter, I began by noting that in many cases enterprise liability, as it is traditionally construed, has no problem grounding respondeat superior. These are cases in which the wrong and its attendant harm qualifies as a negative externality of undertaking the kind of business the employer is running. These wrongs might range from negligence in driving to negligence in handling volatile materials. We can also chalk up some intentional torts to the risks of running a business. These are limited to cases in which the tort is committed against a party that did no business with and had no interest in the enterprise in question. Suppose the employee of a cybersecurity company uses the specialized technology available at her place of work to steal personal information from random individuals. It might be fairer for the owner to be held financially liable for these harms, even if the owner is non-culpable, since she is the one who risked these negative externalities in furtherance of running a business from which she primarily benefits. It is fairer to hold her legally liable insofar as the alternative places the costs of addressing the harms on victims who did not impose or stand to benefit from such risk-taking.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

206

, ,  

But how do we accommodate enterprise liability in cases where the owners are not a cause of what their employees do? Sometimes one set of owners “takes the reins” of a business without immediately exercising any causal influence on the employees who might be wholly unaware of the transition. In such cases, the harm that the employees commit does not count as the manifestation of a risk that the owner imposed. After all, a harm resulting from a risk I take counts as such only if I am a cause of that harm. So, grounding respondeat superior in enterprise liability means that respondeat superior is hostage to the causally contingent relationship between owner and employee. And what about cases in which the enterprise imposes risks on those who willingly, knowingly, and reasonably accepted the risk of harm by choosing to accept services from the enterprise? Enterprise liability, as traditionally construed, is grounded in the claim that those who undertake a gamble from which they stand to benefit are among those who ought to shoulder the burdens of bad option luck. This would suggest that, for example, a nursing home is not legally liable should the caretakers culpably harm those under their care, so long as such a harm is foreseeable—which it is in any service where employees are placed in positions of power over their clients. I responded to these two worries univocally. I argued that service-providing vendors—from doctors to taxicab drivers—possess special moral and legal duties of care toward their patrons. This special duty of care yields defeasible agentrelative obligations to refrain from unduly harming the patron. Should a vendor harm the patron in such a way, the vendor would have reparative duties, even if the patron was aware of and accepted the risks of doing business with the vendor. The vendor is morally and legally liable in this way because the patron entrusts the vendor with her body, personal property, or wellbeing in general. The result is a role-based asymmetry in the vulnerability between the two parties—an asymmetry that yields the special duty in question. In cases where the vendor’s employees are the ones who violate the duty of care, the owners of the enterprise are thereby morally and legally liable since they conferred the special duties of care owed to the patrons upon the employees via a division of agential labor. Thus, the owners fail in their duties should their employees do so, even if the owners did not cause and are not accountable for what the employees have done. The upshot is that a comprehensive analysis and justification of respondeat superior requires more than an appeal to the doctrine of enterprise liability. In addition, we need to appeal to the special duties of care that owners have toward their patrons. And I argued that in order to understand how the owners come to be implicated in the violation of the duties that their employees commit, we need to appeal to the argument for authority-based accountability.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

8 Institutional Racism At the broadest level of generalization, types of institutions include educational, familial, religious, economic, political, legal, linguistic, penal, military, and medical. These in turn nest further, more specific institutions. Put roughly, an institution consists at least partly in roles and rules enabling and regulating cooperation in furtherance of some specified end.¹ Sometimes these roles, rules, or ends (or side-effects thereof) wrong members of a racialized group. This might warrant attributing racism to the institution itself—what is known as “institutional racism”. The harms of institutional racism are hard to overstate, in part because the racist effects of multiple institutions can synergize absent any coordinated attempt to do so. Some refer to this as “structural racism”—a term also used to refer to the systemic and ongoing racial inequality resulting from racist practices in the recent past.² Here, though, I am concerned with institutional racism only; I will focus on the accountability of the members in these institutions who, individually, have little control over the institution’s roles, rules, or ends. Are they accountable for the institution’s racism? On the one hand, we might think that insignificant influence yields insignificant accountability. Yet, at the same time, the institution’s racism is possible only because such individuals implement its problematic roles, rules, and ends. How, then, do we make sense of accountability for institutional racism? It is characteristic of institutional racism that many of the individuals currently constituting the institution might not themselves be racist. For instance, Blacks in the US routinely pay mortgage rates higher than what Whites pay, not because (or not just because) lending agents covertly adopt racialized accreditation criteria, but because mortgage rates depend on the economic risk of lending credit to the applicant, where that risk is analyzed by reference to the neighborhood in which the applicant is purchasing a residence. Hence, residential segregation can result in a disparity in mortgage rates.³ This is, of course, just one example. All such examples share this feature: the institution in question exacerbates racial inequality even absent any intention to do so.

¹ This is not to say that institutions consist solely in roles, rules, and persons. See (Epstein, 2015, p. 46). ² (Shelby, 2016, pp. 26–28). ³ See (Dymski & Mason, 2005).

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0009

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

208

, ,  

In many such cases, racists once operated the institution in question; in doing so, they established norms with the overt or covert purpose of achieving racist ends. Even when the racist individuals leave, the problematic norms sometimes persist. The result seems to be a racist institution absent any racists. I will argue, though, that if an institution is racist, then at least some of its current members are acting in a racist way. Contrary to what others have maintained, it is conceptually impossible for institutional racism to exist absent any racism in that institution. That is, the effects of an institution qualify as racist only if those effects flow from the racism of its members. At first, this might seem to narrow the extension of ‘racist institution’. But I argue that an attitude of indifference toward racial injustice itself counts as a racist attitude. The result is that racist institutions remain prevalent, even given the view that an institution is racist only insofar as its members are racist. Once I defend this “reductionist” account of institutional racism, I argue that authority-based accountability can shed some light on individual accountability for institutional racism. The chief claim I make is about the accountability of nonracist members. Suppose I am a relatively low-level employee at an institution whose leaders do not harbor any racial animus but who wrongly maintain policies that happen to perpetuate racial injustice. Absent any excuse, these leaders are accordingly guilty of indifference-racism. But what about me? Suppose my actions qua employee do not contribute to racial injustice. And suppose that I do not harbor the problematic attitudes of my superiors. Am I off the hook? It seems I must be, since my actions, my attitudes, and their effects are morally unproblematic. But I argue this is not so. The indifference-racism of my superiors inculpates me at the bar of authority-based accountability. I argue that the moral effects of racism—including indifference-racism—can “transmit” across individuals, given the argument for authority-based accountability. This means an employee of a racist institution is inculpated in the indifference-racism of her employers, even if a) the employee harbors no racist attitudes, and b) what the employee does is permissible when described extensionally. This conclusion does not by itself wholly determine who is accountable for what in the context of institutional racism. But it is an important step in that direction. This chapter proceeds in an importantly different way from those that precede it. So far, I’ve argued that deliberators are inculpated in the actions of their executors at the bar of authority-based accountability. In this chapter, though, I proceed in the opposite direction: I argue that executors can be accountable for what their deliberators do at the bar of authority-based accountability. More specifically, I apply the theory of authority-based accountability to determine whether employees working within a racist institution (but who are not themselves racist) can be accountable for the racism of their employers.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

209

If an employee knows or should have known that her conduct enacts her employer’s racist aims, then she is accountable for so acting (absent any excusing conditions, of course). To explain this, we don’t need anything as involved as authority-based accountability. Rather, the employee’s accountability wholly depends on her own causal control and knowledge of what she is doing. The problem, though, is that some acts, which are otherwise harmless or even beneficial, can be made wrongful by the racist attitudes of the actor.⁴ In such cases, it’s not obvious what an employee should do when the employer but not the employee harbors the wrongful attitudes in question. Institutional racism is replete with such cases. I will argue that the employee, in such cases, does indeed commit a wrongdoing—or at least a pro tanto wrong. The argument for authority-based accountability explains why. In presenting this argument, I will rely on three simplifying assumptions. As mentioned, I argue that the effects of an institution qualify as racist only if those effects flow from the racism of its members; in doing so, I will assume for the time being that the relevant members are those who are in charge of the institution in question. This will include owners, executives, supervisors, administrators, directors, and so on. This is a simplifying assumption since the relevant members can include employees who do not occupy positions of unilateral authority. But to keep the argument tractable, I will assume otherwise. Also, in addressing whether and how individuals are inculpated in the racism of an institution, I will focus on the inculpatory effects of obeying instructions, directives, orders, and the like. Again, this is to keep the argument tractable. As I will indicate in the conclusion, the argument I present generalizes to a variety of bureaucratic arrangements, including arrangements in which individuals work together as equals. Finally, I will assume that all the instances of racism I mention—including racist attitudes and actions—are culpable absent any indication otherwise. This assumption is also simplifying since circumstances might partially or fully excuse (though not justify) racism. I will begin in section 8.1 by canvassing views of racism as such. After outlining arguments for non-reductionist views of institutional racism, I will then present reasons for thinking that such views are mistaken. In doing so, I will argue that an agent who is indifferent toward the racial injustice her actions cause, thereby harbors a racist attitude—even if it is only incidental that her actions cause a racial injustice. In section 8.2, I address the following issue: if racist attitudes affect the moral valence of an act, does an employer’s racist attitudes affect the moral valence of an employee’s actions? The answer, I argue, is “yes”. I appeal to the argument for authority-based accountability to explain why. I end in section 8.3 by considering the implications of this view for other kinds of institutional arrangements. ⁴ For further discussion of this contentious claim that attitudes can be deontically relevant, see section 2.3.1, chapter 2.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

210

, ,  

8.1 Metaphysics of Racism With respect to individual agents, the debate about the nature of racism has focused on whether racism is located fundamentally in beliefs, feelings, or behavior. Some adopt a cognitivist approach according to which racism is located fundamentally in attitudes with propositional content. For example, Tommie Shelby locates racism in beliefs about racial superiority and racial essentialism that rationalize oppression.⁵ Some non-cognitivists, like J. L. A. Garcia, locate racism in non-cognitive ill will and disregard.⁶ Other non-cognitivists, like Michael Dummett, locate racism in behavior resulting from feelings of prejudice.⁷ But as others have noted,⁸ these three accounts—cognitive, conative, and behavioral—all analyze racism by reference to individual agents. On these views, racism is fundamentally manifested in persons. This is itself a contentious presumption. In light of this, recent debates focus on addressing what Lawrence Blum calls the “categorical plurality” of racist phenomena: a variety of ontological kinds can, apparently, be irreducibly racist.⁹ These include not just beliefs, emotions, and actions, but also, laws, institutions, societies, and so on. The contention lies between those on the one hand who argue that as a property racism applies only to individual agents, and those who argue on the other hand that as a property racism can apply to social phenomena. So, for example, Garcia argues at length in favor of what I will call a “reductionist” view of racism. On this view, social entities are not themselves metaphysically capable of racism (group agency notwithstanding). Any apparent claim to the contrary is best understood as the claim that the individual agents on which the social entity supervenes are racist.¹⁰ On this view, the racism of social entities derives from and is reducible to the racism of the subvening agents. On the other hand, Haslanger¹¹ and Shelby¹² have independently argued in favor of what I will call a “non-reductionist” view of racism. On this view, social entities can indeed be racist, in and of themselves. Though such racism supervenes on the agency of individual persons, it doesn’t derive from and is not reducible to the racism of those persons. As a result, there can be, for example, a racist institution without any racist individuals. In what follows, I canvass what I take to be the most convincing arguments for non-reductionism about racism, after which I argue against them.

⁵ (Shelby, 2002). ⁶ See in particular (Garcia, 2004, pp. 39–40, 43–45) who has argued for this point over many articles. See also (Arthur, 2007, p. 17). ⁷ (Dummett, 2004, p. 28). See also (Philips, 1984, p. 77). ⁸ See for example (Haslanger, 2004). ⁹ (Blum, 2004, pp. 62–63). ¹⁰ See (Garcia, 1996), (Garcia, 1997), (Garcia, 2004). See also the debate between Mills and Garcia (Garcia, 1999), (Mills C. W., 2002), (Mills C., 2003). ¹¹ (Haslanger, 2004). ¹² (Shelby, 2016, p. 24).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

211

8.1.1 Non-Reductionist Accounts of Racism An institution with a policy “race-neutral” in its content and rationale might unfairly disadvantage a racialized group even if it imposes no greater burden on that group than any other, provided that the racialized group in question is already unfairly disadvantaged. It might be that no one in the institution aims at that effect. More generally, it might be that no one in the institution harbors antipathy toward the racialized group in question. Yet, as Shelby points out, such an institution can play a role in keeping groups “already disadvantaged by racism [ . . . ] in their disadvantaged condition, thus leading some to conclude that they occupy this low station because of the disadvantaged groups’ culpable failings or inherent inferiority.”¹³ Shelby calls this kind of institution “extrinsically racist”. Likewise, Michael Hardimon calls the kind of racism attributable to such institutions “attitude-independent institutional racism” since the institutions in question “promote racial inequality in the absence of personal racism on the part of office holders.”¹⁴ Either way, the institutional racism in question is irreducible to the racism of its members. Or as Hardimon puts it, the racism of these institutions “is found in the way in which they are organized, not the “bad attitudes” of their officers.”¹⁵ So far, I’ve only described the non-reductionist’s view. I haven’t presented any arguments for it. One such argument is best characterized as a reductio. Assume that non-reductionism about racism is false. Suppose that in a given institution, no one harbors any discriminatory attitudes against members of racialized group R as such. Yet the institution systematically disadvantages members of that racialized group. Moreover, their membership in that racialized group best explains why they are disadvantaged: the institution imposes unfair burdens on the poor, and R-members are disproportionally poor partly because of the legacy of racism against their forebearers. The fact that 1) the institution systematically disadvantages R-members, and 2) their membership in that group best explains why they are systematically disadvantaged in that way, is arguably enough to warrant the claim that they are victims of racism. (I return to this contentious claim shortly.) But, by hypothesis, no one in the institution harbors any discriminatory attitudes or feelings toward R-members as such. Given the premise that only individuals can be racist, we are forced to conclude that the members of the racialized group are not victims of racism after all, contrary to what was assumed. This suggests that we ought to abandon reductionism about racism. Or so it is argued. Haslanger accordingly points out that “most of the practices and institutions that structure our lives, although made up of individuals and influenced by

¹³ (Shelby, 2016, p. 24).

¹⁴ (Hardimon, 2020, p. 502).

¹⁵ (Hardimon, 2020, p. 503).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

212

, ,  

individuals, are not designed and controlled by anyone individually.” Rather, she says, they “are collective enterprises that are maintained through complex social conventions and cooperative strategies.” This point, that institutions “are often beyond the control of individual agents, counts against what we might call an individualistic approach to oppression.”¹⁶ If there are no racist individuals, then that leaves the institutions themselves as the racist party. “Making room for this point,” Hardimon says, “is one of the reasons for which the notion of institutional racism is introduced.”¹⁷ Though neither Hardimon nor Haslanger put it in this way, this is what’s known elsewhere as a “responsibility gap” argument for attributions of wrongdoing to entities other than persons and their actions.¹⁸ Consider the following fictional example.¹⁹ Suppose Blanc is a private commercial bank. It was founded almost two centuries ago, at a time when the chattel slavery of a particular racialized group, R, was legal. The owner of Blanc financially facilitated slavery by lending money to slave-owners. He did so not only because he profited from it, but because he harbored racist attitudes toward R-members. The owner of Blanc also permitted his clients to list their slaves as assets and to use their slaves as collateral. Call this period of Blanc’s existence “Phase 1”. After chattel slavery is abolished, but before segregation and discrimination are made illegal, Blanc’s subsequent owner instructed his employees either to refrain from doing business with R-members, or to otherwise charge greater interest rates and to demand greater collateral from them. If an R-member nonetheless managed to afford the loan under these conditions, the applicant would still be denied credit if he sought to purchase land in areas where the racially privileged members of the society reside. Again, the owner adopted these policies in part because he harbored racist attitudes toward R-members. Call this period “Phase 2”. Decades later, the state passes laws against segregation and discrimination. Blanc’s newest owner, however, covertly maintains the goal of facilitating racial segregation. To this end, he adopts a host of policies meant to prevent R-members from securing loans. In particular, the owner adopts a policy that determines interest rates according to the applicant’s neighborhood; this results in higher

¹⁶ (Haslanger, 2004, p. 104). See also Iris Marion Young, who argues that we cannot blame individuals for participating in a “structural injustice,” where a structure is an inter-institutional arrangement comprising many individuals interacting on the basis of established norms (Young, 2011, pp. 103–104). According to Young’s “social connection” theory, such participants are nonculpably “implicated” in the structural injustice by “producing and reproducing” it, as a result of which they possess forward-looking responsibility to engage in activism necessary to change the status quo (Young, 2011, p. 112). Several theorists have argued that Young understates individual culpability in structural injustice. See for example, (Reiman, 2012, pp. 747–749), (Barry & Ferracioli, 2013), (Sangiovanni, 2018), (Atenasio, 2019). ¹⁷ (Hardimon, 2020, p. 506). ¹⁸ I find such arguments unconvincing. For helpful discussion, see (Collins, 2019). ¹⁹ Iris Young uses an example roughly of this type to argue in favor of her social connection theory of responsibility. See (Young, 2011, p. 52). Joshua Glasgow develops it in (Glasgow 2009, 72–73). See also (Shelby, 2016, pp. 44–45).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

213

rates for R-members due to de facto residential segregation. The owner also adopts loan accreditation criteria requiring a substantial credit history, in part because such a criterion will disproportionally prevent R-members from qualifying for loans. In addition, the accreditation criteria automatically disqualify any felons convicted in the past decade; the owner includes such a provision in part because he knows that R-members are over-incarcerated and are thus disproportionately likely to be disqualified on those grounds. These policies, in combination with similar policies from other institutions, effectively ensure that many R-members economically stagnate in comparison to the privileged racialized group. This is Phase 3. Eventually, due to a cultural shift, racial antipathy and derogation are less prevalent. The new owner of Blanc harbors none of the covert racist goals that the previous owner did. However, the new owner does not change any of the lending criteria the previous owner authorized. That is, loans tend to be granted solely to applicants free of felonies, with good credit ratings, and from good neighborhoods. Such applicants tend to be those from economically privileged families, disproportionately few of which are R-members due to the generational effects of the overtly and covertly racist policies that the country’s institutions adopted in the past. Blanc’s lending policies, in combination with similar policies from other institutions, effectively lock R-members into a generational cycle of poverty, while the standard of living among members of the privileged racialized group rises at a disproportionately higher rate. This is Phase 4. The “racially neutral” lending policies of institutions like Blanc in Phase 4, might, in principle, be just under conditions in which members of the racialized groups in question start at roughly the same socioeconomic level. But as is, by adopting those policies, institutions like Blanc not only perpetuate but exacerbate the economic and social disadvantage that has been inflicted on R-members for generations. It seems undeniable that these R-members are victims of racism. But this racism, it might be thought, has to be located somewhere. Since, by hypothesis, neither the owner of Blanc (or any employee) is racist (insofar as she harbors no racial antipathy or disapprobation), we are left to conclude that the institution of Blanc is itself racist. Thus, reductionism about racism is mistaken (or so it is argued). Another consideration seems to lend further support to this “responsibility gap” argument for non-reductionism. Suppose we deny that Blanc, during Phase 4, is racist. We are then forced to conclude that when Blanc’s last racist owner dies or retires at the end of Phase 3, the institution “magically” transitions from racist to nonracist absent any change in the institution’s policies.²⁰ Such a view might seem implausible. This further suggests reductionism about racism is mistaken.

²⁰ (Glasgow, 2009, pp. 73–74).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

214

, ,  

Reductionists might repair to what Joshua Glasgow calls a “genetic” account of institutional racism. This view, in effect, expands the “reductionist base”—i.e., the individuals whose racism subvenes the racism of the institution. On this view, the reductionist base includes not just the institution’s current members, but its past members as well. Thus, an institution is racist if at one point in time racism motivated its members to either establish or implement policies targeting a racialized group as such, where those policies remain in force today and continue to disproportionally burden the racialized group in question. On this view, the institution remains racist even if the current members do not harbor the racism of their predecessors. Yet this view remains reductionist in that the racism of the institution ultimately derives from the racism of individuals—albeit individuals who might be long dead. But institutions like Blanc are criticizable not simply on the grounds that its previous owners, now long since dead, harbored racial antipathy. Rather, it is alleged, an institution similar to Blanc counts as racist in virtue of its current policies. We can imagine a newly established private commercial bank—call it “Azul”—that happens to adopt the very same lending policies as Blanc in Phase 4. Non-reductionists want to say not only that Azul counts as racist in and of itself, but that it does so in the same way and for the same reasons that Blanc counts as racist (though Blanc’s unjust history might mean that it owes more by way of reparations). Why not simply say that the institutions in question—such as Blanc or Azul— “perpetuate an unjust racial inequity”? Why believe that, in addition to this, they qualify as racist? Glasgow says that perpetuating an unjust racial inequity, “seems to qualify those institutions as racist.”²¹ On this view, any institution (including those that do not qualify as group-agents) that foreseeably and wrongly perpetuates an unjust racial inequity counts as racist even if the institution’s constituents are not themselves racist. Reductionists, though, adopt a more restricted ontology for racism: the property of racism can be attributed to an object only if there is an agent whose attitudes, emotions, or conduct serves as the basis for the attribution of that property. Thus, absent any racist persons, there is no racism, on this view. Yet the unjust negative effects of institutions whose members adopt raceneutral policies are so corrosive and widespread that there is, naturally, a need to find someone—or something—accountable. Insofar as this injustice is racial, the charge of racism seems warranted. And insofar as institutions perpetuate this injustice, they seem like the appropriate subjects of blame. But in what follows I argue that we can accommodate these intuitions while also respecting the reductionist’s ontological parsimony.

²¹ (Glasgow, 2009, p. 73).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

215

8.1.2 Reductionist Accounts of Racism Reductionists do not deny that institutions can be racist. They admit as much in cases where the institutional racism derives from the racism of its members. Even reductionists will agree, for example, that Jim Crow laws qualifies as a form of institutional racism. This is because the agents who established and enforced such laws were largely White supremacists whose purpose was to maintain a racist social order in the United States. Thus reductionists, like Garcia, can accommodate the theoretical category of institutional racism by maintaining that “institutional racism exists when and insofar as an institution is racist in the aims, plans, and the like that people give it, especially when their racism informs its behavior.”²² But absent such motives from the members occurrently constituting the institution, it does not qualify as racist, according to reductionists. Thus, reductionists might seem forced to admit that Blanc in Phase 4 is not a racist institution. But I believe that instead of arguing that an institution qua institution can be racist, we ought to argue that Blanc’s owner in Phase 4 harbors racist attitudes after all. If correct, the reductionist can accommodate the claim that Blanc in Phase 4 is a racist institution. By hypothesis, Blanc’s owner in Phase 4 does not aim at perpetuating racial injustice or at mistreating R-members. It seems, then, that his conduct is not racist. But we might accordingly advert to what Hardimon calls “indifference racism”. Hardimon describes this variety of racism as follows: “To be racially indifferent to Rs is to be indifferent to the harm, suffering, or disadvantage Rs experience, simply because Rs are Rs (members of a particular racialized group).”²³ But Blanc’s owner is “indifferent to the harm, suffering, or disadvantage Rs experience” not because he discounts the welfare of R-members as such, but because he denies or ignores the importance of racial justice. The owner acknowledges that his “race neutral” policies will continue to disproportionately disadvantage R-members, but he just doesn’t care. And he is indifferent not because he has anything against R-members per se but because he (wrongly) believes that so long as he is acting in a “race-neutral” way, R-members as such have no basis for complaint against him. So, by hypothesis he is indifferent toward racial injustice. But is he an indifference racist? The answer is “yes,” only if he is indifferent toward R-members because they are R-members. Consider, now, those R-members to whom he is indifferent. He doesn’t recognize (or otherwise doesn’t care) that the policies

²² (Garcia, 1996, pp. 404, cf. 416–18, 423–24). ²³ (Hardimon, 2020, p. 506). See also the concept of ‘laissez-faire racism’ (Bobo, Kluegel, & Ryan, 1997). Jorge Garcia has long maintained that “racial indifference” qualifies as a form of racism. See (1996), (1997), (1999).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

216

, ,  

he adopts exacerbates racial injustice. Suppose, counterfactually, that those R-members weren’t R-members. Imagine they were instead members of the privileged racialized group (call them “P-members”). The policies the owner adopts wouldn’t count as an instance of racial injustice against the P-members. And absent any such racial injustice, there would not (a fortiori) be any injustice for the owner to be indifferent toward in the first place. The point here is that the status of R-members as R-members is indeed what explains at least part of the owner’s indifference toward them. We can even stipulate that the income-bracket of the R-members is the same as the P-members we’re imagining in their stead. That is, suppose the P-members who counterfactually replace the R-members are equally impoverished. In such a case, the owner’s policies might still very well exacerbate economic injustice. But whatever injustices the owner’s policies might inflict, they won’t inflict racial injustice since by hypothesis the victims are P-members and not R-members. So, it follows that the owner is indifferent toward R-members at least partly because they are R-members. One might respond by alleging that I’ve cheated. Though the owner is indifferent toward R-members insofar as he is indifferent toward racial injustice, the owner cannot be guilty of indifference-racism or any other racism because, by hypothesis, the owner sincerely denies that R-members qua R-members are less important or less worthy. But, at the same time, the owner wouldn’t be indifferent toward R-members—or, he would be less indifferent toward R-members—if they weren’t R-members. Given this, the proper conclusion is that whether an individual counts as an indifference-racist depends not on whether she means to single out a racialized group but on whether she actually does so. The two can come apart when she is ignorant of or otherwise ignores the moral relevance of racial justice, as exemplified in the case of Blanc’s Phase 4 owner. So, for example, in 1971 the U.S. Supreme Court considered a case in which Blacks were routinely disqualified for jobs in which they failed to pass tests unrelated to successful job performance. The Court found that “practices, procedures, or tests neutral on their face, and even neutral in terms of intent, cannot be maintained if they operate to freeze the status quo of prior discriminatory practices” (my emphasis).²⁴ On the view that that I have argued for here, such tests would count as racist even if they are “neutral in terms of intent.” Or more to the point, the individuals who authorize such tests as a basis for hiring would be guilty of indifference-racism. Or consider child welfare policies in the US. Alexander Roberts has noted that the economic status of a family rather than its racialized group is what best explains the disparity in the likelihood that child welfare services will intervene

²⁴ (Griggs v. Duke Power Co. 401 US 424) (my emphasis).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

217

in that family. However, a history of racial injustice is at least partly what explains the disparity in economic status between Black and White families.²⁵ Thus, the legacy of historical racism partly explains racial inequality in the child welfare system. If what I have argued is correct, child welfare policies, insofar as they tend to further disadvantage the poor, qualify as indifference-racist, even though such policies are not designed to target Black families as such. Again, the individuals who authorize these child welfare policies would be guilty of indifference-racism. Some theorists understandably worry that we engage in “conceptual inflation” when we classify conduct racially neutral in its intent as a species of racism. They argue that an agent’s behavior is racist only if the agent means to single out members of a racialized group as such.²⁶ Likewise, on this view, policies are racist only if they are adopted “because of ideas connected to the race of its targets.”²⁷ Some argue for this less permissive analysis of “racism” on the grounds that the term invariably expresses severe disapprobation.²⁸ They take this to be a normative characteristic of “racist” in that it should only be used to expresses severe disapprobation.²⁹ Those who argue for the more permissive analysis of “racism” respond by pointing out that even if calling someone a “racist” expresses severe disapprobation, there is a difference between attributing the property of racism to a person sans phrase, and attributing the property of racism to a trait the person possesses.³⁰ We might claim that someone has racist traits without thereby implying that she is a racist.³¹ In light of this, it seems to me appropriate to say that Blanc’s owner possesses an indifference-racist trait insofar as he wrongly ignores the racial inequality that his unjust policies promote. I conclude, then, that reductionists can felicitously claim that institutions like Blanc (in Phase 4) are racist. More specifically, they can claim that Blanc exhibits racial indifference toward R-members. This claim is made true in virtue of the racial indifference that Blanc’s owner exhibits. Racial indifference toward R-members is not as bad as, say, racial antipathy toward R-members. We might, accordingly, argue that racial indifference does not rise to the level of ordinary racism. Blanc’s owner does not intend to adopt policies that single out R-members as such. But he ignores or doesn’t care about racial justice. As a result, he not only perpetuates and exacerbates racial injustice in that society but does so in a way that warrants the attribution of “indifference racist” toward the attitude in ²⁵ (Roberts, 2002). ²⁶ (Anderson, 2010, pp. 47–48). ²⁷ (Anderson, 2010, pp. 60–61). ²⁸ (Blum, 2010, pp. 20, 31). ²⁹ For helpful discussion of this issue, see (Hardimon, 2019). Some also argue that the more permissive analysis risks “over-using” ‘racism’ in a way that will diminish both its moral force and concomitant concern about race-related wrongs. See (Blum, 2004). ³⁰ Those who argue for the more permissive stance include (Shelby, 2003), (Shelby, 2014), (Hardimon, 2019). ³¹ (Blum, 2010, p. 15). As Hardimon puts it, there is “a step from saying that something someone has said or done is racist to saying that the person herself is a racist” (Hardimon, 2019, p. 235).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

218

, ,  

question—or so I’ve argued. An upshot is that the reductionist can have her cake and eat it too in that she can claim both that a) an institution is racist only insofar as its individual members are racist, and that b) an institution can be racist even if none of its members harbor any racial animus. So far, I’ve analyzed Blanc’s institutional racism in terms of the indifferenceracism of its owner. As mentioned at the outset, this is a simplifying assumption. Obviously, an institution like Blanc will include many individuals in positions of authority who, in establishing and implementing policy, might exhibit indifference-racism as well. I return to this issue shortly.

8.2 Accountability for Institutional Racism I argued that reductionists can make sense of this claim: institutions devoid of members harboring racial animus can still be racist. I did so by analyzing the racism of such institutions in terms of the indifference of those in charge of the institution. Consider again Blanc in Phase 4. What policies should the owner have adopted? There are various possibilities. The owner might end the policy of automatically denying loans to convicted felons. The owner might also rescind the policy of indexing interest rates to the applicant’s residential neighborhood. Alternatively, the owner might fund a program providing free financial management services to R-members residing in economically marginalized communities— especially to those who were denied a loan. There are many other possibilities. The owner’s failure to avail himself of them reveals an attitude of racial indifference— which, I have argued, is a racist trait even if the owner doesn’t mean to single out R-members as such.³² This argument, in combination with the argument for authority-based accountability, has significant implications when it comes to accountability for an institution’s racism. In what follows, I consider whether the racial indifference of the employers inculpate employees.

8.2.1 Implicated Employees Suppose Employee is a loan management associate at Blanc. She is thereby tasked with implementing her institution’s loan accreditation policies, set in place by Owner. Employee recognizes and laments the injustice of the loan accreditation criteria she is instructed to enforce. She does not share Owner’s racial indifference toward R-members. Yet, as others have put it, individuals like her are still

³² This isn’t to say that the owner is necessarily accountable for harboring that racist trait. Some, such as Hardimon (Hardimon, 2020, p. 505) argue that racist ideology excuses. Others are more skeptical. See, for example (Mills C., 2003, pp. 60–61).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

219

“implicated” in racism or are “agents of racism”.³³ Or, as Haslanger says, “we are each complicit in the perpetuation of unjust structures, practices and institutions.”³⁴ Because Owner perpetuates and exacerbates racial injustice via Employee’s agency, Employee should rightly feel something like agent-regret for the role that she plays. This view is plausible enough, at first glance. The problem, though, is it presumes the following contentious claim: we have reason to avoid enacting the ends that others have when those ends are made wrongful solely in virtue of their attitudes. To better elucidate the claim at issue, consider the following example. Racist Detainment Police Officer, who is in a patrol car with Sergeant, pulls over a car in response to a minor traffic violation. During the stop, Driver, who is Black, behaves suspiciously. Driver’s behavior warrants detaining him. Sergeant orders Police Officer to do so. But Sergeant issues the order partly on the grounds that he believes Black persons are qua Black persons more inclined to criminal conduct. Police Officer, who obeys the order, harbors no racist attitudes. Police Officer’s act, extensionally described, is (by hypothesis) perfectly permissible. But given Sergeant’s racist reasoning, does Police Officer mistreat the driver by implementing Sergeant’s orders? This is the sort of question I want to address, as it applies to employees in institutions who are tasked with implementing policies exhibiting the racial indifference of their employers. Does an employer’s racial indifference affect the moral valence of what the employee does, in cases where a) the employee lacks the employer’s racial indifference, and b) what the employee does is perfectly permissible when described extensionally? Put metaphorically, the point of this inquiry is to determine whether a supervisor’s racist attitudes “infect” an otherwise innocent subordinate—and whether the subordinate is at all accountable for that “infection”. Return now to Blanc. Recall that in Phase 4, Owner adopts loan accreditation policies exhibiting racial indifference toward R-members. In particular, Owner adopts the policy of denying small business loans to anyone convicted of a felony within the past ten years. Such a policy is unjust on the grounds that R-members are unjustly over-incarcerated. Denying small business loans to felons not only compounds the injustice that many of them have already suffered, but also denies lower-income communities the commercial businesses that they might have otherwise successfully established. Recall that Owner retains this policy not because he harbors any discriminatory attitudes toward R-members, but because, statistically, convicted felons are more likely to default on their loans. He could, alternatively, adopt loan accreditation

³³ See (Hardimon, 2020, p. 505).

³⁴ (Haslanger, 2004, p. 12).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

220

, ,  

criteria that do not automatically deny loans to all felons; a more nuanced policy would consider the nature and potential relevance of the felony. Owner’s failure to do so is evidence (albeit defeasible evidence) of racial indifference toward R-members. Suppose that Applicant—an R-member—meets with Employee. Applicant would like to apply for a small business loan. He is, however, a convicted felon, recently released from federal prison. On these grounds, he is automatically denied the loan. Despite the injustice of that policy, it so happens in this particular case that by all rights Applicant should not be granted the loan since he was convicted specifically of committing repeated acts of loan fraud. In implementing Owner’s accreditation policies, does Employee mistreat Applicant? In this particular case, Owner’s morally unjust accreditation policies accidentally “gets it right.” This is analogous to the example Racist Detainment in which Sergeant’s order to detain Driver, though motivated by racist beliefs, accidentally “gets it right” in that Driver should indeed be detained. Of course, Sergeant’s attitude is racist in that it discriminates on the basis of race, whereas Owner’s attitude is racist in that it exhibits racial indifference. But in both cases, the issue is whether a superior’s racist attitudes “taint” the actions of a subordinate. Given the executor’s mental states, the acts in question are perfectly permissible. But given the deliberator’s mental states, they are not. So, what is the executor to do? This issue can be generalized. Sometimes we work in institutions whose leaders adopt racist policies toward R-members. Suppose we are tasked with implementing those racist policies. Does this in and of itself mean we mistreat R-members, even on those occasions when we treat them exactly the way they would have been treated had the owners adopted morally unproblematic policies? I claim that we do indeed: we commit a directed wrong against the member of R. If this is correct, the racial indifference of the employers in racist institutions inculpates the employees in these institutions. To be clear, the point is not simply that employees are, absent excuses, accountable for the wrongs they commit. (That point verges on the trivial.) Rather, the issue is what counts as a wrong they commit. On the view I defend, what non-racist employees do is made wrong (or at least pro tanto wrong) by the racist attitudes of their employers. I turn to a defense of this view next.

8.2.2 Relevance of Others’ Intentions We might think that Employee does not herself wrong Applicant. After all, by hypothesis, Applicant should not be granted the loan. And by hypothesis Employee’s reasons for denying the loan do not exhibit racial indifference. Rather, the problematic reasons belong to Owner. Though Employee

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

221

serves as the means by which Owner satisfies his problematic intentions, this in itself does not inculpate Employee if what she does is otherwise permissible. There are reasons for thinking that intentions are only first-personally and not third-personally morally relevant to conduct.³⁵ On this view, intentions are relevant only from an agent-relative standpoint. If so, Owner’s bad intentions, in and of themselves, ought not to play a role in Employee’s moral decisionmaking. Given her situation, it is best for Employee to deny Applicant the loan, even though Owner’s reasons for doing so are bad; the fact that those reasons are bad do not even provide a pro tanto consideration in Employee’s deliberations. On this view, when we deliberate, we are required to take into consideration what others will do in response to the decisions we make. But we need not take into consideration the motivating reasons why they do it. Though it is incumbent upon each of us to act for the right sort of motivating reasons, that requirement is wholly first-personal in that no one else has the responsibility to ensure that we adopt those reasons. The desirable features of adopting the right sort of motivation provide you with a reason to do so, and they would provide each of us with that reason were we in your shoes. But they do not provide each of us with reasons to promote that outcome in you. The point can be put differently: all things being equal, you have a reason to prevent me from committing harmful rights-violations, but you don’t have a reason to favor preventing me from committing culpable rights-violations over non-culpable (or less culpable) rights-violations, assuming we hold fixed the amount of harm each rights-violation inflicts. To see why, consider the following example. Mountain Pass We are delivering emergency medical supplies through a mountain pass to a remote village; without them, thousands will die. There are two paths you can take. Each path takes us near a hermit. The hermit on the first path, in response to our presence, will rush home and slam his door, since he is unused to other people. By doing so he will unintentionally cause an avalanche that will kill one random innocent below and will break another innocent’s leg. The hermit on the second path will respond to our presence with anger and frustration rather than fear; he will slam his door with the murderous intention of causing an avalanche which he knows will kill an innocent below. If we have reason not only to prevent the loss of life and limb, but also to prevent others from acting on malicious reasons, then we have to weigh the badness of a) facilitating an intentional killing against the badness of b) facilitating an ³⁵ (Nagel, 1989, pp. 175–179). See also section 2.3.1, chapter 2, in which I address the possibility that intentions are deontically irrelevant.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

222

, ,  

unintentional killing and a broken leg. Depending on how heavily we weigh the reasons against facilitating an intentional killing, we might have to choose the path that causes more harm. This is a difficult conclusion to accept. It seems more plausible to maintain that in deciding what to do, we should take into consideration only the harms the hermit causes. If I am correct in surmising that the practical reasons we take ourselves to have are morally relevant only from an agent-relative standpoint, then we’re not accountable for the motivating reasons why others do what they do. Hence, the fact that Employee is enacting Owner’s racist intentions provides no reason—not even a pro tanto one—against doing so, if Employee’s conduct is otherwise permissible (which, by hypothesis, it is). This argument is, I think, the best reason for thinking that Employee does not mistreat Applicant. But if intentions are indeed morally relevant from a firstpersonal standpoint, then the features explaining why this is true will also provide reasons for thinking Employee is indeed accountable for enabling Owner’s bad intentions, after all. I turn to this argument next.

8.2.3 Modal Tracking and Authority-Based Accountability Consider a terror-bomber who intentionally targets innocent villagers, in contrast to a strategic bomber who collaterally kills the villagers as a side-effect of destroying a munitions factory. If different conduct were necessary to kill the civilians, the terror-bomber, unlike the strategic bomber, would alter his behavior accordingly. On Thomas Nagel’s view, this situation is morally worse. Here, the agent modally “tracks” the harm—where the conditions for causing the harm change, his actions change accordingly. This robust counterfactual interdependence between his agency and the harm in question relates him to that harm in a morally inimical way.³⁶ This, of course, is just a standard counterfactual description of an ordinary intention. Similarly, Employee does not just happen to enact Owner’s practical reasons. Instead, it is quite literally her function to do so. Recall from chapter 1 that according to the argument for authority-based accountability, Employee and the owners of Blanc have established a division of agential labor in which, at a suitably broad level of generalization, Owner counts as the deliberator and Employee counts as an executor. In accordance with this division of agential labor, Owner establishes the purpose of Employee’s actions with respect to her role as a loan management associate. Employee has, then, the agentive function of enacting the practical reasons Owner takes there to be. Though Employee disavows Owner’s

³⁶ (Nagel, 1989, pp. 175–188).

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

223

practical reasons, Employee’s conduct still counterfactually depends on Owner’s practical reasons in a systematic way: where Owner’s reasons change, Employee’s purpose ipso facto changes as well (at least within the remit of the Employee’s job as loan management associate). The counterfactual sensitivity that Employee’s conduct bears to Owner’s ends, is the same counterfactual sensitivity that an agent’s actions bear toward her own ends. Similar to how, in ordinary cases, an intention disposes you to change your actions (within limits) in ways instrumental to achieving its object, Employee’s relationship to Owner disposes Employee to change her actions in ways instrumental to achieving the object of Owner’s intentions. This is because the reasons Owner takes there to be, determine the Employer’s protected reasons which constitutively determine the purpose of her conduct. The result is that no matter how sincerely Employer disavows racial indifference toward Applicant, her purpose is nonetheless to enact the aims of someone who does indeed harbor racial indifference. This should be no surprise: after all, Owner and Employee have established a division of agential labor in which Owner counts as a deliberator and Employee counts as an executor. As a result, Employee does not unilaterally determine the purpose of what she does. So, if we think that intentions are first-personally relevant because they counterfactually relate the individual to an end in a systemically robust way, then we should also think that Owner’s reasons are morally relevant to Employee’s actions. This doesn’t mean that Employee should grant the loan. Instead, the fact that Employee is enacting Owner’s racist intentions provides a pro tanto reason against doing so, even if, typically, intentions have only first-personal moral relevance. An upshot is that even if Employee is herself free of racism, and even if Employee makes the correct decision in denying Applicant the loan, and even if intentions are in general morally relevant only first-personally, Employee still mistreats Applicant by serving as the means by which Owner enacts his racist intentions, in that there is a defeasible reason against denying the loan. I take this to be a welcomed conclusion. There is, intuitively, a sense in which Owner’s bad motivating reasons “taint” Employee’s treatment of Applicant, even though Employee’s treatment of Applicant is in and of itself unproblematic. The argument for authority-based accountability provides a means of making sense of this intuition. If Applicant wants to know Employee’s purpose in implementing the loan accreditation criteria, she will have to refer him to Owner’s motivating reasons in favor of adopting those criteria. Recall from chapter 1 that Employee’s reasons in favor of implementing the criteria are protected; as a result, those reasons refer anaphorically to the practical reasons that Owner takes there to be in favor of implementing them. So, though Employee acted on good reasons in implementing the criteria, Blanc’s reasons for authorizing those criteria were racist insofar as they exhibit contingent racial indifference. Owner is accountable for constitutively determining the purpose of Employee’s conduct—a purpose

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

224

, ,  

exhibiting Owner’s racial indifference. But, in addition, Employee is accountable for providing Owner with the means for doing so. Recall that establishing a division of agential labor requires attributing an active agential function which is only possible if the executor agrees to play that role.³⁷ Though Employee is a mere executor in a division of agential labor, she volunteered to play that role, and is thus accountable for the foreseeable consequences of that decision, as I suggested in section 4.3. The general moral here is that it is possible for an executor (like Employee) to be accountable for the wrong that a deliberator (like Owner) commits, where that consists in endowing the executor’s conduct with a morally problematic purpose (such as a purpose manifesting Owner’s racial indifference). To be sure, Employee will not be as accountable as Owner. There are at least two reasons for this. First, Employee might have little choice but to work at an institution where those in charge harbor racial indifference, since such institutions are ubiquitous. It is characteristic of institutional racism that it is virtually inescapable. This accordingly diminishes her accountability. Second, I argued that the functional relationship between Employee’s conduct and Owner’s ends suggests that even if Employee lacks Owner’s morally problematic intentions, they are still relevant to our evaluation of Employee’s conduct. But unlike Owner, Employee does not personally endorse those ends. This does not eliminate her accountability for the racial indifference her conduct vicariously exhibits, but it does suggest that the Owner is more accountable than Employee.

8.2.4 Alternative Explanations of Employee’s Wrongdoing I have argued that even if intentions have only agent-relative moral relevance, a deliberator’s intentions in a division of agential labor can affect the moral valence of an executor’s actions. The result is that Employee commits a pro tanto wrong in denying Applicant the loan, even though, absent Owner’s racial indifference, denying the loan would be unproblematic. To be clear, none of this is to say that Employee ought to deny the loan. Again, the point is that she commits a pro tanto wrong in doing so in that it mistreats Applicant. There might, though, be other reasons for thinking that Employee mistreats Applicant by denying the loan. We might think that though Employee makes the right decision in denying Applicant the loan, she nonetheless commits a wrong by remaining employed in, and by implementing the racist policies of, a racist institution. She is causally contributing, albeit in a small way, to racial injustice. This, in turn, is what wrongs Applicant. This account is compelling in its

³⁷ See section 1.1.2, chapter 1, for further discussion.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

225

simplicity. There are, though, two ways we might criticize it. The first fails whereas the second succeeds. According to the first criticism, Employee has little choice but to work at that institution or one like it; we can imagine that quitting her job in protest while avoiding employment at institutions whose owners exhibit racial indifferent will impose significant hardships on Employee. Given these circumstances, Employee is permitted to remain employed at that institution, on the grounds that she has an agent-relative permission to give extra weight to her own welfare. Since she is permitted to do what she is doing, she does not wrong Applicant. Or so it might be argued. The circumstances Employee faces, however, do not show that she isn’t mistreating Applicant. Rather, they only show that mistreating him might be a mere pro tanto wrong given the alternatives. A pro tanto wrong still qualifies as a wrong. Even if Employee is, all things considered, permitted to wrong Applicant, she still thereby wrongs Applicant. (Against this it might be argued that agentrelative reasons override rather than merely outweigh competing considerations. The result would be that Employee does not commit even a pro tanto wrong. I think this view of agent-relative reasons is a mistake; addressing it, though, would take us too far afield.) The second criticism is more effective: if the basis of Applicant’s complaint is that Employee is causally contributing to racial injustice, then Applicant’s complaint is no different from the complaint that any other member of R might have against Employee—including those R-members who never applied for and were consequently never denied a loan. The proposed basis for Applicant’s complaint fails to accommodate the directed nature of the wrong committed against Applicant. Put differently, Applicant has no basis for complaint qua applicant. This point is subtle but important. R-members who are not denied loans have a basis for complaint against Blanc—but only as morally wronged bystanders to the primary victim. To use a vivid example of this phenomenon, suppose a member of R is the victim of a racially motivated instance of police brutality. The direct victim is of course the brutalized individual. The indirect victims are all the other R-members in the community.³⁸ If we think that the basis of Applicant’s complaint against Employee is simply that she causally contributes to racial injustice against R-members in general, then we illicitly treat Applicant as an indirect victim rather than as a direct victim of the wrong in question. The upshot is that we cannot accommodate Applicant status as a direct victim of Employee’s conduct by adverting to the racial injustice that Employee promotes. Employee wrongs Applicant qua member of R-who-applied-for-a-loan-at-Blanc, rather than just qua member of R. ³⁸ See (Hull, 2015) for further discussion of the distinction between direct and indirect victims of racism.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

226

, ,  

Another attempt to explain Employee’s wrongdoing appeals to circumstantial moral luck. By hypothesis, Employee makes the right decision in denying him the loan. But we might think that it is only due to circumstantial moral luck that this counts as the right decision. By hypothesis, if Applicant’s felony hadn’t involved loan fraud, he would still have been denied the loan on the grounds that he is a convicted felon. Since denying him a loan on these grounds wrongs him, it is only due to circumstantial moral luck that he is not wronged. This might serve as a basis for thinking that what Employee actually does is morally criticizeable; though she actually makes the right decision, Employee is nonetheless disposed to make the wrong one. That is to say, she lacks a robust disposition to treat loan applicants in a way that consistently refrains from exhibiting racial indifference. Applicant can, accordingly, argue that there is a straightforward sense in which Employee mistreated him in that it was only due to luck that Employee happened to make the right decision regarding the loan. But even if circumstantial luck is morally relevant—which is a big “if”—the resulting explanation of Employer’s wrongdoing would still fail to distinguish between Applicant and other R-members in that community, including those who did not need and never applied for a loan in the first place. There are presumably among them some who are convicted felons whom Employee would have denied a loan on those grounds if they had applied. It is only a matter of circumstance that they didn’t apply. If what grounds the claim that Employee mistreats Applicant is the relevance of circumstantial moral luck, then we are once again unable to morally distinguish Employee’s direct victims—such as Applicant—from her indirect victims—such as others in the community. Appealing to circumstantial moral luck, then, does not help us show that Employee wrongs Applicant qua applicant. The upshot is that if we are inclined to think that Employee has mistreated Applicant, the argument for authoritybased accountability provides the best explanation why.

8.3 Conclusion I have focused on a single case of a loan management officer working in a private commercial bank. The example is highly schematized, in that the agential division of labor involves just two persons: the owner of the bank, and the loan management officer. Real cases involve many individuals in complex organizational relationships. For most private banks there is typically more than one owner; these partners together control the bank which typically comprises three divisions: lending, retail, and trust. The bank will also include various corporate management, operations, and administrative departments and sub-committees, as well as semi-autonomous groups such as internal auditors. The organizational relationships become even more complex given a board of directors, a chief executive

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

 

227

officer and (in a public bank) shareholders. But no matter how byzantine the agential division of labor, authority-based accountability follows the rules outlined in chapters 1 and 2. I abstracted from the complexities of the organizational characteristics of private commercial banks to keep the discussion and analysis of authority-based accountability tractable. A further advantage of abstracting from the specific institutional peculiarities of a private commercial bank is that the ensuing discussion of institutional racism is generalizable to a variety of other institutions—medical, penal, educational, religious, etc. The circumstances that the loan management officer faces will be all too familiar to individuals across institutions: an employer tasks an employee with enacting policies that have the side-effect of perpetuating injustice toward members of an historically oppressed racialized group. The employer’s instructions reveal her racial indifference toward members of an historically oppressed racialized group. But the employee has a protected practical reason to comply with those instructions. So, even if the employee’s practical reasons are morally benign, her conduct has the purpose of implementing the employer’s practical reasons, by virtue of the division of agential labor they have established. The employer’s wrongful attitude toward members of a racialized group can morally “infect” what the employee does—even in cases where what the employee does would have been permissible when described extensionally. This does not mean that what the employee does is impermissible; rather, it means that the employee mistreats the victim, albeit in a way that might be justifiable all things considered. Either way, the employer is accountable for “infecting” the employee’s conduct, by fixing the content of the practical reasons to which the employer’s protected reasons anaphorically refer. But most employees agree to serve as an executor in a division of agential labor with the employer. If it is foreseeable that this means enacting the employer’s problematic ends—viz., ends that manifest the employer’s contingent racial indifference toward members of an historically oppressed racialized group—then the employee is also accountable for the purpose she has qua employee (absent any excusing conditions). Continuing the septic metaphor, the employee effectively volunteered to become “infected” and accordingly bears some accountability for doing so (though, as I argued, this is not to say that she is as accountable as the employer). Moreover, the infection might not spread from the topmost rung of a hierarchically organized collective; we can imagine a case where the individual or individuals in charge of an institution are free of racist attitudes, but a particular mid-level manager is not. In this case, the manager’s racist attitudes would potentially “infect” the subordinates working under her. Likewise, the “infection” can spread laterally.³⁹ An agent harboring racist attitudes, working with otherwise

³⁹ See the Appendix for more on this.

OUP CORRECTED PROOF – FINAL, 22/7/2022, SPi

228

, ,  

innocent colleagues in a project where no one has unilateral authority over anyone else, might thereby “infect” her colleagues by way of the same mechanism described in this chapter. The moral here is the that racist attitudes subvening an institution’s racism spreads easily, in a variety of ways, and in a variety of contexts. I have assumed in my discussion of the racist financial policies the bank implements, that the racism must be located either in the bank qua bank, or in its personnel—i.e., either in the bank as an institution or in the individuals it comprises. But there is a third possibility: there might be other aspects of society— such as de facto residential segregation—that partly explains why the bank’s policies result in further injustice. On this view, what needs to change is not the bank’s lending policies, but rather the aspects of society in virtue of which those lending policies further injustice. I am, though, quite skeptical of such a view. It might be true that de facto residential segregation is partly what makes the bank’s lending practices racially problematic. But this doesn’t let the bank off the hook; it has a duty to adopt lending practices sensitive to the moral relevance of de facto residential segregation. I hope to have shown that what morally matters when we participate in racist institutions is not just the harm that we do or don’t do, but the role we play in enacting the morally problematic intentions of our employers—whether those intentions manifest racial indifference or outright unapologetic racial hostility. Even if, as employees, we tread carefully, thereby managing to avoid personally harming members of the racialized group in question, our own personal moral regard and respect for the racialized group in question does not absolve us. We nonetheless commit a pro tanto wrong against them by effectively allowing our employers to enact their intentions vicariously through us.

OUP CORRECTED PROOF – FINAL, 16/7/2022, SPi

Conclusion I have argued, over the course of this book, that we routinely accept the role of enacting the ends that others have, and that we routinely assign to others the role of enacting our ends. In such cases, we “outsource” the deliberative and executory aspects of our agency to others. If you and I establish a division of agential labor in which you serve as the deliberator and I serve as the executor, I have the function of enacting your ends, and your ends have the function of guiding my actions. In such a case, you confer upon me a protected reason to do as you say, where that protected reason is supposed to “settle the matter” for me. Your ends constitutively determine the object of my protected reasons, thereby determining the purpose of what I do. Though the machinery here is complex, the relationships it describes are implicit in a host of pedestrian practices, including promising, requesting, and demanding—and in shared action generally. I have argued that divisions of agential labor occur more often, and that their moral implications are more radical, than we have recognized. When you serve as a deliberator and I serve as an executor, you thereby constitutively determine the purpose of what I do qua executor. If what I do is wrongful, my victim can demand that I explain why I wronged her; after divulging my own reasons, I will have to refer her to you, since your reasons constitutively determine the purpose behind what I did. Where that purpose is wrongful, you are thereby accountable for a wrong-making feature of what I did: the wrongful purpose for which it was done. Hence, you are accountable for the fact that what I did counts as a wrong. I called this “authority-based accountability”. The dyadic relationship between a deliberator and an executor serves as a building block for more complex cases of cooperative action. The authoritybased accountability such relationships yield, provides a way to hold an individual to account for more that her own causal contribution to a cooperatively committed harm. The result is that an individual who contributes negligibly to a cooperatively committed harm can be accountable for more than that contribution. If each cooperant in that collectively committed harm agrees to contribute, then each serves at once as both a deliberator and as an executor for every other contributor; the result is that each bears authority-based accountability for what every other contributor does. Authority-based accountability has moral implications for a host of issues in legal theory and in applied ethics. In legal theory, authority-based accountability helps us make sense of accomplice liability and of respondeat superior. And in

Authority, Cooperation, and Accountability. Saba Bazargan-Forward, Oxford University Press. © Saba Bazargan-Forward 2022. DOI: 10.1093/oso/9780192862419.003.0010

OUP CORRECTED PROOF – FINAL, 16/7/2022, SPi

230

, ,  

applied ethics, it helps us make sense of revisionism in war ethics, and of individual accountability for institutional racism. The range of applications is far and wide. I’ve only scratched the surface here. For example, it is easy to imagine chapters on the authority-based accountability of individuals working for defense contractors, oil companies, private prisons, pernicious political parties, and the like. Even if such individuals causally contribute very little or nothing to the wrongs that these groups cause, they can still end up on the hook at the bar of authority-based accountability. But rather than further explore the implications of authority-based accountability or reiterate the principle claims and conclusions of this book, I will end with some remarks about the connection between authority-based accountability and integrity. In a famous example belonging to Bernard Williams, a pacifist named “George” is considering a job manufacturing weapons of mass destruction.¹ His refusal will not only impoverish his family but will result in more harm overall since a more zealous applicant will take the job in his stead thereby producing the weapons more effectively. Yet for him to take the job would mean contributing to an end he personally finds morally abhorrent. Thus, George is faced with an ethical dilemma. Williams’s discussion of this example is less about what it is that George should do, and more about how he should decide what to do. Whatever else might be true of morality, Williams says, we cannot demand of him that he treat the violation of his deepest convictions as a value in the costs-column of a Utilitarian balance sheet. To do so would be “to alienate him in a real sense from his actions and the source of his action in his own convictions.”² This would constitute an “attack” on his integrity, since integrity requires that our deepest convictions inform our actions. This is because our deepest convictions, Williams says elsewhere, “will characteristically be what gives one’s life some meaning, and gives one some reason for living it.”³ If what I’ve argued in this book is correct, the attack on George’s integrity lies not simply in that he contributes to an end he finds morally abhorrent. The attack on George’s integrity also lies in accepting the purpose of contributing to such an end. According to Sophie-Grace Chappell, “[a]n agent’s integrity, in Williams’ sense, is his ability to originate actions, to further his own initiatives, purposes or concerns, and thus to be something more than a conduit for the furtherance of others’ initiatives, purposes or concerns . . . .” Of course, Williams focused on the “initiatives, purposes, or concerns” deriving from the impartial point of view that Utilitarianism demands of us. But the point Williams made is generalizable. For George, “his projects and his decisions have to be seen as the actions and decisions

¹ (Smart & Williams, 1973, pp. 97–99). ³ (Williams, 1995, pp. 169–170).

² (Smart & Williams, 1973, pp. 116–117).

OUP CORRECTED PROOF – FINAL, 16/7/2022, SPi



231

which flow from the projects and attitudes with which he is most closely identified.”⁴ By taking on the job in which he serves as an executor and his employer servers as a deliberator, George has relegated himself to the role of “a conduit” in furtherance of “initiatives, purposes, or concerns” antithetical to his own. His employer, after all, confers upon George protected reasons to comply with his instructions; inasmuch, George is supposed to refrain from weighing the pros and cons of fulfilling the tasks assigned to him. He is thereby ceding deliberation regarding a critical issue to unscrupulous opportunists. His deepest moral convictions demand the opposite: that he grant his tasks precisely the sort of critical moral deliberation that his role as executor proscribes. This is, in and of itself, a violation of his integrity quite apart from whether it results in causal contributions to an unjust end. The upshot is that the divisions of agential labor in which we participate are crucial not just in assessing accountability for cooperatively committed harms, but in reckoning with the demands of integrity. Your decision to function as an executor is perfectly compatible with the demands of integrity when the deliberator in that division of agential labor furnishes a purpose compatible with your convictions. In such a case, the deliberator’s practical reasons—which constitutively determine your purpose qua executor—cohere with your convictions. But when your convictions and your deliberator’s practical reasons fail to cohere in this way, your decision to nonetheless serve as an executor undercuts your integrity. In such cases, by serving as a “conduit” for another’s “projects and decisions” you risk violating your integrity. Divisions of agential labor, then, have moral implications beyond authoritybased accountability. Indeed, given their prevalence, I believe divisions of agential labor play a crucial role in the moral analysis of social activity. I hope to have shown over the course of the book that what matters morally is not just our causal contributions to what we together do; the purposes we confer upon one another matter as well. We cannot make moral sense of cooperation without them.

⁴ (Smart & Williams, 1973, pp. 116–117) (emphasis in original).

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

Appendix The simplest authority-relation possible is one in which a single individual has authority over another. Here I investigate how this simple, dyadic relationship can serve as a building block for organizing progressively complex cooperative arrangements. In doing so, I investigate whether and how authority-based accountability is distributed in these cooperative arrangements (absent excusing conditions). Cooperative action can be organized, formally, in all sorts of ways. For example, some are organized so that the decisions are made hierarchically. Others are organized in a more egalitarian fashion. Still others combine these two traits. How do these various formal arrangements affect authority-based accountability? In what follows I will assume that the deliberators in the molecular arrangements I discuss all harbor one and the same end. I make this simplifying assumption in order to keep the discussion tractable.

Vertical Arrangements I will use an arrow (displayed horizontally, vertically, or diagonally) to represent an authority-relation. For example:

AΩ ↑ BΩ

1

This means person B has vested authority in person A with respect to end Ω, and person A has accepted that authority. The result is that as far as the two of them are concerned, A has authority over B with respect to that end. So, if B promises to do Ω for A, and A accepts that promise, then (1) represents that relationship. Often, though, we need or otherwise accept the help of more than one person. Suppose person A wants Ω done. But doing Ω is a two-person job. So, A enlists both B and C to help out. The result is the following pyramidal arrangement:



↑ BΩ



2



In this case, A has authority over B and C. But neither B nor C have authority over A or over each other. Given the argument for authority-based accountability, A bears authority-based accountability for what B does and what C does, which means she bears authority-based accountability for Ω as a whole. But B bears no authority-based accountability for what C does, and C bears no authority-based accountability for what B does. In this case, B and C are normatively “cut off” from each other in that B’s practical reasons do not normatively guide C’s conduct, and C’s practical reasons do not normatively guide B’s conduct.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi



233

An example might be helpful. Suppose Ω is the act of stealing a priceless artifact broken into two halves and kept in two different locations. Accordingly, A hires B to retrieve one half of the artifact and she hires C to retrieve the other half. Given the analysis of (2), it turns out that B and C are each individually accountable only for stealing one half of the artifact, whereas A bears authority-based accountability for having stolen the artifact in toto. But suppose B’s contribution is causally necessary for C’s contribution. Imagine that Ω is the act of stealing a priceless artifact housed in a private residence. To retrieve the artifact, A hires B to neutralize the residence’s alarm system, and hires C to crack the safe in which the artifact is stored. In this case, neutralizing the alarm system is a sine qua non of cracking the safe and stealing the artifact. Inasmuch, it seems that B should bear some accountability not only for neutralizing the alarm, but also for the theft of the artifact. But she is accountable for the artifact’s theft wholly in virtue of the consequences of neutralizing the alarm system. That is, she does not bear authority-based accountability for the artifact’s theft. We can contrast this pyramidal arrangement with a nested hierarchical arrangement, in which A has authority over B, who in turn, has authority over C:







3



CΩ So, suppose B promises to do Ω for A, who accepts that promise. But B either can’t or won’t do Ω on his own. So, he asks C to do Ω, who agrees to do so. In this case, A has no authority over C. For example, suppose again that Ω is the act of stealing a priceless artifact broken into two halves. To retrieve them, A hires B. However, B informs A that retrieving one of the two halves will require the expertise of a third party, which B promises to enlist; A agrees to this arrangement. So, A has authority over B, who in turn has authority over C. That is, A’s practical reasons normatively guides B’s practical reasons, which, in turn, normatively guides C’s conduct. However, A’s authority over B and B’s authority over C does not itself entail A’s authority over C. The authority-relation is not transitive. Nonetheless, A bears authoritybased accountability for what C does, qua B’s subordinate. This is because A bears authority-based accountability for what B foreseeably does in furtherance of Ω—and one thing B foreseeably does in furtherance of Ω is to enlist the C’s assistance. So, A bears authority-based accountability for what B does in foreseeable furtherance of the promise she makes, and B bears authority-based accountability for what C does in foreseeable furtherance of the promise C makes. So, as in (2), A bears authority-based accountability for both what B does and C does. But unlike (2), so is B. This is because B is accountable for the half of the artifact she steals, and she bears authority-based accountability for the half that C steals. She is accountable for the former by committing that act herself; she is accountable for the latter in virtue of her authority over C. Note, though, that as in (2), C does not bear authority-based accountability for what B does.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

234



Horizontal Arrangements So far, I’ve considered “vertical” relationships, in which authority is unilateral among the parties involved. Consider, though, “horizontal” relationships in which authority is bilateral. In the following case, A and B have authority over one another with respect to end Ω. I will depict this using a double-sided arrow (displayed horizontally, vertically, or diagonally):

AΩ ↔ BΩ

4

For example, (4) represents the relationship between two persons who agree to move a fallen tree onto a road, which requires A to lift his end and B to lift her end. In such a case, the two parts add up to a nice mereologically felicitous whole—the moving of the fallen tree, which is what “Ω” refers to. They each bare authority-based accountability for that event. But suppose that both A and B want Ω done, and they both agree to do their parts, though this isn’t enough—they require C’s contribution as well. Consequently, B enlists C’s help. In doing so, they promise each other to do their parts in furtherance of Ω. However, A and C remain normatively apart—neither makes any agreement with the other. The following represents this arrangement:

AΩ ↔ BΩ ↔ CΩ

5

As in (3), A’s authority over B and B’s authority over C does not itself entail A’s authority over C. Likewise, C’s authority over B, and B’s authority over A, does not itself entail C’s authority over A. Nonetheless, A bears authority-based accountability for what C does, qua B’s subordinate. This is because A bears authority-based accountability for what B foreseeably does in furtherance of Ω – and one thing B foreseeably does in furtherance of Ω is to enlist C’s assistance. The result is that A bears authority-based accountability for what C does, even though A has no agreement with C. But unlike (3), the relationship A bears to B is mirrored in the relationship C bears to B. The result is that C bears authority-based accountability not only for what B does, but for what A does as well. If this is right, (5) does not differ in any morally significant way from a relationship in which A and C do indeed have authority over each other, as represented in the following:



↕ ↕





6



Here, everyone has authority over everyone with respect to their contributions to Ω. As a result, each members of this triad bears authority-based accountability for what every other member foreseeably does in furtherance of Ω. So far, I’ve focused on arrangements that are either horizontal or vertical. Hybrid arrangements, however, are also possible.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi



235

Hybrid Arrangements Consider the following pyramidal arrangement, similar to (2) but in which B and C have bilateral authority over one another:



↑ ↕





7



This is just like (2) except B promises to do her part not only for A, but for C as well; and C promises to do his part not only for A but for B as well. Assuming they accept these promises, B and C each bear authority-based accountability for what they together do, as does A. So, even though A has authority over B and C, whereas neither of them have authority over A, each of B and C, like A, bears authority-based accountability for what they together do. The moral here is this. We might have thought that when one person enlists several others to help accomplish some task, that the “boss” bears more authority-based accountability than any of the underlings alone. But this depends on whether the underlings have authority over one another—the practical authority to demand that they do their part. If they do not have such authority—as depicted in (2)—then the boss is indeed more accountable than the underlings. But if they do have such authority—as depicted in (7)— then the underlings can indeed bear as much authority-based accountability for what they together do as the boss does. In hybrid pyramidal arrangements, there is one person in charge, at the top. This is in contrast to hybrid arrangements in which power is shared between multiple persons. To best exemplify this sort of arrangement, begin again with (4):

AΩ ↔ BΩ

4

Assume, again, accomplishing Ω is a two-part task. So, A agrees to do one part and B agrees to do the other. Suppose, though, that B discovers that his task is more complex than he thought. So, he solicits C for assistance, who subsequently agrees to do what B says in furtherance of Ω. This arrangement can be represented as follows:

AΩ ↔ BΩ



8

CΩ Here, A bears authority-based accountability for what C does, for the same reason that A bears authority-based accountability for what C does in (3): though A’s authority over B and B’s authority over C does not itself entail A’s authority over C, A does indeed bear authority-based accountability for what B foreseeably does in furtherance of Ω—and one thing B foreseeably does in furtherance of Ω is to enlist C’s assistance. The result is that A bears authority-based accountability for what B and C do, as does B. But C bears no authority-based accountability since he has no authority over anyone. In (8), C is under B’s authority rather than A’s. We can imagine, though, that B had made a promise to each of them. An inverted pyramid best represents this relationship:

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

236



AΩ ↔ BΩ

9







Note, though, that authority-based accountability does not differ between (8) and (9); in both cases, A and B each bear authority-based accountability for what the three of them do in furtherance of Ω, whereas C bears authority-based accountability only for what she does. In (8), the participants started off with a horizontal arrangement, and then added to it a vertical arrangement. Suppose, though, that they started off with a vertical arrangement, as exemplified in (1):

AΩ ↑ BΩ

1

Here, A and B agree that B will act in furtherance of Ω. Suppose, again, that they discover that doing Ω will require the help of a third party. Consequently, B solicits C’s assistance:





10

↔ CΩ



Unlike (9), B promises to do her part not just to A but to C as well. This is in contrast to (9) in which B had no obligations to C. So, unlike (9), in which C was accountable only for her own actions, in (10), C bears authority-based accountability for what B does as well in furtherance of Ω. (In both cases, though, A bears authority-based accountability for what both of B and C do). The arrangements discussed so far serve as base-cases for analyzing hybrid arrangements, no matter how complex. From these atoms, we can build molecules of arbitrary complexity. We can then determine who bears authority-based accountability for what, by adverting to the rules governing how authority-based accountability disseminates within these molecules. Consider, for example, this arrangement:



↑ BΩ





11























In this case, A bears authority-based accountability for what everyone else does in foreseeable furtherance of Ω. C bears authority-based accountability for what she does, as well as what E, F, G, and H do, but not for what A, B, or D do. The only individual who bears no authority-based accountability in (11) is D.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

Bibliography Alonso, F. (2009). Shared Intention, Reliance, and Interpersonal Obligations. Ethics, 119, 444–475. Altham, J. (1985). Wicked Promises. In I. Hacking (Ed.), Exercises in Analysis (pp. 1–21). Cambridge: Cambridge University Press. Anderson, E. (1996). Reasons, Attitudes, and Values: Replies to Sturgeon and Piper. Ethics, 106(3), 538–544. Anderson, E. (2010). The Imperative of Integration. Princeton: Princeton University Press. Anscombe, G. (1978). Rights, Rules, and Promises. Midwest Studies in Philosophy, 3, 318–323. Anscombe, G. (1981). Ethics, Religion and Politics. Oxford: Blackwell. Arruda, C. (2020). Commitments in Collective Responsibility. In S. Bazargan-Forward, & D. Tollefsen (Eds.), The Routledge Handbook of Collective Responsibility (pp. 184–201). New York: Routledge. Arthur, J. (2007). Race, Equality, and the Burdens of History. Cambridge: Cambridge University Press. Atenasio, D. (2019). Blameless Participation in Structural Injustice. Social Theory and Practice, 42(2), 149–177. Bach, K. (1995). Terms of Agreement. Ethics, 105, 604–612. Bacharach, M. (1999). Interactive Team Reasoning: A Contribution to the Theory of Cooperation. Research in Economics, 53(2), 117–147. Baier, A. (1985). Postures of the Mind: Essay on Mind and Morals. Minneapolis: University of Minnesota Press. Baier, A. (1994). Moral Prejudices. Harvard: Harvard University Press. Barry, C., & Ferracioli, L. (2013). Young on Responsibility and Structural Injustice. Criminal Justice Ethics, 32(3), 247–257. Bazargan, S. (2012). Complicitous Liability in War. Philosophical Studies, 165, 177–195. Bazargan, S. (2013). Morally Heterogeneous Wars. Philosophia, 41(4), 959–975. Bazargan-Forward, S. (2016). Accountability and Intervening Agency: An Asymmetry Between Upstream and Downstream Parties. Utilitas, 29(1), 110–124. Bazargan-Forward, S. (2017). Complicity. In M. Jankovic, & K. Ludwig (Eds.), Routledge Handbook on Collective Intentionality (pp. 327–337). New York: Routledge. Bazargan-Forward, S. (2018). The Identity-Enactment Account of Associative Duties. Philosophical Studies, 176, 2351–2370. Bazargan-Forward, S. (2020). Responsibility for Shared Action in War. In S. BazarganForward, & D. Tollefsen (Eds.), The Routledge Handbook of Collective Responsibility (pp. 433–446). New York: Routledge. Bazargan-Forward, S. (2021). Grounding the Beneficiary Pays Principle. In D. Sobel, & S. Wall (Eds.), Oxford Studies in Political Philosophy (Vol. 8, pp. 3–36). Oxford: Oxford University Press. Benbaji, Y., & Statman, D. (2019). War by Agreement. Oxford: Oxford University Press. Björnsson, G. (2020). Collective Responsibility and Collective Obligations Without Collective Moral Agents. In S. Bazargan-Forward, & D. Tollefsen (Eds.), The Routledge Handbook of Collective Responsibility (pp. 127–141). New York: Routledge.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

238



Black, O. (2004). Agreements, Undertakings, and Practical Reason. Legal Theory, 10, 77–95. Black, O. (2007). Two Theories of Agreement. Legal Theory, 13, 1–22. Blomberg, O., & Hindriks, F. (2020). Collective Responsibility and Acting Together. In S. Bazargan-Forward, & D. Tollefsen (Eds.), The Routledge Handbook of Collective Responsibility (pp. 142–154). New York: Routledge. Blum, L. (2004). What Do Accounts of ‘Racism’ Do? In M. P. Levine, & T. Pataki (Eds.), Racism Mind (pp. 56–77). Ithaca: Cornell University. Blum, L. (2010). “I’m Not a Racist, but . . .”: The Moral Quandary of Race. New York City: Cornell University Press. Bobo, L., Kluegel, J., & Ryan, S. (1997). Laissez-faire Racism: The Crystallization of a Kinder, Gentler, Antiblack Ideology. In S. A. Tuch, & J. K. Martin (Eds.), Racial Attitudes in the 1990s: Continuity and Change (pp. 93–120). Westport: Praeger. Brake, E., & Millum, J. (August 2, 2016). Parenthood and Procreation. Retrieved February 9, 2017, from https://plato.stanford.edu/archives/win2016/entries/parenthood/ Bratman, M. (1987). Intention, Plans, and Practical Reason. Cambridge, MA: Harvard University Press. Bratman, M. (1992). Shared Cooperative Activity. The Philosophical Review, 101(2), 327–341. Bratman, M. (1993). Shared Intention. Ethics, 104, 97–113. Bratman, M. (1999). Shared Intentions and Mutual Obligation. In M. Bratman (Ed.), Faces of Intention (pp. 130–141). Cambridge: Cambridge University Press. Bratman, M. (2009). Intention, Belief, and Instrumental Rationality. In Sobel David, & S. Walls (Eds.), Reasons for Action (pp. 13–36). Cambridge: Cambridge University Press. Brodie, D. (2007). Enterprise Liability: Justifying Vicarious Liability. Oxford Journal of Legal Studies, 27(3), 493–508. Broome, J. (2001). Are Intentions Reasons? In C. Morris, & A. Ripstein (Eds.), Practical rationality and preference: Essays for David Gauthier (pp. 98–120). Cambridge: Cambridge University Press. Broome, J. (2004). Reasons. In J. R. Wallace, P. Pettit, S. Scheffler, & M. Smith (Eds.), Reason and Value: Themes from the Moral Philosophy of Joseph Raz. Oxford: Oxford University Press. Broome, J. (2005). Does Rationality Give Us Reasons? Philosophical Issues, 15, 321–337. Broome, J. (2009). The Unity of Reasoning? In S. Robertson, S. John, & T. Jens (Eds.), Spheres of Reason (pp. 92–92). Oxford: Oxford University Press. Cane, P. (2002). Responsibility in Law and Morality. Portland: Hart Publishing. Cohen, L. (2008). A Consumers’ Republic: The Politics of Mass Consumption in Postwar America. New York: Alfred A. Knopf. Coleman, J. (1992). Risks and Wrongs. Oxford: Oxford University Press. Collins, S. (2019a). Collective Responsibility Gaps. Journal of Business Ethics, 154, 943–954. Collins, S. (2019b). Group Duties: Their Existence and Their Implications for Individuals. Oxford: Oxford University Press. Copp, D. (1979). Collective Actions and Secondary Actions, American Philosophical Quarterly, 16, 177–186. Copp, D. (2006). On the Agency of Certain Collective Entities. In P. French, & H. Wettstein (Eds.), Shared Intentions and Collective Responsibility (pp. 194–221). Malden: Blackwell. Cullity, G. (2008). Decisions, Reasons and Rationality. Ethics, 119, 57–95. Dancy, J. (2004). Ethics without Principles. Oxford: Clarendon. Darwall, S. (2003). Desires, Reasons and Causes. Philosophy and Phenomenological Research, 67, 435–443.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi



239

Darwall, S. (2004). Respect and the Second Person Standpoint. Proceedings and Addresses of the American Philosophical Association, 78, 43–59. Darwall, S. (2011). Demystifying Promises. In H. Sheinman (Ed.), Promises and Agreements (pp. 255–276). Oxford: Oxford University Press. Davidson, D. (1963). Actions, Reasons, and Causes. In D. Davidson (Ed.), Essays on Actions (pp. 3–19). Oxford: Oxford University Press. de Moore, A. (1987). Are Contracts Promises? In J. Eekelaar, & J. Bell (Eds.), Oxford Essays in Jurisprudence (Vol. 3). Oxford: Oxford University Press. Dill, J., & Shue, H. (2012). Limiting the Killing in War: Military Necessity and the St. Petersburg Assumption. Ethics & International Affairs, 26(3), 311–333. Dummett, M. (2004). The Nature of Racism. In T. Pataki, & P. M. Levine (Eds.), Racism in Mind (pp. 27–34). Ithaca: Cornell University Press. Dymski, G. A., & Mason, P. L. (2005). Racial Inequality and African Americans’ Disadvantage in the Credit and Capital Markets. In C. A. Conrad, J. Whitehead, P. Mason, & J. Stewart (Eds.), African Americans in the US Economy (pp. 151–157). Oxford: Rowman & Littlefield. Enoch, D. (2011). Giving Practical Reasons. Philosophers’ Imprint, 11(4), 1–22. Enoch, D. (2014). Authority and Reason-Giving. Philosophy and Phenomenological Research, 89(2), 296–332. Epstein, B. (2015). The Ant Trap. Oxford: Oxford University Press. Feinberg, J. (1968). Collective Responsibility. Journal of Philosophy, 65(21), 674–688. Feinberg, J. (1970a). Action and Responsibility. Princeton: Princeton University Press. Feinberg, J. (1970b). Doing and Deserving. Princeton: Princeton University Press. Feinberg, J. (1986). Harm to Self. New York: Oxford University Press. Ferrero, L. (2010). Decisions, Diachronic Autonomy, and the Division of Deliberative Labor. Philosophers’ Imprint, 20(2), 1–23. Fischer, J. M., & Tognazzini, N. A. (2009). The Truth About Tracing. Noûs, 43(3), 531–556. Foot, P. (2001). Natural Goodness. Oxford: Clarendon. Frankfurt, H. (1973). Coercion and Moral Responsibility. In T. Honderich (Ed.), Essays on Freedom of Action (pp. 65–86). London: Routledge & Kegan Paul. Freiderich, D., & Southwood, N. (2011). Promises and Trust. In H. Sheinman (Ed.), Promises and Agreements (pp. 277–294). Oxford: Oxford University Press. French, P. (1979). The Corporation as a Moral Person. American Philosophical Quarterly, 16(3), 207–215. French, P. (1984). Collective and Corporate Responsibility. New York: Columbia University Press. French, P. (1995). Corporate Ethics. Fort Worth: Harcourt Brace. French, P. (1998). Individual and Collective Responsibility. Rochester: Schenkman. Frowe, H. (2015). Defensive Killing. Oxford: Oxford University Press. Garcia, J. (1996). Heart of Racism. Journal of Social Philosophy, 27(1), 5–45. Garcia, J. (1997). Current Conceptions of Racism: A Critical Examination of Some Recent Social Philosophy. Journal of Social Philosophy, 28(2), 5–24. Garcia, J. (1999). Philosophical Analysis and the Moral Concept of Racism. Philosophy and Social Criticism, 25, 1–32. Garcia, J. (2004). Three Sites for Racism: Social Structurings, Valuings, and Vice. In T. Pataki, & P. M. Levine (Eds.), Racism in Mind (pp. 35–55). Ithaca: Cornell University Press. Gardner, J. (2007). Moore on Complicity and Causality. University of Pennsylvania Law Review, 156, 432–443.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

240



Gauthier, D. (1986). Morals by Agreement. Oxford: Clarendon Press. Gibbons, J. (2010). Things That Make Things Reasonable. Philosophy and Phenomenological Research, 81(2), 335–361. Gilbert, M. (1989). On Social Facts. Oxford: Princeton University Press. Gilbert, M. (1990). Walking Together: A Paradigmatic Social Phenomenon. Midwest Studies in Philosophy, 15(1), 1–14. Gilbert, M. (1993). Agreements, Coercion, and Obligation. Ethics, 103, 679–706. Gilbert, M. (1996). Living Together. Plymouth, UK: Rowman & Littlefield Publishers. Gilbert, M. (1999). Obligation and Joint Commitment. Utilitas, 11, 143–163. Gilbert, M. (2000). Sociality and Responsibility. Lanham: Rowman & Littlefield. Gilbert, M. (2006). A Theory of Political Obligation. Oxford: Oxford University Press. Gilbert, M. (2008). Two Approaches to Shared Intention: An Essay in the Philosophy of Social Phenomena. Analyze & Kritik, 30, 483–514. Gilbert, M. (2009). Shared Intention and Personal Intention. Philosophical Studies, 144, 167–187. Gilbert, M. (2011). Three Dogmas about Promising. In H. Sheinman (Ed.), Promises and Agreements (pp. 73–101). New York City: Oxford University Press. Glasgow, J. (2009). Racism as Disrespect. Ethics, 120(1), 64–93. Glover, J. (1975). It Makes No Difference Whether or Not I Do It. Proceedings of the Aristotelian Society, Supplementary Volumes, 49, 171–209. Goodin, R., & Pasternak, A. (2016). Intending to benefit from wrongdoing. Politics, Philosophy & Economics, 15, 280–297. Gravely, W. (2019). They Stole Him Out of Jail: Willie Earle, South Carolina’s Last Lynching Victim. Columbia, South Carolina: University of South Carolina Press. Haque, A. (2017). Law and Morality at War. Oxford: Oxford University Press. Hardimon, M. (2019). Should We Narrow the Scope of “Racism” to Accommodate White Sensitivities? Critical Philosophy of Race, 7(2), 223–246. Hardimon, M. (2020). Institutional Racism and Individual Responsibility. In S. BazarganForward, & D. Tollefsen (Eds.), Routledge Handbook of Collective Responsibility. New York: Routledge. Hart, H. (1955). Are there any natural rights? Philosophical Review, 64(2), 175–191. Hart, H. (1958). Legal and Moal Obligation. In A. Melden (Ed.), Essays in Moral Philosophy (pp. 82–107). Seattle: University of Washington Press. Hart, H. (1982). Essays on Bentham: Jurisprudence and Political Philosophy. Oxford: Oxford University Press. Hart, H. (1990). Commands and Authoritative Legal Reasons. In J. Raz (Ed.), Authority (pp. 92–114). New York: New York University. Hart, H. L., & Honoré, T. (1958). Causation in the Law (2nd ed.). Oxford: Clarendon Press. Haslanger, S. (2004). Oppressions: Racial and Other. In M. P. Levine, & T. Pataki (Eds.), Racism in Mind (pp. 97–123). Ithaca: Cornell University. Heafey, R. J., & Kennedy, D. M. (2006). Product Liability: Winning Strategies and Techniques. New York: Law Journal Press. Hess, K. M. (2014). The Free Will of Corporations (and Other Collectives). Philosophical Studies, 168(1), 241–260. Heuston, R., & Buckley, R. (1996). Salmond and Heutson on the Law of Torts (21st ed.). London: Sweet & Maxwell. Hieronymi, P. (2005). The Wrong Kind of Reason. Journal of Philosophy, 102(9), 437–457. Hinchman, E. (2003). Trust and Diachronic Agency. Noûs, 37(1), 25–51. Hobbes, T. (2008). Leviathan. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi



241

Holton, R. (2004). Rational Resolve. Philosophical Review, 113(4), 507–535. Honoré, T. (1999). Responsibility and Fault. Oxford: Hart Publishing. Hooker, B. (2011). Promises and Rule Consequentialism. In H. Sheinman (Ed.), Promises and Agreements: Philosophical Essays (pp. 237–254). Oxford: Oxford University Press. Hoover, D. (2003, March 16). The lynching of Willie Earle, SC’s last, foreshadowed changing times. Retrieved October 15, 2020, from Greenville News: https://www.greenvilleonline. com/story/news/local/greenville-roots/2018/04/09/lynching-willie-earle-scs-lastforeshadowed-changing-times/499199002/ Howden, D. (September 9, 2008). ‘Don’t kill me,’ she screamed. Then they stoned her to death. Retrieved October 12, 2020, from Independent: https://www.independent.co.uk/ news/world/africa/dont-kill-me-she-screamed-then-they-stoned-her-to-death-1003462. html Hull, G. (2015). Affirmative Action and the Choice of Amends. Philosophia, 43(1), 113–134. Hurd, H. M. (2001). Is it Wrong to Do Right When Others Do Wrong? Legal Theory, 7(3), 307–340. Isaacs, T. (2011). Moral Responsibility in Collective Contexts. Oxford: Oxford University Press. Kadish, S. (1985). Complicity, Cause and Blame. California Law Review, 73, 323–410. Kadish, S. H. (1997). Reckless Complicity. The Journal of Criminal Law and Criminology, 87(2), 369–394. Kaiserman, A. (2018). ‘More of a Cause’: Recent Work on Degrees of Causation and Responsibility. Philosophy Compass, 13(7), 1–10. Keatings, G. C. (2014). Strict Liability Wrongs. In J. Oberdiek (Ed.), Philosophical Foundations of the Law of Torts (pp. 292–311). Oxford: Oxford University Press. Kiely, T. F., & Ottley, B. L. (2006). Understanding Products Liability Law. Newark NJ: Matthew Bender. Kolodny, N. (2005). Why Be Rational? Mind, 114, 509–63. Kolodny, N. (2011). Scanlon’s Investigation: The Relevance of Intent to Permissibility. Analytic Philosophy, 52(2), 100–123. Kolodny, N., & Wallace, R. J. (2003). Promises and Practices Revisited. Philosophy & Public Affairs, 31(2), 119–154. Korsgaard, C. (1986). Skepticism about Practical Reason. Journal of Philosophy, 83, 5–25. Kutz, C. (2000). Complicity: Ethics and Law for a Collective Age. Cambridge: Cambridge University Press. Lazar, S. (2010). The Responsibility Dilemma For Killing in War: A Review Essay. Philosophy & Public Affairs, 38(2), 180–213. Lepora, C., & Goodin, R. E. (2013). On Complicity and Compromise. Oxford: Oxford University Press. Lewis, D. (1973). Causation. Journal of Philosophy, 70, 556–567. Lewis, H. D. (1948). Collective Responsibility. Philosophy, 23(84), 3–18. List, C. (2005). Group Knowledge and Group Rationality: A Judgment Aggregation Perspective. Episteme, 2(1), 25–38. List, C., & Pettit, P. (2013). Group Agency: The Possibility, Design, and Status of Corporate Agents. New York: Oxford University Press. Ludwig, K. (2007). Collective Intentional Behavior from the Standpoint of Semantics. Noûs, 41(3), 355–393. Ludwig, K. (2018). From Plural to Institutional Agency (Vol. II). Oxford: Oxford University Press. MacCormick, N. (1972). Voluntary Obligations and Normative Powers I. Proceedings of the Aristotelian Society, 46(1), 59–78.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

242



Mackie, J. (1980). The Cement of the Universe: A Study of Causation. Oxford: Oxford University Press. Marmor, A. (1995). Authorities and Persons. Legal Theory, 1(3), 337–359. Mason, E. (2005). We Make No Promises. Philosophical Studies, 123(1), 33–46. May, L. (1987). Morality of Groups. Notre Dame: Notre Dame Press. May, L. (2005). Crimes Against Humanity: A Normative Account. Cambridge: Cambridge University Press. McCann, H. (1991). Settled Objectives and Rational Constraints. American Philosophical Quarterly, 28, 25–36. McGary, H. (1986). Morality and Collective Liability. Journal of Value Inquiry, 20(2), 157–165. McMahan, J. (1994). Innocence, Self-Defense and Killing in War. Journal of Political Philosophy, 2(3), 193–221. McMahan, J. (2009). Killing in War. Oxford: Oxford University Press. McNeilly, F. (1972). Promises De-Moralized. Philosophical Review, 81, 63–81. Mellema, G. (2016). Complicity and Moral Accountability. Notre Dame: University of Notre Dame Press. Miller, S. (2006). Collective Moral Responsibility: An Individualist Account. In P. A. French (Ed.), Midwest Studies in Philosophy (Vol. XXX, pp. 176–193). Minneapolis: Wiley Blackwell. Miller, S. (2020). Collective Moral Responsibility as Joint Moral Responsibility. In S. Bazargan-Forward, & D. Tollefsen (Eds.), The Routledge Handbook of Collective Responsibility (pp. 38–50). New York: Routledge. Mills, C. (2003). “Heart” Attack: A Critique of Jorge Garcia’s Volitional Conception of Racism. Journal of Ethics, 7(1), 29–62. Mills, C. W. (2002). The ‘Racial Contract’ as Methodology (Not Hypothesis): Reply to Jorge Garcia. Philosophia Africana: Analysis of Philosophy and Issues in Africa and the Black Diaspora, 5, 75–99. Mintoff, J. (2004). Is an Agreement an Exchange of Intentions? Pacific Philosophical Quarterly, 85, 44–67. Moore, M. S. (2000). The Metaphysics of Causal Intervention. California Law Review, 88, 827–878. Moore, M. S. (2007). Causing, Aiding, and the Superfluity of Accomplice Liability. University of Pennsylvania Law Review, 156(2), 396–452. Moore, M. S. (2009). Causation and Responsibility. New York: Oxford University Press. Moran, R. (2001). Authority and Estrangement. Princeton: Princeton University Press. Morris, C. R. (1961). Enterprise Liability and the Actuarial Process: The Insignificance of Foresight. Yale Law Journal, 70, 554–601. Morton, J. (2013). Deliberating for Our Far Future Selves. Ethical Theory and Moral Practice, 16(4), 809–898. Murray, M., & Dudrick, D. (1995). Are Coerced Acts Free? American Philosophical Quarterly, 32, 118–123. Nagel, T. (1989). The View From Nowhere. Oxford: Oxford University Press. Nefsky, J. (2019). Collective harm and the inefficacy problem. Philosophy Compass, 14(2), 1–17. Nolan, V., & Ursin, E. (1995). Understanding Enterprise Liability: Rethinking Tort Reform for the Twenty-First Century. Philadelphia: Temple University Press. Owen, D. G. (2015). Products Liability Law (3rd ed.). St. Paul: West Academic Publishing.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi



243

Owens, D. (2008). Rationalism about Obligation. European Journal of Philosophy, 16(3), 403–431. Owens, D. (2012). Shaping the Normative Landscape. Oxford: Oxford University Press. Parfit, D. (1997). Reasons and Motivation. Proceedings of the Aristotelian Society, 71(1), 99–130. Penner, J. (1996). Voluntary Obligations and the Scope of the Law of Contract. Legal Theory, 2, 325–57. Pettit, P. (2007). Responsibility Incorporated. Ethics, 117(2), 171–201. Philips, M. (1984). Racist Acts and Racist Humor. Canadian Journal of Philosophy, 14, 75–96. Pink, T. (1996). The Psychology of Freedom. Cambridge: Cambridge University Press. Prichard, H. (1940). The Obligation to Keep a Promise. In J. MacAdam (Ed.), Moral Writings (pp. 257–266). Oxford: Clarendon Press. Prosser, W. L., Keeton, W. P., Dobbs, D. B., Keeton, R. E., & Owen, D. G. (1984). Prosser and Keeton on Torts (5th ed.). St. Paul: West Group. Rawls, J. (1971). A Theory of Justice. Cambridge, MA: Harvard University Press. Raz, J. (1977). Promises and Obligations. In P. Hacker, & J. Raz (Eds.), Law, Morality and Society: Essays in Honour of H.L.A. Hart. Oxford: Clarendon Press. Raz, J. (1978). Practical Reasoning. Oxford: Oxford University Press. Raz, J. (1979). The Authority of Law. Oxford: Clarendon Press. Raz, J. (1981). Authority and Consent. Virginia Law Review, 67, 103–31. Raz, J. (1986). The Morality of Freedom. Oxford: Oxford University Press. Raz, J. (1990). Practical Reasons and Norms. Princeton: Princeton University Press. Raz, J. (1999). Agency, Reason and the Good. In J. Raz (Ed.), Engaging Reason. Oxford: Oxford University Press. Raz, J. (1999). Engaging Reason. Oxford: Oxford University Press. Raz, J. (2014). Is There a Reason to Keep a Promise? In G. Klass, P. Saprai, & G. Letsas (Eds.), Philosophical Foundations of Contract Law (pp. 58–77). Oxford: Oxford University Press. Reiman, J. (2012). The Structure of Structural Injustice: Thoughts on Iris Marion Young’s ‘Responsibility for Justice’. Social Theory and Practice, 38(4), 738–751. Roberts, D. (2002). Shattered Bonds: The Color Of Child Welfare. New York: Civitas Books. Robins, M. (1976). The Primacy of Promising. Mind, 85(339), 321–340. Rosati, C. (2011). The Importance of Self-Promises. In H. Sheinman (Ed.), Promises and Agreements: Philosophical Essays (pp. 124–155). Oxford: Oxford University Press. Roszkowski, M. E., & Roszkowski, C. L. (2005). Making Sense of Respondeat Superior: An Integrated Approach for both Negligent and Intentional Conduct. Southern California Review of Law and Women’s Studies, 14(2), 235–288. Roth, A. (2004). Shared Agency and Contralateral Commitments. The Philosophical Review, 113(3), 359–410. Roth, A. (2014). Indispensability, the Discursive Dilemma, and Groups with Minds of Their Own. In S. R. Chant, F. Hindriks, & G. Preyer (Eds.), From Individual to Collective Intentionality (pp. 136–162). Oxford: Oxford University Press. Roth, A. (2015). Practical Intersubjectivity and Normative Guidance: Bratman on Shared Agency. Journal of Social Ontology, 1(1), 39–48. Roth, A. (2017). Entitlement to Reasons for Action. In D. Shoemaker (Ed.), Oxford Studies in Agency and Responsibility (Vol. 4, pp. 75–92). Oxford: Oxford University Press. Rovane, C. (1997). The Bounds of Agency: An Essay in Revisionary Metaphysics. Princeton: Princeton University Press.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

244



Rundle, B. (1997). Mind in Action. Oxford: Oxford University Press. Sadler, B. J. (2006). Shared Intentions and Shared Responsibility. Midwest Studies in Philosophy, XXX, 115–144. Sangiovanni, A. (2018). Structural Injustice and Individual Responsibility. Journal of Social Philosophy, 49(3), 461–483. Sartorio, C. (2019). More of a Cause? Journal of Applied Philosophy, 37(3), 346–363. Scanlon, T. (1998). What We Owe to Each Other. Cambridge, MA: Harvard University Press. Scanlon, T. (2008). Moral Dimensions: Permissibility, Meaning, Blame. Cambridge, MA: Harvard University Press. Schmitz, M. (2017). What is a mode account of collective intentionality? In G. Preyer, & G. Peters (Eds.), Social Ontology and Collective Intentionality: Critical Essays on the Philosophy of Raimo Tuomela with his Responses (pp. 37–70). Cham: Springer. Schroeder, M. (2004). The Scope of Instrumental Reason. Philosophical Perspectives, 18, 337–64. Schroeder, M. (2007). Slaves of the Passions. Oxford: Oxford University Press. Schwenkenbecher, A. (2018). Making Sense of Collective Moral Obligations: A Comparison of Existing Approaches. In K. Hess, V. Ignesky, & T. Isaacs (Eds.), Collectivity: Ontology, Ethics, and Social Justice (pp. 109–32). London: Rowman & Littlefield. Sciaraffa, S. (2009). On Content-Independent Reasons: It’s Not in the Name. Law and Philosophy, 28, 233–260. Searle, J. (1997). The Construction of Social Reality. Free Press. Searle, J. (2001). Rationality in Action. Cambridge, MA: MIT Press. Searle, J. (2010). Making the Social World. Oxford: Oxford University Press. Sellars, W. (1980). On Reasoning about Values. American Philosophical Quarterly, 17(2), 81–101. Setiya, K. (2007a). Cognitivism about Instrumental Reason. Ethics, 117, 647–673. Setiya, K. (2007b). Reasons without Rationalism. Princeton: Princeton University Press. Setiya, K. (2014). Intentions, Plans, and Ethical Rationalism. In M. Vargas, & G. Yaffe (Eds.), Rational and Social agency: The Philosophy of Michael Bratman (pp. 56–82). Oxford: Oxford University Press. Setiya, K. (2014). What is a Reason to Act? Philosophical Studies, 167(2), 221–235. Shapiro, S. (1998). On Hart’s Way Out. Legal Theory, 4(4), 469–507. Shapiro, S. (2002). Authority. In J. Coleman, & S. Shapiro (Eds.), The Oxford Handbook of Jurisprudence (pp. 382–339). Oxford: Oxford University Press. Sheinman, H. (2008). Promises as Practice Reason. Acta Analytica, 23(4), 287–318. Sheinman, H. (2011). Agreement as Joint Promise. In H. Sheinman (Ed.), Promises and Agreements (pp. 365–396). Oxford: Oxford University Press. Sheinman, H. 2011. Promises and Agreements: Philosophical Essays. Oxford: Oxford University Press. Shelby, T. (2002). Is Racism in the ‘Heart’? Journal of Social Philosophy, 33, 411–20. Shelby, T. (2003). Review: “I’m Not a Racist, But . . .” The Moral Quandary of Race. The Philosophical Review, 112(1), 124–126. Shelby, T. (2014). Racism, Moralism, and Social Criticism. Du Bois Review, 2(1), 57–74. Shelby, T. (2016). Dark Ghettos: Injustice, Dissent, and Reform. Harvard: Belknap Press. Shiffrin, S. (2008). Promising, Intimate Relationships, and Conventionalism. Philosophical Review, 117(4), 481–524. Shiffrin, S. (2012). Are Contracts Promises? In A. Marmor (Ed.), The Routledge Companion to the Philosophy of Law (pp. 241–258). London: Routledge.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi



245

Shiffrin, S. V. (2011). Immoral, Conflicting and Redundant Promises. In R. J. Wallace, R. Kumar, & Samuel Freeman (Eds.), Reasons and Recognition: Essays on the Philosophy of T.M. Scanlon (pp. 155–178). New York: Oxford University Press. Simmons, A. (1979). Moral Principles and Political Obligations. Princeton: Princeton University Press. Simmons, A. (1984). Consent, Free Choice, and Democratic Government. Georgia Law Review, 18, 791–819. Smart, J., & Williams, B. (1973). Utilitarianism: For and Against. Cambridge: Cambridge University Press. Smiley, M. (2010). From Moral Agency to Collective Wrongs: Re-Thinking Collective Moral Responsibility. Journal of Law and Policy, 19, 171–202. Smith, S. (2004). Contract Theory. Oxford: Oxford University Press. Strawser, B. J. (2011). Walking the Tightrope of Just War. Analysis, 71(3), 533–544. Szigeti, A. (2020). The Discursive Dilemma and Collective Responsibility. In S. BazarganForward, & D. Tollefsen (Eds.), The Routledge Handbook of Collective Responsibility (pp. 297–312). New York: Routledge. Tadros, V. (2019). To Do, To Die, To Reason Why: Individual Ethics in War. Oxford: Oxford University Press. Thomson, J. J. (1990). The Realm of Rights. Cambridge, MA: Harvard University Press. Tognazzini, N. (2007). The Hybrid Nature of Promissory Obligation. Philosophy and Public Affairs, 35(3), 203–232. Tollefsen, D. (2003). Participant Reactive Attitudes and Collective Responsibility. Philosophical Explorations, 6(3), 218–234. Tuomela, R. (2006). Joint Intention, We-Mode and I-Mode. Midwest Studies in Philosophy, 30, 35–58. Tuomela, R. (2010). The Philosophy of Sociality: The Shared Point of View. New York: Oxford University Press. Velleman, D. (1997). Deciding How to Decide. In G. Cullity, & B. N. Gaut (Eds.), Ethics and Practical Reason (pp. 29–52). Oxford: Oxford University Press. Waldron, J. (1995). Moments of Carelessness and Massive Loss. In D. G. Ownes (Ed.), Philosophical Foundations of Tort Law (p. 387). Oxford: Oxford University Press. Walzer, M. (2000). Just and Unjust Wars: A Moral Argument with Historical Illustrations (3rd ed.). New York: Basic Books. Watson, G. (1996). Two Faces of Responsibility. Philosophical Topics, 24, 227–248. Watson, G. (2004). Asserting and Promising. Philosophical Studies, 117(2), 57–77. Watson, G. (2009). Promises, Reasons, and Normative Powers. In D. Sobel, & S. Wall (Eds.), Reasons for Action. Cambridge: Cambridge University Press. Way, J. (2015). Reasons as Premises in Good Reasoning. Pacific Philosophical Quarterly, 98, 251–270. Weale, A. (2010). The SS: A New History. London: Brown: Abacus. Weiss, B. (2002). What Were They Thinking?: The Mental States of the Aider and Abettor and the Causer Under Federal Law. Fordham Law Review, 70(4), 1341–1381. Wellman, C. (2005). A Theory of Secession. New York: Cambridge University Press. Wertheimer, A. (1987). Coercion. Princeton: Princeton University Press. West, R. (1947, June 14). Opera in Greenville. The New Yorker. Westlund, A. (2011). Autonomy, Authority, and Answerability. Jurisprudence, 2(1), 161–179. White, G. E. (2003). Tort Law in America: An Intellectual History. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

246



Wieland, J. W., & van Oeveren, R. (2020). Participation and Superfluity. Journal of Moral Philosophy, 17(2), 163–187. Williams, B. (1981). Moral Luck. Cambridge: Cambridge University Press. Williams, B. (1995). Making Sense of Humanity. Cambridge: Cambridge University Press. Wright, R. (2001). Once more into the bramble bush: Duty, causal contribution, and the extent of legal responsibility. Vanderbilt Law Review, 54(3), 1071. Yaffe, G. (2012). Moore on Causing, Acting, and Complicity. Legal Theory, 18(4), 437–458. Yeager, D. (1996). Criminal Justice Ethics, 15, 25–35. Young, I. M. (2011). Responsibility for Justice. New York: Oxford University Press. Zimmerman, M. J. (1985). Intervening Agents and Moral Responsibility. The Philosophical Quarterly, 35(141), 347–358. Zipursky, B. C. (2005). Philosophy of Tort Law. In The Blackwell Guide to the Philosophy of Law and Legal Theory (pp. 122–137). Malden, MA: Blackwell Publishing.

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

Index Note: Figures are indicated by an italic “f ”, respectively, following the page number. For the benefit of digital users, indexed terms that span two pages (e.g., 52–53) may, on occasion, appear on only one of those pages. accomplice liability: according to Christopher Kutz 173–4 actus reus of 163, 165–6 authority-based accountability as an account of, see authority-based accountability, and accomplice liability as derivative liability 162–3, 165–6, 169, 171–3, 179–81 as vicarious liability 13–14, 18, 162–3, 172, 175–81, 175n.30, 176n.31 causal / contributory accounts of 18, 162–9, 172–3, 179–80 and causation, see causation, relevance to accomplice liability culpability in, see culpability, of an accomplice definition of 162–5 ignorance in, see ignorance, in accomplice liability in joint action 164–5, 174 mens rea of 163–4 non-derivative / direct 165–7, 169, 171–4, 179–81 for non-proxyable wrongs 167–9, 179 overdetermination in 18, 172–4, 180 in Pinkerton 177–9 reckless 162, 165, 169–70, 179 and risk, see risk, in accomplice liability superfluity of 165–7 voluntary intervention principle 171, 171n.21 accountability: for advice 73–5 and agent-causation 166 alienation and, see alienation authority-based, see authority-based accountability and causation, see causation, as a basis for moral accountability in Christopher Kutz’s account 58–60, 59n.34 civilian 13, 17, 139–40, 150–1 coercion and, see coercion combatant 139–42, 145, 148–9, 151 compensatory 184–5, 194–5, 202; see also legal liability, compensatory of an employee for an employer’s conduct 47, 208, 219, 222–4, 227

of an employer for an employee’s conduct 18, 47, 206, 227; see also respondeat superior for conferring a wrongful purpose, see authority-based accountability in the context of non-cooperative collective action 12, 12n.21 definition of 15, 35–6 deliberator’s, see deliberators, accountability of executor’s, see executors, accountability of; authority-based accountability, and executor accountability; divisions of agential labor, executor accountability for of group-agents 95–6 for harms, see harms, accountability for in the lynching of Willie Earle 131–6 ‘ordinary’ 5, 11–12, 38, 43–4, 47, 53–4, 67, 198, 209; see also authority-based accountability, ordinary and intervening agency 5n.10, 171 mitigated 13, 97–8, 106–8, 140, 151–2, 158–61; see also ignorance, as a mitigator of accountability for more than what is caused 1–3, 8, 10–11, 47–8, 52, 66, 71, 151, 207, 229 for purposes, see purposes accountability for relevance of intentions to see intentions, relevance of and responsibility, see responsibility vicarious, see agreements, vicarious accountability in for a wrong-making feature 5, 7–10, 36–8, 40–3, 49–50, 53, 57, 61, 64, 70, 75, 83, 97–9, 154–5, 229 zero-sum 7, 43, 170 agency 1–2, 5, 26–7, 29–30, 38–9, 41, 168–70, 174, 210, 218–19, 222 active vs passive role 25–6 agent-causal view of 165–6 corporate 95–6 deliberative 1, 226–7 diachronic 15, 29–30, 32–4, 32n.22, 84 distributing see ‘division of agential labor’ executory 1, 15, 28–9, 35, 226–7 group 3, 16, 95–6, 210 individual planning 92

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

248



agency (cont.) innocent 180–1 intervening 5n.10, 142, 167–71; see also accountability, and mediated agency outsourcing see ‘division of agential labor’ relation in the law 189–90 shared 87–8 vicarious 175–8, 180–1 agreements 5–7, 10, 16, 26, 47–51, 54, 61, 67–71, 75, 87, 93, 128, 143, 170–1, 174, 192, 195–6, 199, 203–4, 223–4, 227, 229 alienated, see alienation among soldiers 143, 147–8, 152–3, 155–6 authority established by way of 7, 15–16, 32, 51, 70, 91 as a basis for authority-based accountability, see authority-based accountability, agreements as a basis for coerced, see coercion conditional 85 content of 52, 175 establishing divisions of agential labor, see divisions of agential labor, established through agreements ignorantly established, see ignorance immoral 10, 16, 84, 95 implicit 7, 147–8 as interdepending promises 87–8 as joint decisions 89–90 in the lynching of Willie Earle 129, 132–6 as offers 88 pairwise 1, 7, 143 in Pinkerton 177 and reasons, see reasons, in agreements redundant 66–7 relevance of protected reasons in 75, 83, 85–6, 93, 95 strength of 174 in suretyships 176 vicarious accountability in 174, 177–8 alienation 16–17, 51, 97–9, 116, 126–7 deliberator’s 97, 102–5, 126–7, 151–4, 156–8 executor’s 97, 99–101, 105, 111, 126–7, 155–6 and integrity 230 and promises, see promises, alienated; promises, weakly alienated; promises, strongly alienated in war 13, 139–40, 151, 154–8, 161 weak vs. strong 97, 100–5, 126, 155–8 Anderson, Elizabeth 217 attributability 58–9, 175, 186–7, 189–90 as a form of responsibility 35–6 authority-based accountability 1, 10–13, 15–16, 36–7, 41, 43, 51–3, 57–60, 71–2, 97, 99,

109–10, 124, 126, 128, 139, 143, 160–1, 208, 218, 226–7, 229–30 agreements as a basis for 10, 75, 88 as a form of vicarious liability 18, 162–3, 175–6, 180–1 and alienation in war 154–8, 161 causation in 11, 66–7, 86–7 of civilians in war 17, 150–1, 160–1 and coercion in war 140, 151, 159–61 and complicity see authority-based accountability, and accomplice liability compared to ordinary accountability 37–8, 43–4, 53–4 and accomplice liability 13–14, 18, 163, 172, 174, 179–80, 229–30 definition of 7 and duties of care 197, 206 and executor accountability 37–8, 45–7, 208, 219–20 and following racist orders 15, 18, 208–9, 222–4, 226 for advice 74–5 and group size 68–9 and group-agency 95–6 in a chain of command 10, 145–6 and ignorance in war 13, 17, 139, 151–4, 161 and immoral promises 16, 82–3, 95 implications for integrity 19, 229–31 intuitions behind 11–12 and joint action in war 17, 142–6, 148 and joint commitments 89–91 in morally heterogenous wars 149 and Pinkerton 179 promises as a basis for 77–8, 175 redundancy of 7, 47–50 relevance of intentions to 42, 60; see also intentions, moral relevance of requests and demands as a basis for 75, 85, 159 of employers for employees 13–14, 18, 183–4, 197–8, 204, 229–30 and revisionism in war ethics 13, 17, 139–40, 148, 151, 160–1 role of protected reasons in 37–8, 44–5, 95 strength / scalarity of 67–8, 174 shared / joint action as a basis for 16, 75–6, 89, 91, 94, 142 and alienation in war 139, 151, 154–8, 161 in the lynching of Willie Earle 128–9, 131–6 and war-crimes 148–9, 161 autonomy 32, 166–7, 226–7 in divisions of agential labor, see divisions of agential labor, autonomy in

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

 executor’s, see executors, autonomy of role of in promises, see promise, autonomybased normative power view of blame 35–6, 171, 214 mitigated 98, 105n.2, 118, 139, 212n.17 see also culpability Blum, Lawrence 109–10, 210 Bratman, Michael 1, 75–6, 88–9, 91–4 Brodie, Douglas 184–5 Cane, Peter 188 causation: and action-attribution 58–9 and agent-causal accounts 165–6 and agent-relative constraints 168–9 but-for 9, 43, 49–50, 66, 196 concurrent 50–1, 143, 180–1 counterfactual accounts of 172–3 and deliberators, see deliberators, causal influence of; see also, causation, role of in deliberator’s influence on executor and deviant causal chains 66 in diachronic decision-making 29, 32–3 and enterprise liability 14, 18, 183–5, 191–2, 194–7, 206; see also enterprise liability, role of causation in in expectationalism 80 and integrity 231 interpersonal 29, 165–7, 170–1, 179 INUS / NESS accounts of 50, 66, 196 in the Lynching Willie Earle 132, 134 as a basis for moral accountability 2–3, 5, 10–13, 15, 18, 36, 41, 43, 47–8, 50, 52–4, 66, 69, 71, 74–5, 86–7, 134, 140, 143, 143n.12, 151, 168–9, 171–4, 209, 224–5, 229–31 in non-cooperative collective committed harms 11 in non-proxyable wrongs 167–8 and redundant promises 66 overdetermined 13–14, 18, 50–1, 69, 71, 172–4, 180–1 and product liability 200 and racial injustice 209, 224–5; see also racism, causal contributions to and reckless complicity 162, 170–1 regularity accounts of 172–3 relevance to accomplice liability 162, 165–73, 175–80 and respondeat superior 18, 182–7, 190–1 role of in deliberator’s influence on executor 11, 34–6, 41, 43, 48–51, 69, 74–5, 86–7, 180, 208 singularist accounts of 173–4

249

and strict liability 197, 200–2 tracing 136, 191 versus causal influence 166 versus constitutive determination 11, 41, 74–5 versus formal relationships 183–4, 197 in vicarious liability 13–14, 175–7, 180–1, 185 in war 142–3, 143n.12 chain of command 51, 115, 147–8, 150–1, 157–8, 172–3; see also coercion, in a chain of command; authority-based accountability, in a chain of command Chappell, Sophie-Grace 230–1 coercion 10, 16, 26, 51, 60, 97–8, 105–16, 126–8 in a chain of command 115 definition of 105n.2, 106–7 of a deliberator 97–8, 105–6, 112–16, 127, 159–60 in diachronic self-governance 29–30 of an executor, 97–8, 105–6, 108–12, 115–16, 127, 159 general mitigatory effects 107–8 and promises, see promises, coerced, possibility of, coerced into accepting, coerced into making in war 13, 17, 140, 151, 158–61; see also authority-based accountability, and coercion in war Cohen, Lizabeth 201 commands 27–8, 32–3, 32n.23, 51, 61, 74n.2, 85–6, 149, 161, 163n.3, 178 common knowledge 52, 92, 92n.45 compensation, see legal liability, compensatory; accountability, compensatory complicity, see accomplice liability cooperants 2–3, 7, 9–11, 18, 47, 52, 71, 229 accountability among, see authority-based accountability agreements among, see agreements authority among, see reasons, protected causal reach of, see authority-based accountability, causation in coerced, see coercion deceived, see ignorance ignorant, see ignorance cooperative activity 7, 11–12, 67, 70–1, 76, 105, 121, 124–5, 148, 164, 229 complicity in, see accomplice liability agreements in, see agreements alienation in, see alienation authority in, see reasons, protected coercion in, see coercion furnishing a purpose in, see divisions of agential labor, purposes in ignorance in, see ignorance

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

250



cooperative activity (cont.) negligible/overdetermined contributions to 2–3, 9, 12–13, 18, 47, 67–9, 71, 160–1, 180 in war 17, 139, 143, 148, 153 cooperatively committed harms, see cooperative activity culpability 9–13, 35–6, 51, 89, 118, 167, 171, 173–5, 178, 180, 197–8, 210n.7, 214, 221 of an accomplice 13, 18, 162, 165–6, 171, 178–80 of civilians 13, 17, 150–1 and coercion, see coercion in contributing to a wrong 2, 12n.21, 18, 52, 162–3, 175, 179 of deliberators, see deliberators, culpability of of employees 192, 205, 218, 220–1 of employers 163, 204, 208, 218, 220 of vendors 197, 200, 202–6 of executors, see executors, culpability of and ignorance, see ignorance in the lynching of Willie Earle 132, 135–6 of racist attitudes 15, 209, 211 of soldiers 139, 145–6, 153, 160 see also blame Davidson, Donald 56 decisions, see deliberation, decisions deliberation 38, 221 about advice, see accountability, advice about demands, see demands in Margaret Gilbert’s account 90–1 and integrity 231 interpersonal 28–9, 34–5, 39–40, 44, 84–5, 91 intrapersonal 23–4, 29–32, 84–5 normative freedom to engage in 45 about promises, see promises, role of protected reasons in about protected reasons, see reasons, protected purpose of 27–30 in Abraham Roth’s account 55 about requests, see requests vs. thinking about 27–8 deliberators 5, 7–10, 12, 15, 19, 33, 41, 42n.5, 43, 47–9, 51–3, 58–9, 70, 96–8, 143, 157–8, 229, 231 accountability of 1, 7–11, 15, 34, 36–8, 40–6, 49, 57, 59–61, 64–7, 71–3, 75, 83, 95, 97–102, 104–5, 109–10, 112–14, 116–22, 125–8, 143, 151–2, 158, 179–80, 197–8, 208, 224 alienated, see alienation, deliberator’s authority of 6, 26–7, 29, 32–5, 51, 83, 95, 101–4, 111

becoming 5, 26–7, 33, 35, 45, 75–6, 82–3, 95, 151–2, 157–9, 223–4 as a but-for cause 43 causal influence of 11–12, 15, 75; see also causation, role of in deliberator’s influence on executor coerced, see coercion, of a deliberator conferring a purpose, see purposes, furnishing/ conferring in a criminal enterprise 13–14, 18, 172, 174, 179–80 in a democracy 150–1 in a division of agential labor 1, 26–7, 29, 32–3, 40–2, 46, 55, 59, 64–5, 70, 74–5, 83, 86–7, 98–9, 101, 103–4, 109–13, 115–16, 126, 143, 157, 161, 179–80, 223, 229 furnishing duties of care 14, 18, 204 ignorant, see ignorance, deliberator’s in the lynching of Willie Earle 132–6 motivating reasons of 7–8, 15, 26, 37–46, 50, 55–7, 61, 64, 71–2, 74–5, 98, 101–3, 105, 158, 197–8, 220, 224; see also reasons, deliberator’s racist 15, 220 employers / supervisors as 197–8, 204, 220, 222–4 among soldiers 147–8, 156–60 demands 5–6, 61, 73, 101, 106, 109, 112–13, 123, 127, 158–9, 226–7 authority to issue 26–9, 35, 70, 77, 79, 86, 91, 102–3, 155–6 as a basis for authority-based accountability, see authority-based accountability, requests and demands as a basis for of integrity 19, 230–1 for an explanation 38, 51, 153, 229 establishing a division of agential labor, see divisions of agential labor, established through demands function of 110–11 in the lynching of Willie Earle 128–31, 133–4 divisions of agential labor 1, 3, 11, 15, 33–5, 43, 47–8, 55, 59, 64, 70, 72, 74–5, 83, 95–7, 143, 226–7, 229, 231 agentive functions in 1–2, 26–9, 33–5, 37, 39–40, 58–9, 65, 70, 204, 222–4, 226–7 alienation in, see alienation, executor’s; alienation, deliberator’s autonomy in 29, 45 as a basis for legal complicity 172, 174, 178–80 as a basis for respondeat superior 183, 197–8, 204–6

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

 coercion in, see coercion, of executor; coercion, of deliberator deliberators in, see deliberators, divisions of agential labor in a democracy 150–1 disobeying 32–4 and duties of care 183, 197–8, 204–6 established through agreements 15–16, 75, 83, 87–8, 98–9 established through demands 109–10, 158–9 established through promises 15–16, 33–4, 75–6, 82–3, 87, 98–9, 101, 103–4, 111–13, 116, 122, 174; see also promises, divisions of agential labor in established through requests 15–16, 75, 83–4, 86–7, 98–9 established through shared action 15–16, 75, 83, 88–9, 98–9 executor accountability for 37–8, 64–5, 145–6, 223–4, 227; see also executors, in a division of agential labor and group agency 95–6 ignorance in, see ignorance, executor’s; ignorance, deliberator’s and integrity 19, 231 intrapersonal 33–4, 84, 94 in the Lynching of Willie Earle 133–5 motivating reasons in 1, 37–42, 55–6, 65, 74–5, 82–3, 98–9, 204, 222–4 purposes in 1, 11, 15, 19, 40–3, 46, 48, 55–6, 59, 64–5, 70, 74–5, 82–4, 95, 97–9, 101, 104, 119, 126, 135, 145–6, 152–3, 155, 161, 178–80, 197–8, 204, 222–4, 226–7, 229, 231 in racist institutions 222–4, 227 in war 145–6, 152–3, 155–9, 161 Dill, Janina 141 Dummett, Michael 210 duties of care 14, 18, 183, 197–206; see also deliberators, furnishing duties of care; divisions of agential labor, and duties of care; executors, enacting duties of care; respondeat superior, special duties of care as a basis for Duhulow, Aisha Ibrahim 1 employee / employer 3, 5–6, 14–15, 18–19, 46, 52–3, 182–3, 205 accountability, see accountability, of an employer for an employee’s conduct; accountability, of an employer for an employee’s conduct culpability, see culpability, of employee; culpability, of employer

251

and authority-based accountability, see authority-based accountability, of employers for employees purposes, see purposes, employee’s in a supervisorship 175–6, 182 scope of employment, see respondeat superior, scope of employment in duty of care, see duty of care racism of 208–9, 218–20, 226–8 Enoch, David 73 enterprise liability 14–15, 18, 182–4, 186, 186n.17, 193n.42, 206 as grounds for respondeat superior 14, 183–4, 187, 190–2, 197, 205–6 role of causation in 194–5, 206 basis for 191–2, 194, 195n.45, 196 and ignorance, see ignorance, in enterprise liability executors: accountability of 15, 37–8, 45–7, 64–5, 116–17, 145–6, 208, 222–4, 227; see also authority-based accountability, and executor accountability alienated, see alienation, executor’s authority over 1, 5–7, 15–16, 26–9, 32–5, 39, 47–8, 51, 70, 76, 82–3, 95, 101–4, 109–10, 122, 125–6, 147–8, 155–9 autonomy of 33–4, 45, 127 becoming 1–2, 5–7, 10–11, 26–7, 35, 51, 70, 72, 74–6, 82–3, 86–7, 104–5, 151–2, 157–9, 223–4 coerced, see coercion, of an executor in a criminal enterprise 172, 174, 179–80 disobeying 32–4 in a division of agential labor 1, 26–7, 29, 32–3, 40–2, 46, 55, 59, 64–5, 70, 74–5, 83, 86–7, 98–9, 101, 103–4, 109–13, 115–16, 126, 143, 157, 161, 179–80, 223, 229 enacting duties of care 18, 198, 204 function of 1–2, 5, 15, 18, 26–7, 32–5, 37, 39–43, 45, 55, 58–9, 65, 70, 73, 75, 82–3, 86–7, 100–1, 109–10, 118, 145–7, 155–6, 158–9, 222–4, 226–7 ignorant, see ignorance, executor’s in the lynching of Willie Earle 132–5 purpose conferred upon, see purposes, executor’s racist 9–10, 15, 208, 220, 223–4 among soldiers 147–8, 151–2, 155–9 Ferrero, Luca 29–32 foreseeability 18, 47, 72, 115, 119, 123, 148–51, 155, 159, 174, 182–3, 186–8, 191–3, 196, 206, 214, 224, 227; see also risk, foreseeable Frankfurt, Harry 105n.2

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

252



French, Peter 95 Frowe, Helen 141n.3 functions 1, 5, 15, 18, 24–9, 31–5, 37, 39–43, 45, 55, 58–9, 73–4, 82–3, 94n.52, 100–1, 118, 134–6, 145–7, 158–9, 163, 204, 222–4, 226–7, 231 active vs. passive 25–6, 35, 39, 70, 223–4 agentive 15, 25, 70, 75, 144n.14, 222–4; see also divisions of agential labor, agentive functions in of combatants 146–7, 150–1, 155–6, 158 of commands, see commands of conduct 1, 33, 39–40, 65, 109–10 of consent 109 deliberator’s see deliberator of demands see demands, function of design-based 25 in a division of agential labor, see divisions of agential labor, agentive functions in executor’s, see executor, function of in intrapersonal decision-making 1, 29–30, 65 natural 25 outsourcing / assigning 24–9, 33, 35, 39, 226–7 of pre-commitments 29–30 of promises, see promises, function of of reasons, see reasons, function of of requests, see requests, function of, status 25n.5 in terms of protected reasons 26, 70, 75 use-based 25, 25n.4 Garcia, J.L.A. 210, 215 Gardner, John 167 Gilbert, Margaret 16, 54n.11, 75–6, 82n.17, 87–91, 93, 93n.50, 108n.5 Glasgow, Joshua 214 group 2–3, 10, 15, 51, 54n.12, 58, 94n.52, 123, 130, 164–5, 168–9, 177–8, 229–30 accountability, see accountability, of group-agents agency, see agency, group conferring a purpose 52–3 intentions, see intentions, group racialized 18–19, 207, 210–17, 226–8 size 68–9 Haque, Adil 142–3 Hardimon, Michael 9, 211–12, 215, 217n.30, 218n.32, 219n.33 harms 1–2, 11–12, 12n.20, 44, 68–9, 104–5, 143, 196–8, 221–2, 230–1 accountability for 7, 11–13, 37–8, 47–8, 53, 58, 71, 200–3, 205

agentially mediated, see agency, intervening avoiding, 108 coercively committed, see coercion compensating for, see compensation cooperatively committed, see cooperative activity deontological constraints against committing 91 enterprises that cause 14, 180–1, 183, 185–9, 191–3, 195–6, 200–2, 204–6 environmental 11–13 evaluating 38, 53–4, 108, 198–9, 202–3, 221–2 ignorantly committed, see ignorance legal liability for 167, 171, 180–1, 183, 185–9, 186n.17, 191–3, 196–7, 204–6 in the lynching of Willie Earle 129–30, 134–5 modal tracking of 222 overdetermined 2–3, 9, 12–14, 68, 71, 140, 160–1, 172, 180, 229 of promise-breaking 78–80 racist, see racism, harms of resulting from alienation, see alienation risk of 14, 168–70, 168n.17, 173–4, 185–7, 189, 191–3, 195–6, 206 special duty not to cause 198–203, 205–6 in war 13, 139–40, 142–3, 160–1 within the scope of employment 183, 185–9, 206 wrong-making features of 41, 53 Hart, H.L.A. 5n.10, 27, 50n.9, 76nn.3,4, 82n.16, 165–6, 171, 171n.19 Haslanger, Sally 210–12, 210n.8, 218–19 Heafey, Richard J. 201nn.54,57 Hobbes, Thomas 175 hierarchies 28, 54n.12, 90n.40, 91, 95, 139, 145, 150–1, 175–6, 227–8 ignorance 10, 51, 60, 110, 116–28 in accomplice liability 164 deliberator’s 16–17, 97–8, 116–21, 148–9, 151–2 in enterprise liability 195, 206 executor’s 16–17, 97–8, 116–17, 127, 152 as a mitigator of accountability 98 non-culpable 98, 116, 125, 152–3 of racial injustice 216 of reasons 55–6 in war 13, 16–17, 139, 151, 154–8, 161 interpersonal divisions of agential labor, see ‘divisions of agential labor’ institutions 14, 18–19, 78–9, 78n.8, 139, 157–8, 161, 175–6, 193n.42, 207–15, 207n.1, 212n.17, 217–20, 224–5, 227–30

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

 intentions 1, 4, 7, 11–12, 25–8, 25nn.4,5, 36, 42–3, 51, 54n.12, 72, 77, 79–80, 88, 92–4, 99–100, 128, 141, 148, 168, 178 acting on another’s 54–5, 57 Bratman’s account of 1, 23n.2, 91–3 conditional 84, 174 group 88–9, 95 joint 58–9, 164–5 in the lynching of Willie Earle 130–2, 134 meshing 92–4 moral relevance of 15, 38, 42, 60–5, 221–4, 228 participatory 58–9, 173–4 purpose of 39–40 racist 15, 207, 220–4, 228 shared 12n.20, 88–9, 88n.26, 91–3, 92n.45, 93n.50, 94n.52, 174 in the law 163–5, 169, 172, 175–6, 179, 182–7, 184n.5, 205 vicariously enacted 15, 18–19, 228 joint action, see shared action Kadish, Sanford 166–9, 171, 173–4 Kantianism 3 Kennedy, Don M. 201nn.54,55 Kutz, Christopher 2n.6, 3n.7, 15, 37, 54, 58–60, 143n.11, 173–4 legal liability 166, 173–7, 179–81, 203, 206 of accomplices, see accomplice liability compensatory 14, 18, 176, 182–3, 202–3, 214 derivative vs. direct 162–3, 165–7, 169, 171–3, 179–81 enterprise, see enterprise liability for non-proxyable wrongs 167–8, 179 for products 200–2; see also causation, and product liability strict 184, 193, 197, 200–2 of successors 196 of superiors, see respondeat superior vicarious 13–14, 18, 162–4, 172, 175–80, 176n.31, 182, 185, 186n.17, 189–90 see also enterprise liability Lewis H.D., 23, 178n.39 liability, see moral liability; legal liability List, Christian 12n.15, 33n.24, 52n.10, 95n.55 Ludwig, Kirk 15, 25nn.4,5, 164n.7 McMahan, Jeff 140–1 metaphysics 5, 41, 166, 179, 210 Moore, Michael 166n.11, 167–8, 169n.18, 171nn.19,22, 173–4, 173n.25, 176n.31, 177–9 moral liability 160, 214 of civilians 150

253

in shared action 142 of soldiers 13, 139–41, 143, 143n.12, 148–9, 150n.18, 151, 160–1 moral luck 66–7, 100, 226 Moran, Richard 4n.9, 29–30 Morris, C. Robert 194 Nagel, Thomas 63–4, 221–2 overdetermination: in accomplice liability, see accomplice liability, overdetermination 18, 172–4, 180 and causation, see causation, overdetermined in complicity, see accomplice liability, overdetermination in in a cooperative harm, see cooperative activity, negligible/overdetermined contributions to in a harm, see harms, overdetermined negligible/overdetermined contributions to 13–14, 50–1, 69, 143, 172–3, 180 Owens, David 2n.6, 3n.7, 28n.11, 76nn.3,4, 77, 79n.11, 82n.16, 85–6, 88nn.28,29 Pettit, Philip 12n.15, 33n.24, 95n.55 Philip Morris International 2, 8, 7–10, 52 Pinkerton 177–9 product liability, see legal liability, product promises: accepted or made insincerely 101–3, 156 alienated 16, 97–100, 102, 111, 116, 156 authority in 16, 28, 34–5, 39, 70, 72, 76–8, 82–3, 93, 95, 101–4, 112, 122–5, 156, 174 authority-based normative power views of 77–8 autonomy-based normative power views of 77–8 as a basis for authority-based accountability, see authority-based accountability, promises as a basis for bindingness of 16, 28, 77–9, 81–2, 82n.17, 87, 97, 101, 103, 108–11, 123, 127, 156, 158 breaking of, see harms, of promise-breaking coerced, possibility of 108–9 coerced into accepting 105–6, 112–16, 127, 159 coerced into making 16, 105, 109–12, 127, 158 conditional 84–5, 87–8, 99, 123–4, 174 as constituents of agreements 75, 87–8 conventionalist views of 16, 78–80, 78n.8 discretion in 112–13, 159 dispositionally defective 100 divisions of agential labor in 15–17, 33–5, 40, 70, 75–6, 82–4, 87–8, 95, 98–9, 101, 103–6, 109–13, 115–16, 122, 127, 134–5, 156, 174 exclusionary force of 44, 79–80

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

254



promises: (cont.) expectationalist views of 78, 80–1, 80nn.14,15 function of 77–8, 101, 108–9 ignorantly accepted 16–17, 117–23, 125–8, 151–2 ignorantly made 16–17, 116–17, 122–3, 125–7, 151–2 immoral 16, 28, 35, 75–6, 81–4, 82nn.16,17, 95, 156; see also authority–based accountability interdepending, see agreements, as interdepending promises in the Lynching of Willie Earle 134–6 redundant 66–7 relation to requests 84–6 role of protected reasons in 6, 15–16, 28, 33–5, 39–40, 70, 75–85, 87–8, 93, 95, 100–1, 103–4, 109, 111–13, 122, 128, 156, 174–5 stringency of 80–1, 174 strongly alienated 101, 103–5, 156 and the Principle of Fidelity 80 the role of trust in 80, 80nn.14,15 in war 151–2, 156, 158 weakly alienated 100–3 proxyship 175, 177–9; see also accomplice liability, for non-proxyable wrongs purposes: accountability for 1, 3–5, 7–9, 11, 15, 36–7, 40–54, 56–7, 59, 61, 64–72, 74–5, 83, 89, 91, 95, 97–9, 102–3, 105, 117–20, 122–3, 125–9, 132–6, 145–6, 148, 150, 152–61, 174–5, 178, 180, 197–8, 204, 222–4, 227, 229, 231 aggregating 52–3 alienated, see alienation among soldiers 145–6, 148, 152–60 coerced, see coercion in the commission of war-crimes 152–3 content of 8n.12, 26, 36–7, 40–1, 48–9, 52–3, 74–5, 104, 113, 204 criminal 152–3, 163, 174–5, 178, 180 decisive 49–51, 69, 71–2 in a division of agential labor, see division of agential labor, purposes in employee’s 204, 222–4, 227 executor’s 1, 5–11, 8n.12, 15, 19, 26, 36–7, 40–52, 42n.5, 55–7, 59, 61, 64–72, 74–5, 82–3, 95, 97–131, 135, 145–6, 155–6, 158, 161, 179–80, 197–8, 204, 222–4, 226–7, 229, 231 furnishing/conferring 3–10, 15, 36–7, 40–3, 42n.5, 45–53, 57, 59, 61, 64, 66–72, 83, 89, 91, 95, 97–9, 101–5, 117–20, 122–3, 125–8,

132–6, 145–6, 148, 150, 153, 155–61, 180, 197–8, 204, 231 of a group, see group, conferring a purpose ignorance of, see ignorance in joint commitments 91 official 52, 152–3 opacity of 8n.12, 40, 56–7 in Pinkerton case 178–9 racist 52, 208, 215, 223–4 redundancy of 49, 68–9, 71–2 role of motivations in 7–8, 15, 26, 36–7, 39–46, 48–50, 52–3, 55–7, 61, 64–5, 69, 71–2, 74–5, 99, 102–4, 113, 119, 127, 132–3, 155–6, 158, 197–8, 223–4 role of reasons in 1, 7–8, 15, 26, 36–7, 39–46, 48–52, 55–7, 61, 64, 70–2, 74–5, 82–4, 89, 95, 97–105, 117–20, 122–3, 126–8, 132–3, 135, 145–6, 148, 153–6, 158, 174, 178, 197–8, 204, 222–4, 227, 229 in shared / joint action 52–3, 58, 83, 89, 91, 95, 123, 128, 226–7 unique 48–51, 68–9, 71–2 violation of 83, 155, 175, 204 wrongful 3, 7–10, 15, 36–7, 40, 42–3, 45–9, 51–4, 59, 64, 68–70, 75, 83, 89, 97, 102–4, 118, 123, 127–8, 132–3, 135, 145–6, 153, 179–80, 197–8, 224, 229 racism accountability for 218–26; see authority-based accountability, and following racist orders absent racists 208, 218 accounts of 210 attitude-independent, see racism, extrinsic causal contributions to 209, 225 conceptual inflation of 109–10, 217 culpability for, see culpability, of racist attitudes in deliberators, see deliberators, racist direct vs. indirect victims of 225 in employees/employers, see employees / employers, racism of examples of 14, 52, 107, 212–13, 215–20 excused 209, 219n.33, 225 in executors, see executors, racist extrinsic 211 harms of 15, 201–2, 207, 214–15, 228; see also harms, racist indifference 208, 215–18 institutional 14, 207, 209, 215; see also divisions of agential labor, in racist institutions and intentions, see intentions, racist

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

 overt vs. covert 208, 213 policies, 207 race-neutral 210, 213, 215–16 reductionist vs. non-reductionist accounts of 208, 210–12, 214–15, 217–18 responsibility gaps for attribution of 213 structural 207, 212n.17 transmitting moral effects of 15, 18, 219–20, 227–8 see also purposes, racist Raz, Joseph 1n.1, 6, 15, 27n.8, 28n.10, 32n.23, 33–4, 76nn.3,4, 77n.7, 82n.16, 85–6, 108n.6, 144n.14 reasons: in Abraham Roth’s account 55–6 action-guiding 27–8, 62–3 advice yield, see accountability, advice agent-neutral 63–4 agent-relative 63–5, 90, 168–9, 221–2, 225 in agreements 1n.1, 7, 10, 15–16, 49–51, 69–70, 75, 83–4, 87–8, 90, 93, 95, 128, 133, 151–3, 156, 174–5, 204, 223–4 alienated from, see alienation, reasons in analysis of authority 6, 27–30, 33–5, 70, 95 anaphoric reference to 31–2, 40, 204, 223–4, 227 and authority-based accountability, see authority-based accountability, role of protected reasons in as premises 8n.12, 55–6 bootstrapping 30–2, 30n.17 in Bratman’s account of shared action 16, 75–6, 89, 93–4 causally influencing 74–5, 170 choosing to see as 62 coerced, see coercion, reasons conferred in a vote 52–3, 150–1 conferring 6–7, 16, 24, 26, 28–9, 33–9, 44, 49, 52, 55, 57, 74–6, 82–4, 86, 95, 100, 103–4, 109–11, 126–8, 132–4, 143, 146–8, 150–1, 156–7, 174, 178, 204, 226–7, 231 convergence of 31 convincing 74 creating ex nihilo 30, 74n.2 decisive 31–2, 45, 49–51, 62, 69, 71–2, 79–83, 195 definition of 1n.1, 38, 55–7 deliberating vs. thinking about, see deliberation, vs. thinking about deliberator’s 1, 6–8, 8n.12, 15, 26, 33–5, 37–46, 50–1, 55–7, 61, 64, 71–5, 82–3, 97–9, 101–5, 117–20, 126–8, 158–9, 197–8, 204, 227, 229, 231 deontic relevance of 38, 60, 64–5, 222–3

255

deriving from authority 7, 27–35, 27n.8, 70, 76, 82–3, 86, 91, 95, 101, 109–10, 112, 144n.14, 156–7, 178 determining a purpose 4–5, 7–8, 8n.12, 15, 26, 36–7, 40–3, 45–6, 48–51, 55–7, 61, 64, 71–2, 74–5, 82–3, 89, 91, 95, 97, 100, 105, 117–19, 126–8, 153, 158, 197–8, 204, 222–4, 227, 231; see also purposes, role of reasons in in diachronic decision–making, see decisions, diachronic directed 49, 71–2 enacting 1, 37, 39–45, 49, 56–7, 64–5, 74–5, 82–3, 86–7, 98–101, 109–10, 118–19, 132–3, 155–6, 158–9, 222–3 exclusionary 6, 26–7, 27n.8, 31–2, 34, 44–5, 76, 79–80, 85–6, 90, 117, 145 executor’s 1, 7–8, 8n.12, 15–16, 33, 35, 37–8, 40–3, 45–6, 55–7, 61, 70–2, 74–6, 82–3, 95, 99–101, 103, 109–10, 118, 127–8, 155, 157, 204, 224, 226–7, 229, 231 explanatory 74, 120 fact-relative vs. belief-relative 1n.1 first-order 6, 27, 31, 44–5, 79–80, 145 first-personal relevance of 63–4, 221, 223 to fulfill promises, see promises function of 1, 39–40, 86–7, 109–10, 158–9 in Gilbert’s account of shared action 16, 75–6, 87–91 giving 4, 27–8, 34, 55, 63–4, 70–1, 73, 73n.1, 74n.2, 76, 80–1, 85–7, 101, 109–10, 158–9, 170, 221–4 ignorance of, see ignorance of, reasons introspective opacity of 8n.12, 40, 55–7 ‘looking through’ 4, 36–7, 118, 126 in the lynching of Willie Earle 132–5 modifying 74–5 motivating 1, 4, 7–8, 8n.12, 18, 26–7, 36–46, 48–50, 52–3, 55–7, 60–5, 70–2, 74–5, 82–3, 97–105, 116–20, 122–3, 126–8, 132–3, 153, 155–6, 158–9, 170, 179, 197–8, 204, 221–4; see also divisions of agential labor, motivating reasons in nested 110 non-moral 84, 101, 103, 111 official 52–3 practical 1, 7–8, 16, 26–8, 34–5, 37–41, 44–5, 48–9, 55–7, 60, 65, 73–5, 82–7, 90, 95, 97, 100–1, 103–5, 109–11, 117–20, 122–3, 126–8, 132–3, 153, 156, 204, 222–4, 227, 231; see also reasoning, practical pre-emptive 27n.8 private / personal 65, 104–5, 153, 157–8 pro tanto 68, 221–3

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

256



reasons: (cont.) promises yield, see promises, role of protected reasons in protected 6–7, 10, 16, 26–9, 27n.8, 31–41, 43–5, 49–52, 57, 61, 70–2, 74–89, 91, 93–5, 100–1, 103–4, 109–13, 122, 124–8, 132–4, 143–4, 144n.14, 146–8, 150, 154–9, 174–5, 204, 223, 226–7, 231; see also agreements, relevance of protected reason in; functions, in terms of protected reasons as rationalizing action 56 in the ‘Reasoning View’ of 55–7 redundancy of 49, 71–2 relation to rationality 1n.1, 32, 93–4 requests yield, see divisions of agential labor, established through requests second-order 6, 27, 45, 76, 79–80, 94, 145 in shared/joint action 15–16, 75–6, 83, 87–91, 93–5, 122, 125, 128, 143–4, 226–7 soldiers confer 143, 148 strength of 144, 148, 174 ‘tainted’ / ‘infected’ 98, 198, 220, 223–4, 227 teleological orientation of 1, 8n.12, 39 triggering vs. robust 73n.1 unique 48–50, 71–2 unprotected 37–8, 44–5 wrongful 37–8, 40–3, 45–6, 57, 64, 73, 75, 98, 119, 128, 197–8, 220–1 reasoning, practical 1, 39, 55–6, 65, 93–4, 198 requests: and authority-based accountability, see authority-based accountability, requests and demands as a basis for divisions of agential labor established through, see divisions of agential labor, established through requests function of 85–7 and promises, see promises, relation to requests and reasons, see reasons, requests yield respondeat superior 13–14, 176, 187–90, 196–7, 206, 229–30 and causation, see causation, and respondeat superior definition of 13–14, 182–3 and divisions of agential labor, see divisions of agential labor, as a basis for respondeat superior enterprise liability as grounds for, see enterprise liability, as grounds for respondeat superior for intentional torts 186–7, 205 scope of employment in 182–7, 189–90 special duties of care as a basis for 183, 197, 202–5

responsibility 3, 59, 89n.33, 183–4, 200, 212n.17, 212n.19, 221 accountability–sense of 15, 35–6 attributability–sense of 35–6 for the action of another 171, 178, 185, 196–7 gaps 211–13 risk 32, 73, 98–9, 118–19, 125, 128, 168–9, 168n.17, 170–1, 207, 231 accepted 14, 18, 183, 191–3, 203–6 in accomplice liability 165, 169, 173–6 foreseeable 12–13, 18, 196 harms of, see harms, risk of of running a business 14, 18, 185–93, 195–7, 200, 203–6 in war 153, 156–8 Roberts, Dororthy 216–17 Roth, Abraham 15, 37, 54–7, 90n.38, 92–3, 93nn.48,49, 94n.52 rule-consequentialism 3 Scanlon, T.M. 1n.1, 62–3, 78–80, 80n.15, 93, 93n.48 Schroeder, Mark 1n.1, 55–6 Shapiro, Scott 27–8, 27nn.6,7, 86 Searle, John 15, 25, 33n.24, 82n.17 shared action 16, 54, 70, 75–6, 83, 88–9, 95, 114 Michael Bratman’s account of 16, 75–6, 89, 91–4 and divisions of agential labor, see divisions of agential labor established through shared action Margaret Gilbert’s account of 16, 75–6, 89–91 Christopher Kutz’s account of 58–9 moral liability in, see moral liability, in shared action Abraham Roth’s account of 54–5 in war 140, 142–3, 146, 148 Shiffrin, Seana 76n.4, 77, 77n.7, 79n.11 Strawser, B.J. 141 Shelby, Tommie 210–11, 212n.19, 217n.30 Shue, Henry 141 special duties 14, 18, 183, 197–202, 204–6 supervisorship 175–9, 182 suretyship 176–9 Tadros, Victor 62 Tony Honoré 50, 165–6, 171 Walzer, Michael 141 war: agreements in, see agreements, in war; aims in 139–43, 146, 148–54, 156–8, 160–1 alienation in, see alienation, in war; authoritybased accountability, and alienation in war

OUP CORRECTED PROOF – FINAL, 15/7/2022, SPi

 and authority-based accountability for morally heterogenous, see authority-based accountability, in morally heterogenous wars and authority-based accountability of civilians, see authority-based accountability, of civilians in war causation in, see causation, in war coercion in, see coercion, in war; authoritybased accountability, and coercion in war conscription in 140, 151, 158–61 cooperative activity in, see cooperative activity, in war discrimination in 149–50 divisions of agential labor in, see divisions of agential labor, in war harms in, see harms, in war ignorance in, see authority–based accountability, and ignorance in war; ignorance, in war and authority–based accountability for joint action in, see authority–based accountability and joint action in war morally heterogenous 149–50

257

orthodox accounts of 141 promises in, see promises, in war revisionist account of 13, 140–2, 148, 151, 160–1, 229–30 and risk, see risk, in war and shared action, see shared action, in war; authority-based accountability for joint action, see authority-based accountability, and joint action in war and war-crimes, see accountability for war crimes, see accountability, for war crimes; see authority-based accountability, and war-crimes; purposes, in the commission of war-crimes Watson, Gary 35–6 Way, Jonathan 55–6 Wertheimer, Alan 106n.3 Williams, Bernard 230–1 Yeager, Daniel 173–4 Young, Iris Marion 212nn.17,19