Trust : Analytic and Applied Perspectives [1 ed.] 9789401209410, 9789042036802

Citation preview

TRUST

VIBS Volume 263 Robert Ginsberg Founding Editor Leonidas Donskis Executive Editor Associate Editors G. John M. Abbarno George Allan Gerhold K. Becker Raymond Angelo Belliotti Kenneth A. Bryson C. Stephen Byrum Robert A. Delfino Rem B. Edwards Malcolm D. Evans Roland Faber Andrew Fitz-Gibbon Francesc Forn i Argimon Daniel B. Gallagher William C. Gay Dane R. Gordon J. Everet Green Heta Aleksandra Gylling Matti Häyry Brian G. Henning

Steven V. Hicks Richard T. Hull Michael Krausz Olli Loukola Mark Letteri Vincent L. Luizzi Hugh P. McDonald Adrianne McEvoy J.D. Mininger Peter A. Redpath Arleen L. F. Salles John R. Shook Eddy Souffrant Tuija Takala Emil Višňovský Anne Waters James R. Watson John R. Welch Thomas Woods

a volume in Nordic Value Studies NVS Edited by Matti Häyry, The University of Manchester

TRUST

Analytic and Applied Perspectives

Edited by Pekka Mäkelä and Cynthia Townley

Amsterdam - New York, NY 2013

Cover photo: www.dreamstime.com Cover Design: Studio Pollmann The paper on which this book is printed meets the requirements of “ISO 9706:1994, Information and documentation - Paper for documents Requirements for permanence”. ISBN: 978-90-420-3680-2 E-Book ISBN: 978-94-012-0941-0 © Editions Rodopi B.V., Amsterdam - New York, NY 2013 Printed in the Netherlands

CONTENTS

Editors’ Introduction ONE

TWO

THREE

FOUR

vii

Trust, Trustworthiness, and Responsibility Floora Ruokonen 1. Trust as Encapsulated Interest 2. Hardin’s Analysis of Trustworthiness 3. Trust as Entrusting 4. Trust as an Affective Attitude 5. Components of Trustworthiness 6. From Competence, Trust-responsiveness, and Goodwill to Responsibility 7. Trust and Responsibility 8. A Strong Account of Trust and Trustworthiness

8 10 11

Trusting Interpretations Karen Jones 1. Trust and Interpretation 2. Trust and Trusting Interpretations

15 16 25

Trust in Wittgenstein Olli Lagerspetz and Lars Hertzberg 1. On Psychological Expressions 2. The Use of “Trust” 3. Certainty and Evidence 4. Summary

31 32 36 43 49

A Trust-Based Argument Against Paternalism Simon Clarke 1. The Trust Argument 2. Justifying the Trust Condition 3. Is it Reasonable to Trust Paternalistic Government? 4. Conclusion

53 55 58 64 73

1 2 2 5 5 6

FIVE

SIX

SEVEN

EIGHT

NINE

Trust and Risky Secrets Timo Airaksinen 1. Dangerous Mistrust 2. The Risks of Shared Secrets 3. The Risks of Trust 4. Undisclosed Secrets and Social Power 5. Lost Secrets: A Pro-Surveillance Argument as a Test Case

77 77 78 81 83 92

Public Trust Cynthia Townley and Jay L. Garfield 1. Trust, Reliance and Reactive Attitudes 2. Institutions and Trust 3. Why Public Trust is Important 4. The Limits of Trust 5. Modernity, Community, and Trust

95 96 100 102 103 106

Can You Buy Trustworthiness? Vittorio Pelligra 1. The Birth of a Market 2. Trust, Trustworthiness, and Incentive 3. The Experiment 4. Market for Trustworthiness? 5. Conclusions

109 110 111 113 116 117

Desire for Esteem as Reason for Trust? Pekka Mäkelä 1. The Notion of Trust 2. Mechanisms of Trustworthiness: Loyalty, Virtue, and Prudence 3. Trust-responsiveness Mechanism 4. Discussion

123 124 126

Trust in Nature and Trust in People: Necessity, Care, and Rightness Per Ariansen

131

119 120

About the Contributors

143

Index

147

EDITORS’ INTRODUCTION Pekka Mäkelä and Cynthia Townley The importance of trust can hardly be overstated. Trust is significant in a multitude of human practices, and studied in a range of disciplines. “Whatever matters to human beings, trust is the atmosphere in which it thrives” writes Sissela Bok (Bok, 1978, p. 31n). Although trust is ubiquitous, understanding trust is a non-trivial challenge. Philosophical studies of trust have usually followed Annette Baier’s seminal 1986 paper “Trust and Antitrust,” by taking paradigm cases to be those of one person trusting another, often in intimate, cooperative contexts. Much less philosophical work has examined trust in institutional and more generally impersonal contexts of modern societies. These latter domains of trust are important, not only for their inherent interest, but also because they form the backdrop and climate against which intimate trust relationships take place, and hence they condition at least some interpersonal trust. Outside of philosophy, trust is often treated as a broad, relatively undifferentiated, social phenomenon. It is common, for example, to hear that modern societies are seriously deficient in trust, because of alienation and/or increasing social change. Social institutions are increasingly disconnected from people’s everyday concerns, and thus less able to supply people with meaningful orientation and reliable systems of support. Some social scientists regard the accelerating pace of social change as one of the defining features of a modernity in which the very idea of society as a set of stable institutions and clear boundaries and tasks has come into question. Increasing trust may be a solution to broad social challenges, as trust is considered critical to both social integration through cooperation and to people’s well-being. A significant body of empirical research has been used to support such claims as trust helps to solve large scale collective action dilemmas, trust furthers economic growth and prosperity, trust makes democracy work better, and trust is an enabling condition of the institutions of the welfare state. Many philosophers agree that a society where people are disposed to be trusting, and where their trust is generally well placed, is almost certain to work more harmoniously and fruitfully than a society where trust fails to appear or spread. (See e.g. Hardin, 1993; Gambetta, 1988). Despite the increasing interest in and the growing amount of research on trust, there still is a gap between much of the theoretical and conceptual work on trust taking place within philosophy, and the empirical studies of social scientists. Much of the recent empirical work on trust – be it based on surveys or experiments – does not seem to proceed from a clear account of what is meant by trust in the first place. Trust is sometimes discussed as if it

viii

Pekka Mäkelä and Cynthia Townley

encompasses any and all forms of social cooperation and interdependence. While trust is a pervasive social phenomenon, it is useful to distinguish between different notions of trust, and especially between reliance and stronger, moral notions of trust. This is not merely a matter of theoretical precision: there are distinct risks and benefits associated with different kinds of dependence, and different ways to manage the risks and cultivate the beneficial dimensions of cooperation, in public as well as intimate contexts. Diverse forms of constructive cooperation include but are not limited to trust. Different ways of understanding trust can have direct implications for institutional design. If there are intangible and non-fungible components of trust relationships, then these will be difficult to capture in simple cost-benefit analyses. Since trust and reliability are complex and interacting phenomena, bolstering compliance in some areas may undermine trustful interactions in other ways. It is important to be clear about the characteristics of trust, about the conditions favorable for the prevalence of trust, and about the reasons why people might withhold trust or invest trust in one another and in the institutions of their society. It is challenging to provide a characterization of trust that will suit all contexts and phenomena. Trust comes in different forms, whether between intimates, larger groups or institutions, and trust phenomena differ in various ways from other kinds of cooperation and dependence. A standard, although not uncontested, way to distinguish trust from other kinds of dependence appeals to the trustee’s motivation. Roughly, trust proper involves good will, or something similar (in other words, the emphasis is on the moral motivation of a trustee) and non-trust reliance emphasizes a cooperator’s capacities to behave in the expected manner, and the likelihood of her so behaving (motivation matters less than competence and predictability). In this collection, Floora Ruokonen examines and defends such a distinction that is taken up in later papers, most prominently those by Timo Airaksinen, Simon Clarke and Cynthia Townley and Jay Garfield. The distinction between trust (proper) and (trust as) reliance need not reduce trust to a simple cognitivism, exclusively focused on beliefs, judgments or expectations about the trustworthiness of the other. Richer accounts of trust acknowledge the relevance of affective and conative attitudes, as well as its cognitive dimensions. The papers in this volume by Karen Jones and by Olli Lagerspetz and Lars Hertzberg exemplify these richer accounts of trust. Social cooperation requires both trust and systems of accountability. Institutions, public and private, are entrusted with care for goods that are of value to society as a whole, but at the same time, managing such complex organizations often, and perhaps necessarily, involves incentives and checkable procedures. Two papers deal directly with the tensions that can arise between regulative frameworks and the good will and intrinsic values of cooperation. Townley and Garfield apply a trust/reliance distinction to

Editors’ Introduction

ix

institutions of public trust, distinguishing the conditions for each to thrive. Institutions such as universities incorporate checkable reliance systems and intrinsic values such as integrity and academic freedom. Trust may not be replaced by reliance, in part because monitoring is costly, and because excessive monitoring can corrode integrity and commitment, which trust, conversely can enhance. A related phenomenon is shown empirically in a study of responses to a child-care center’s introduction of a late fee. The charge fails to motivate the desired behavior: it appears to reduce instead of increase the parents’ tendency to collect children on time. Vittorio Pelligra argues that one reason is that the introduction of a financial penalty has replaced a trustful expectation. This experiment gives empirical support to the view that trust is important and distinctive domain of social cooperation, not only worth preserving for its own sake, but also because alternatives may be more, not less, costly. This result might be explained in terms of a framingeffect, the introduction of the material rewards leads to a reframing of the situation, as a fiduciary relation is reframed as a market relation. This is an important result in light of institutional design in general, and in light of the trend of marketization in particular. Trust appears to be a good thing, something to be promoted, from almost every conceivable evaluative perspective. Thus, it is no wonder that the questions concerning generation of trust, maintenance of the atmosphere of trust, and reasons for trust should be and are considered of utmost relevance and importance in different trust related discussions and debates. For instance, from the point of view of institutional design these questions are most pressing. If institutional design aims at furthering and developing wellfunctioning society and at promoting the good ends of the society, and if trust is a constituent ingredient and/or a necessary facilitating condition of such societies, then the question of how to generate and maintain trust should be one of the core issues in the field of institutional design. Pekka Mäkelä’s paper discusses one trust generating mechanism introduced by Philip Pettit. Roughly, Mäkelä criticizes Pettit’s analysis of a trust-responsiveness mechanism arguing that such a mechanism does not provide a promising grounding for socially desirable and robust proper trust-relationships. Like the result of Pelligra’s experiment, Mäkelä’s conclusion is relevant for social planning aiming at the increase of social capital and well-being. Lagerspetz and Hertzberg draw on Ludwig Wittgenstein’s work to enhance, and perhaps challenge standard philosophical understandings of the concept of trust. Building on Wittgenstein’s discussion of psychological expressions they claim that by invoking the language of trust and betrayal, we do not simply identify facts out there. Instead we invoke a certain perspective. We are invited to see someone’s behavior in a certain light. The question to ask here is “what is the role, in human interaction, of speaking about trust?” Lagerspetz and Hertzberg criticize the current analyses of trust for having been shaped by the faulty assumption that our grasp on reality is ultimately to

x

Pekka Mäkelä and Cynthia Townley

be spelled out in terms of factual assertions. They argue for a non-reductive understanding of trust. According to their view, in many cases the trust we have for another individual will ground the beliefs we come to form, instead of the other way around. Such a view is supported by psychoanalysts Kenneth S. Isaacs, and Ernest A. Haggard (Isaacs, Alexander and Haggard, 1963) who claim that trusting and non-trusting persons live in different worlds. It appears, further, that a trusting person can see more opportunities to flourish in the world than a person lacking trust. However, it is not quite clear whether trust is a matter of a worldview. Jones argues that when viewed with trust the world looks different from how it looks when viewed with distrust or neutrality, and this difference matters for both practical and theoretical rationality by way of shaping our reasons for action and belief. Jones illuminates the relationship between trust and interpretation: does trust give rise to trusting interpretations, do the interpretations give rise to trust, or is the relation constitutive? Jones argues for the claim that the relation between trust and interpretation is both constitutive and causal. Considering trust and institutions raises the question of how power differences can affect trust. Airaksinen explores responsibility and power in contexts of secrets and trust. He points out that a secret may require mutual trust from the owner and hearer of the secret. Institutions with requirements for disclosure include those concerned with medical treatment, and Airaksinen argues that a benevolent institution must trust its clients, and allow them secrets. When the power difference is reversed, the trust that citizens may have in authority or government is quite different. Clarke is also directly interested in citizens’ trust of government. He applies the good will element of trust to the justification for trusting government, specifically one that engages in coercive moral paternalism. Joseph Raz has argued that coercive moral paternalism is problematic because it undermines the conditions for reasonable trust, and such trust would be required if that paternalism were to be justified. Clarke presents a distinct analysis of coercive moral paternalism, concluding, like Raz, but for different reasons, that it is problematic. Governments will rarely if ever achieve the conditions that would permit trust in their paternalism to be reasonable. The most radical extension of the concept of trust occurs in Per Ariansen’s study of our usage of “trust” and of the functioning of the concept in the context of nature as well as human relationships. The notion of trust is illuminated in terms of interplay of the notions of trust, trustworthiness, trustability and dependability. Dependability is identified as the notion at the core of trustability (counterpart of trustworthiness in non-human contexts) and trustworthiness. Ariansen distinguishes between different kinds of dependability grounding or, in part, constituting trustworthiness or trustability. Trustworthiness of a person can be constituted by either a dependability grounded on the necessity that arguably characterizes mechanisms of nature,

Editors’ Introduction

xi

or a dependability grounded on free will related to agential features of the person. Ariansen uses this analysis of trust to discuss the crisis of trust in science caused by environmental problems. The papers in this volume address critical and analytical issues of trust. Collectively, the papers both address and extend discussions about trust taking place both within philosophy and in the public domain, with particular attention to institutional trust, and trust that involves social roles. The first three papers in the book offer general characterizations of trust, from a perspective more conceptual than applied. The remaining (six) papers consider trust in practical contexts ranging from the public sphere broadly understood to particular social institutions, such as universities and medical care. The collection as a whole explores what kind of good trust is, what kind of goods it can protect and how it can bring about goods, and develops subtle distinctions between trust and other virtues, and between trust and other forms of dependence. The complexity of trust and the multiplicity of values associated with trust and mechanisms for its cultivation are evident in the collection as a whole. The pluralism of the collection reflects the diversity of real world contexts and theoretical perspectives indispensable in the search of deeper understanding of trust. Without such an understanding of the nature of trust and the good reasons why people might trust one another or the institutions, we are in danger of designing institutions that will reduce trust or even drive it out. The significance of trust to social institutions and the public sphere, is far from exhausted by this the collection. These papers explore different aspects of trust, both analytically, and as pertaining to important social institutions and the public sphere. Trust: Analytic and Applied Perspectives aims at shedding new light on the intersecting dimensions of our social cooperation, in which both trust and reliance can be responsibly undertaken.

Pekka Mäkelä Cynthia Townley

WORKS CITED Baier, Annette. (1986). “Trust and Antitrust,” Ethics, 96:2 (January), pp. 231–260. Bok, Sissela. (1978). Lying. New York: Pantheon Books. Isaacs, Kenneth S., James M. Alexander and Ernest A. Haggard. (1963). “Faith, Trust, and Gullibility,” International Journal of Psychoanalysis, 44:4, pp. 461–469. Gambetta, Diego. (1988). Trust: Making and Breaking Cooperative Relations, Oxford, England: Blackwell. Hardin, Russell. (1993). “The Street-Level Epistemology of Trust,” Politics and

xii

Pekka Mäkelä and Cynthia Townley

Society, 21:4 (December), pp. 505–529. Raz, Joseph. (1986). The Morality of Freedom, Oxford, England: Clarendon Press.

One TRUST, TRUSTWORTHINESS, AND RESPONSIBILITY Floora Ruokonen Introduction Is trust an inherently moral notion? Or does trust have a morally neutral core, so that it has both moral and morally neutral forms? What about trustworthiness: if trust is a moral notion, is trustworthiness likewise, or is its moral status independent of that of trust? I will argue here for the benefits of understanding both trust and trustworthiness as moral notions. The argument is based on an analysis of trust, which places holding someone morally responsible for something at its core. Three claims by Russell Hardin prompt these questions. First, Hardin claims that in current literature on trust the most central conceptions are of trustworthiness instead of trust (e.g., Hardin, 1991, p. 186; 1996, p. 28; 2006, p. 16). Second, he states that there might be a moral approbation that goes with trustworthiness but trust is fully explicable (Hardin, 1996, p. 28; 2002, p. 36). Third, he maintains that that if we want to increase the amount of trust in society, we should promote trustworthiness (Hardin, 2002, p. 30). In light of these statements, understanding the conditions of trustworthiness appears absolutely pivotal for a study of trust. However, when studying Hardin’s own account of trustworthiness one soon runs into the problem that although he claims that there might be a moral approbation associated with trustworthiness, his account does not explain this approbation save for some distinct forms of trustworthiness. I will therefore shortly introduce two different kinds of accounts of trust to see whether they can do better in this respect. My claim is that they can. Then I suggest that a further account of trust, built on the notion of moral responsibility, can be drawn from some salient features of trustworthiness pointed out by these accounts. Against Hardin’s insistence that analyses of trust based on character dispositions or emotions of the trustee are too restrictive to be useful tools for explicating the workings of trust in interpersonal behavior and social institutions (Hardin, 2006, pp. 16, 26), I claim that morally grounded expectations of trustworthiness should be distinguished from other forms of reliance. Otherwise we might, for example, accidentally end up destroying trust relationships by misguided attempts to enhance trustworthiness. I begin by

2

FLOORA RUOKONEN

introducing Hardin’s analysis of trust and the account of trustworthiness arising from it. 1. Trust as Encapsulated Interest In Hardin’s theory of “trust as encapsulated interest” trust is seen as an intentional relation, the rational analysis of which must depend on the rationality of both the trustor and the trustee and on the commitments of the trustee (Hardin, 1991, p. 189). According to the theory, trust is essentially a set of expectations that depend on rational assessments of the trustee’s motivations (ibid., p. 187). The most common motivating facts assessed by the trustor are the interests of the trustee, hence the name of the theory. Trust as encapsulated interest can be analyzed in the following way: (TEI): A trusts that B does X in circumstances C if (1) A expects that B does X in circumstances C, and (2) A’s expectation (1) is grounded on A’s rational assessment of B’s motivations. According to this analysis, trust is reducible to A’s rationally grounded beliefs about B’s behavior, and these beliefs are grounded on an assessment of B’s contingent motivations. In other words, the analysis does not make any particular motivational orientation a necessary condition for trust. As a result, the analysis is non-normative. According to Hardin, the theory purports to be a descriptively adequate analysis of trust, an analysis that is true to the everyday use of the term “trust.” He claims that the term can be used in a wide range of trust-situations. This would mean that I am speaking of trust in just the same sense, whether I am that I trust that my friend’s advice to me is genuinely based on her thinking of my best interest, or I trust the justice system of my country, or that I trust my competitor for a position to vilify me in front of the board. 2. Hardin’s Analysis of Trustworthiness Hardin thinks it strange that so many theorists worried about changes in society claim that their worries concern trust. According to him, they are usually, against their testimony, discussing trustworthiness instead of trust, and quite rightly so. If we are worried about the decline of trust in society, we should investigate trustworthiness, since trustworthiness creates trust (Hardin, 2002, p. 30). In the encapsulated-interest account, “trustworthiness is just the capacity to judge one’s interests as dependent on doing what one is trusted to do” (Hardin, 2002, p. 28). The emphasis in Hardin’s account of trustworthiness is on sticking to promises or expected actions. His examples are of the following kind: what would make us think that somebody who forswears love forever is

Trust, Trustworthiness, and Responsibility

3

trustworthy?; how do I ensure you that you will pay me for your purchase of my car?; or, what would make us believe that somebody who leaves the society in order to live in a convent stays with her decision? Answers to these questions, according to Hardin, indicate ways of creating trustworthiness. In Hardin’s opinion, there is little point in exploring the direct willing of trustworthiness, that is, purely internal commitment to future action. There are indirect ways of “willing,” devices that can be called into action for making ourselves trustworthy in some matters. These devices are largely external, provided by social controls. They range from small-scale controls of ongoing relationships with those near us to large-scale controls of the law and other pervasive social institutions, with “mixed” devices, such as broad social norms and religious controls in between (Hardin, 1996, p. 31). Of institutional constraints that can be used to create trustworthiness, Hardin’s primary example is contract law. He regards it as an almost ideal type of institutional oversight that controls individual commitments. Economic relations between individuals are especially well governed by contracts when compared with many other types of relations. Furthermore, there is “great trustworthiness in contracts because performance is easy to assess and enforcement is relatively easy” (Hardin, 1996, p. 34). Here contracts differ from such arrangements as, for example, marriage where, according to Hardin “there is far less trustworthiness (. . .) in many societies and times, because performance is too hard to measure to make enforcement work” (ibid.). When it comes to creating trustworthiness, the most effective social conventions are, according to Hardin, also simply stated, easily monitored and sanctioned, and relatively stable (Hardin, 1996, p. 38). The example of Ukifune in Lady Murasaki’s Tale of Genji is used to illuminate the working of such conventions (with or without the support of law) (ibid., pp. 35–36). Ukifune decides to shirk from making difficult decisions in her love life by entering a convent. The commitment to leave the court and enter a convent is sealed by the cutting of Ukifune’s six feet long hair that it has taken her all life to grow. The convention is easily stated: “cut your hair and there is no coming back to the world of court.” Adherence to the convention is easily monitored, since a person leaving the convent without hair would instantly be noticed, and sanctioning would be easy, since everyone at the court would immediately shun Ukifune on her return. Finally, the convention is stable, since it has existed before and will exist after the life of Ukifune. The convention of cutting the hair is thus in Ukifune’s society one that secures her commitment of leaving the court quite effectively, and would by Hardin’s account of trustworthiness have made her extremely trustworthy in her decision to leave the court. The example shows that Hardin’s analysis of trustworthiness is built on his analysis of trust. If trust is based on rational expectations concerning the motivational set of the trustee, making oneself trustworthy in the eyes of

4

FLOORA RUOKONEN

others amounts to successful communication of a motivation to fulfill the trust placed on one. The most effective way of communicating one’s motivation to fulfill the trust placed on one is organizing one’s circumstances so that it will be evident to the potential trustor that it is in the trustee’s interests to act in the expected way. Remembering Hardin’s claim that there might be a moral approbation that goes with trustworthiness, but without which trust is fully explicable, it is interesting that the above examples of making oneself trustworthy do not explain such approbation. In his account of trustworthiness he does mention the morality of others as one possible reason to believe that they are trustworthy (Hardin, 2002, pp. 36, 52), but this does not account for some kind of moral approbation going with trustworthiness in general. It might be assumed that since contract breaking is prima facie believed to be wrong, there is an implicit moral demand of trustworthiness. Hardin, however, also thinks that although probably widespread, moral dispositions are relatively thin and unreliable for the vast number of relations people have with others whose moral character is not sufficiently well known to be judged reliable (Hardin, 2002, p. 53). We can perfectly rationally believe that other people are trustworthy, even if it would be irrational to rely on their morality. In Hardin’s account, we are just as justified in talking about trustworthiness when we believe the one we trust would act in the expected way even if there were no contract or norm to secure the commitment, or we believe that she would not do so, were it not for the contract or norm. In both cases there are reasons to believe that the trustee is motivated to act in the expected way. So we are back with the morally neutral analysis of trust. I think many people would intuitively claim that there is a relevant difference in the above two cases, and this difference makes it more justified to attribute trustworthiness to the first instead of to the second person. Many are also ready to claim that the difference has something to do with morality. Theorists who advocate a more exclusive notion of trust refer with the word “reliance” to dependence grounded on external motivations such as legal contracts or norms with sanctions, and seek to distinguish it from “trust.” Trust has something to do with morality whereas reliance does not. One way to point at the difference between mere reliance and trust proper is to refer to the difference in the emotions aroused by breaches of trust and reliance. A breach of trust is met with the feeling of being betrayed, whereas “disappointment” is a better word for describing the emotion associated with misplaced reliance. An adequate analysis of trust should be able to explain this difference. What is it about trust that makes us susceptible, in the case of it being breached, to the reactive attitude of feeling betrayed? It appears that to explain this reaction to breaches of trust we need more exclusive accounts of trust and trustworthiness than those Hardin offers. I will next briefly introduce Annette Baier’s theory of trust as entrusting and then Karen Jones’s affective-attitude theory. Then, following the leads given by

Trust, Trustworthiness, and Responsibility

5

Baier and Jones, I will consider three features which might be taken as components of trustworthiness in these more exclusive accounts.

3. Trust as Entrusting Like many others (though not Hardin), Baier and Jones want to distinguish “trust proper” from “mere reliance” (see Townley and Garfield, this volume; Thomas, 1990; Holton, 1994). According to Baier and Jones, in cases of mere reliance one relies on the other to perform a particular action for any reason whatsoever. Trust on the other hand requires that the trustor believes in the goodwill of the trustee toward her. In these terms, we can rely on our employees to show up at work if we know that they need their jobs and that they know that not showing up will result in getting fired, but we do not trust them to do so. In Baier’s theory trust is analyzed as entrusting. Trust is accepted vulnerability to another’s possible but not expected ill will or lack of goodwill, as one entrusts something one values to her (Baier, 1986, p. 235). The analysis of trust in this theory goes as follows: (TE): A trusts (trust proper) B if (1) A entrusts B with a valued thing C; (2) A leaves B some control over how to take care of C; (3) A is confident that B is competent in taking care of C; and (4) A has confidence in B’s good will toward her or at least does not expect ill will or indifference from B. A’s trust of B is rational in so far as she has good grounds for 3) and 4). In this account, proper trust is about beliefs concerning the competence and goodwill of the trustee. Baier’s discussion of trust also emphasizes the risk involved in trust. When trusting, the trustor puts the trustee in a position to harm her, although she is confident that the trustee will not take advantage of her position. 4. Trust as an Affective Attitude Karen Jones’s (Jones, 1996) account of trust as an affective attitude is another theory of trust that is stronger than Hardin’s. As Jones’s article in this volume makes clear, she has changed her view on trust since presenting that account. As her affective attitude account has been extremely influential in the philosophical discussion on trust and as I do not think that the aspects of the account on which I draw later in this article are in tension with her new view on trust, I will here discuss the earlier account. Like Baier, Jones emphasizes the difference between reliance and trust

6

FLOORA RUOKONEN

proper. According to Jones, reliance is an attitude that resembles a prediction, a belief concerning the future. To be justified, reliance requires more evidence than trust. Trust in turn is partly a distinctive attitude, not a belief, and therefore does not need as much evidence as reliance in order to be justified. The analysis of trust in the theory of trust as an affective attitude is the following: (TAA) A trusts (trust proper) B if (1) A has on optimistic attitude about the goodwill of B; (2) A has an optimistic attitude about B’s competence when it comes to B’s expected behavior; (3) A believes that B will be directly moved by acknowledging that A is counting on her. The core idea of Jones’ theory is that trust is a feeling as well as a judgment and a disposition to act. Trusting someone means that one has an attitude of optimism about the goodwill and the competence of the other as it extends to the domain of one’s interaction with them and an expectation that the other will be directly moved by the thought that we are counting on them (Jones, 1996, p. 15) Jones takes resistance when it comes to evidence to be an essential feature of trust. This means that a person can validly trust that P and at the same time believe that P happening is less likely than it not happening (cf. also Holton, 1994). According to Jones, this feature of trust offers one of the adequate success criteria for a theory of trust; it is a feature that any theory of trust has to be able to explain. If trust were a belief, for example, a belief that P happening is more likely than it not happening, then the trustor would be inconsistent in her beliefs. A model in which trust is essentially a belief, cannot explain all instances of trust. If, however, the mental state belonging to trust proper is not a belief but some other attitude, this problem does not arise. Trusting is more like hoping than predicting and we do not therefore need the same amount of evidence as we do for relying on someone. According to the theory of trust as an affective attitude then, trust is a way of seeing the trustee and an attitude taken toward offered evidence instead of a prediction concerning the behavior of the trustee. Optimism within this theory means an attitude that leads the trustor to anticipate that the other will have and display goodwill, and to base her expectations of the trustee’s behavior on this anticipation. (Jones, 1996, pp. 6, 11.) 5. Components of Trustworthiness Do the above more exclusive accounts of trust explain the moral approbation

Trust, Trustworthiness, and Responsibility

7

that even Hardin admits might be going with trustworthiness? The most central features of trustworthiness arising out of the analysis of trust as entrusting are competence and goodwill. The analysis of trust as an affective attitude adds trust-responsiveness to these. Of these, competence and trustresponsiveness are features that Hardin also takes to be aspects of trustworthiness, although his interpretation of trust-responsiveness differs significantly from that of Jones. Competence is in a sense a self-evident precondition of relying on someone to perform some action. In terms of trust, the belief that the trustee is competent is entailed by the expectation that the trustee will perform some act. In such an analysis the intentional object of trust is the expected behavior of the trustee. When speaking of the trustworthiness of a person there is, however, more to say about competence. Here the intentional object of trust is in the first place the trusted person, and only derivatively her actions. When competence contributes to trustworthiness, we are not merely thinking about the technical competence necessary for specific acts, but a competence based on more or less permanent features of the trustee. Baier emphasizes one feature or skill, discretion, as especially salient for trustworthiness. When one trusts someone with taking care of something that matters to one, one normally leaves some room for the trustee to exercise her discretionary powers in order to discern what this care taking does, and does not, involve. The more reason we have to think that those that we trust are trustworthy, the more we tend to rely on their judgment about just what to do to fulfill the trust we place on them. Competence requires that one knows the appropriate means for fulfilling a given task and knowing the appropriate means requires knowledge of their proper scope and limits. A trustworthy person discerns the requirements and limits posed by a trust-relationship. Jones notes that this kind of competence is sometimes more of a moral than a technical kind: a trustworthy friend understands what loyalty, kindness, and generosity call for in different situations (Jones, 1996, p. 7). Hardin and Jones agree that some kind of trust-responsiveness is a feature of trustworthiness. They differ markedly on how the trustresponsiveness is to be understood. According to Hardin, “to say we trust you means we believe you have the right intentions toward us and that you are competent to do what we trust you to do” (Hardin, 2002, p. 17). The right intentions in his account count as the interest of the trustee to maintain a relationship with the trustor, and this interest gives the trustee an incentive to be trustworthy (ibid.). According to Jones, to have trust, the attitude of optimism when it comes to the goodwill of the trustee needs to be supplemented with a belief that the trustee will be “directly and favorably” moved by the thought that someone counts on them, which means that a manifestation of trust becomes a motivating factor for them. Thus, “one is not trustworthy unless one is willing to give significant weight to the fact that the other is counting on one, and so will not let that consideration be overruled by

8

FLOORA RUOKONEN

just any other concern one has” (Jones, 1996, p. 8). The qualification “directly moved” distinguishes Jones’ notion of trust-responsiveness from Hardin’s. According to Jones, we have reliance instead of trust if the trustor believes that the trustee’s incentive to be trustworthy is indirect in the sense that it is primarily based on considerations such as the continuation of relationship instead of the thought that one is being counted on. (Ibid., p. 9.) Goodwill as an aspect of trustworthiness distinguishes Baier’s and Jones’s accounts of trust most obviously from that of Hardin. According to them, we can rely on people on the grounds of any contingent motivations we take them to have, but trust requires a belief in the goodwill of the trustee toward the trustor. In both accounts, goodwill is to be understood in quite broad terms, encompassing the minimal sense of goodwill as lack of ill will, that is, lack of an intention to harm the other, and a strong sense of goodwill as explicit consideration of the particular interests of the trustor (and whatever in between). We expect a trustworthy stranger to respect our integrity and privacy when walking the street, and we may trust a trustworthy friend to discern what would be in our interest and act accordingly even in cases where we cannot ourselves form a clear picture of our situation (cf. Jones, 1996, p. 7). Both are instances of goodwill. 6. From Competence, Trust-responsiveness, and Goodwill to Responsibility It appears that with competence, trust-responsiveness, and goodwill we have many ingredients at hand to account for the moral approbation going with trustworthiness. The competence needed for trustworthiness is sometimes of a moral kind (for example knowing what loyalty calls for); trustworthy persons are responsive to others counting on them (take other persons’ needs into account); and “goodwill” has a moral sound to it. Yet it appears that we would still need to spell out how, and whether, these ingredients hang together. So far the moral aspects of competence, trust-responsiveness, and goodwill as features of trustworthiness have been presented as contingent. Seeing how they fit the idea that reactions to breaches of trust and reliance are different can illuminate this. None of them can explain why a breach of trust would always be met with emotions of being betrayed. That the trustor finds that belief in the trustee’s competence was mistaken cannot explain why one should feel betrayed when someone trusted turns out untrustworthy. It is utterly unfair to blame anyone for not being able to do something. Trustresponsiveness in turn can at times be morally required of others. We expect our friends to help us even at some cost to them if they notice that we depend on them, and we often expect, mostly more implicitly than explicitly, that people unknown to us recognize that we count on them to respect our physical integrity. Yet we cannot require others to be responsive to everything we happen to count on them to do. We should not feel betrayed if we have placed

Trust, Trustworthiness, and Responsibility

9

unreasonable expectations on someone and she does not give weight to these expectations in her reasoning. We might even try to exploit someone’s trustworthiness by communicating expectations that we do not ourselves consider to be justified. The reaction to finding that our expectations are not taken to merit consideration is here much more likely to be disappointment than a feeling of betrayal. Finally, it appears that goodwill does not explain the difference between reactions to breaches of trust and reliance. Why we should feel betrayed instead of disappointed when we learn that we have been mistaken in our confidence in the goodwill of the trustee? Why should situations of trust be in this respect different from other situations where we learn that, contrary to our expectations, someone is not well disposed toward us? We may in such cases feel frustrated, angry, or disappointed, but not necessarily betrayed. But no doubt we feel betrayed when trust is broken. I suggest that we can account for this feature of trust by formulating the motivational expectation to which we hold the trustee in terms of responsibility instead of terms of goodwill. In this description, there is a definitional connection between trust and reactive attitudes. The description can also incorporate trust-responsiveness and competence as aspects of trustworthiness. The link between trust and the emotion of being betrayed is found in an account of holding someone responsible for something. Of the different senses of responsibility the following are relevant here. Firstly, someone can be taken as a responsible agent. A responsible agent has the capacities of understanding, reasoning, and control of conduct. Secondly, an agent can be (retrospectively) held responsible for what she does or fails to do. In this second sense, one is responsible for at least some events or outcomes which can be ascribed to one. This second kind of responsibility implies the first kind of responsibility, and both are implied in an account of responsibility which defines holding someone responsible as being susceptible to reactive attitudes. If the task is to assess the relationship of the exclusive form of trust to other forms of dependence this is the most illuminating account of responsibility. The following sketch of this account is based on R. Jay Wallace’s elaboration of Peter F. Strawson’s idea (Wallace, 1996; Strawson, 1962/2001). According to Wallace’s account, to “hold someone morally responsible is essentially to be subject to a distinctive range of moral sentiments in one’s interactions with the person” (Wallace, 1996, p. 8). The distinctive range of moral sentiments refers here to reactive attitudes such as resentment, indignation, and guilt. Responding to a person with moral reactive attitudes indicates that we take the person to be, first, a responsible agent, and therefore fit to be the object of moral evaluation. The other is taken to be an agent, someone whose actions reflect personhood. Responsible agents are taken to be autonomous in the sense that they have their values and conceptions of the good, are able to reflect on these conceptions and their relations to their

10

FLOORA RUOKONEN

commitments, and to act in accordance with them. The assumptions of autonomy, rationality, and reflective capacities make a person a potential target of assessments of responsibility and thus, of reactive attitudes. As Wallace points out, the liability to reactive attitudes toward other persons reflects a relationship of deep assessment toward them (Wallace, 1996, p. 52). They are recognized as persons in the moral sense, which is the sense that makes one justified to hold them to moral expectations and conversely makes them justified in holding others to moral expectations. Secondly, responding to someone with reactive attitudes indicates that we hold the person responsible for violating some expectation. The propositional object of reactive attitudes is the belief that an expectation to which one holds a person has been breached. Holding someone to an expectation is to be susceptible to reactive emotions should the expectation be breached, or to believe that reactive emotions would be appropriate when an expectation is breached (Wallace, 2004). The paradigmatic reactive attitudes of resentment, indignation, and guilt are moral in nature. Moral reactive attitudes are distinguished from other attitudes having to do with expectations, such as disappointment, in that moral reasons — agent-neutral or agent-relative — figure in the justification of the expectation (Wallace, 2004, pp. 30, 36). They are “reasons that we ourselves accept as a basis for practical deliberation and normative criticism and discussion” (Wallace, 2004, p. 63). As reactive attitudes are responses within interpersonal human relationships, there appears to be a demand for others to share at least some of our moral reasons. The social aspect of rationality is expressed in the need to have one’s reasoning recognized by other autonomous rational beings. In interpersonal relationships with a moral dimension we ask the other party to share our practical reasoning. And when this does not happen, that is, the other party does not recognize the reasons we have for our moral expectations, we are susceptible to reactive attitudes. 7. Trust and Responsibility I argue that trusting someone consists partly of holding the person morally responsible in the sense that the trustor is susceptible to reactive attitudes, should the expectations to which she holds the person in question be breached. We do not always trust people even if we hold them responsible for something. For example, we can hold parents in general responsible for their children and are likely to feel resentment and indignation toward parents who neglect their children, but we do not trust every parent to act accordingly. What differentiates trust from other forms of holding someone responsible is, I suggest, a specific reason for action among those which we expect another person to recognize. This reason is our dependence on her. We expect the trusted person “to be directly and favorably moved by the thought of being counted on” as Jones put it in her analysis. In terms of the account of

Trust, Trustworthiness, and Responsibility

11

trust given above, this would mean that when trusting someone, we expect her to be moved by the demand of recognition implied in our trust in her. We ask her to recognize us as persons with justified reasons for counting on her. This is, in my opinion, the fundamental moral dimension in trust: the demand to be taken as a person whose dependence on the trusted person needs to be taken into account as a reason in practical deliberation. The content of the normative demand for recognition is, in a trust relationship, grounded on beliefs that concern the nature of the relationship and the mutual recognition of this nature. For optimism concerning the trustworthiness of another person it is crucial to believe that the other shares with us an understanding of the nature of our relationship with each other. A more metaphorical way of putting this is to say that in a trusting relationship, the trustor expects that the trustee will not hurt her, since there is a common perspective out of which both parties view some — wider or narrower — range of affairs and the normative reasons for action that it provides. The depth and extension of the common perspective — both the weight of interests as reasons and the kind of interests considered — depends on the nature of the relationship. The perspective on reasons for action is, for example, far more widely shared in friendships than in accidental encounters, and trust in a friend encompasses different things than our trust in someone we happen to meet in the street. Friends may have knowledge of our particular interests which makes it possible for them to take our interests much more widely into account in their practical reasoning. Moreover, we expect our particular relationships with them, the kind of dependence involved in these relationships, to provide them with reason to do so. What exactly is the extension of the sphere within which recognition of dependence is justified? How much are the parties in a relationship expected to recognize each other’s reasons and interests and incorporate them into their horizons of reasons, that is, how wide is the shared horizon, and how attentive should the parties be to its details? These are crucial questions in such paradigmatic cases of strong forms of trust as, for example, trust in friends, close relatives, or lovers. Such relationships give rise to special obligations deriving from the particular nature of the relationship. It takes judgment and discretion to recognize these duties (on the importance of discretion for trust, see Baier, 1989). Thus, not only the existence of malice and treachery makes one who trusts vulnerable. Trusting is also risky because of the possibility of different interpretations of the nature of a relationship and, as a consequence, the range of justified demands within it. 8. A Strong Account of Trust and Trustworthiness Some accounts of trust include holding someone responsible for something. Such accounts yield a strong or narrowly defined notion of trust. Hardin thinks that in the study of social phenomena broadness instead of narrowness

12

FLOORA RUOKONEN

of the concept is an advantage. In social sciences a major topic of research in the last couple of decades has been “social capital,” that is, social structures that create social harmony and cooperation which in turn help the members of the society to realize their individual or collective goals. Trust has been taken to be a central part of social capital. As seen, Hardin holds that trustworthiness, not trust, leads to beneficial cooperation (Hardin, 2006, p. 91), and anyone concerned with increasing social capital should be interested in the conditions of trustworthiness instead of trust. He believes that his account of trust as encapsulated interest and the accompanying notion of trustworthiness serve well in this context. According to this account, social capital in an interpersonal relationship is grounded in the rationality of maintaining a cooperative stance in the relationship. “This, of course, recommends the use of institutional capital as well as, to a likely limited extent, norms to secure cooperation at the interpersonal level” (ibid., p. 91). By institutional capital Hardin means for example “legal institutions that stand behind contracting” and “enable us to enter into exchanges that would be prohibitively risky without legal enforcement of relevant obligations” (Hardin, 2006, p. 88). With the narrower definition of trust the problem with statements of the above kind becomes apparent. Trust based on moral expectations and reliance based on institutional sanctions have different conditions, as Cynthia Townley’s and Jay Garfield’s paper in this volume shows. Both trust and reliance are crucial for the functioning communities. But quite likely the same measures cannot enhance both. Instead, they can even pull in different directions. Studies of social capital should attend to this tension. Take the example of friendship, an important social relation. We do not usually ask our friends to sign contracts where they assure that they will be sensitive to our dependence on them and take our interests into account in their practical reasoning. Such a measure would not have the kind of functions that trust is often believed to have: it does not decrease the complexity or costs of interaction. On the contrary, it would be a highly costly and complicated matter to draw up explicit contracts on the responsibilities of the partners in each of our friendships. This kind of proceeding might even have the cost of degrading or destroying the friendship. Trust is often thought to be a constitutive feature of friendship. But if we try to explain to our friends that we want to use “institutional capital,” in this case in the form of a contract, to make sure they are trustworthy, they might conclude that our relationships with them lack the trust necessary for friendship in the first place. This can only be explained by our different conceptions of trust and trustworthiness. The conceptions are different in kind and not only in degree, since what we take to be merely an extra measure to ensure a larger amount of trust in a relationship, is interpreted by our friends as a denunciation of trust. Our friends see trustworthiness as a moral matter, and trusting as recognition of the moral agency of another person. We have been guided by our conception

Trust, Trustworthiness, and Responsibility

13

of trustworthiness as “the capacity to judge one’s interests as dependent on doing what one is trusted to do” (Hardin, 2002, p. 8), and thought it a good idea to ensure our interests simultaneously by different methods. Our friends have interpreted the measures taken by us to imply that our attitude is not quite optimistic when it comes to them being persons who live up to their moral responsibilities. Instead of creating trust, which was our aim, we have come to destroy it. Similar mechanisms can work at different levels of social relationships, between individuals and between individuals and institutions. I think it is safe to conclude that an inherently moral notion of trustworthiness and a notion based on encapsulated interest are not easily combinable. With this conclusion I have achieved the main aim of this paper. I took Hardin’s claims concerning trustworthiness to indicate that it is extremely important to understand how trustworthiness works. I believe that Hardin is right to claim that there is a moral approbation going with trustworthiness, but noted that his account of trustworthiness does not fit this claim. After reviewing two quite influential philosophical accounts of trust I suggested an interpretation of trust as partly consisting of holding the trustee responsible. My hope is that this account brings out the moral underpinning of both trust and trustworthiness. This underpinning in turn leads to there being more to say about the direction of entailment between trust and trustworthiness than Hardin accounts for. This theme is, however, discussed in more detail in the other articles of the volume at hand.

WORKS CITED Baier, Annette. (1986). “Trust and Antitrust,” Ethics, 96:2 (January), pp. 231–260. Baker, Judith. (1987). “Trust and Rationality,” Pacific Philosophical Quarterly, 68, pp. 1–3. Ekstrom, Laura W., ed. (2001). Agency and Responsibility: Essays on the Metaphysics of Freedom. Boulder, Colorado: Westview Press. Flanagan, Owen J. and Amélie Oksenberg Rorty, eds. (1990). Identity, Character, and Morality: Essays in Moral Psychology. Cambridge, Mass.: MIT Press. Hardin, Russell. (2006). Trust. Cambridge, England: Polity Press. ———. (2002). Trust and Trustworthiness. New York: Russell Sage Foundation. ———. (1996). “Trustworthiness,” Ethics 107:1 (October), pp. 26–42. ———. (1991). “Trusting Persons, Trusting Institutions.” In Strategy and Choice, ed. Richard Zeckhauser. Cambridge, Mass.: MIT Press, pp. 185–209. Holton, Richard. (1994). “Deciding to Trust, Coming to Believe,” Australasian Journal of Philosophy, 72:1, pp. 63–76. Jones, Karen. (1996). “Trust as an Affective Attitude,” Ethics, 107:1 (October), pp. 4– 25. Strawson, Peter. (2001/1962). “Freedom and Resentment.” In Agency and Responsibility, ed. Laura W. Ekstrom. Boulder, Colorado: Westview Press, pp. 183–204.

14

FLOORA RUOKONEN

Thomas, Laurence. (1990). “Trust, Affirmation, and Moral Character: A Critique of Kantian Morality.” In Identity, Character, and Morality: Essays in Moral Psychology, eds. Owen Flanagan and Amélie Oksenberg Rorty. Cambridge, Mass.: MIT Press, pp. 235–238. Wallace, R. Jay. (1996). Responsibility and the Moral Sentiments. Cambridge, Mass.: Harvard University Press. Zeckhauser, Richard, ed. (1991). Strategy and Choice. Cambridge, Mass.: MIT Press.

Two TRUSTING INTERPRETATIONS Karen Jones Introduction The world looks different when viewed with trust from how it looks when viewed with distrust or neutrality. This difference matters for both practical and theoretical rationality. Trust opens up opportunities for action, for relationships, for inquiry, and for knowledge that would otherwise be closed off, in which lies both its power and its danger. Trust is able to work its magic, or wreak its mayhem, through the way that it changes how agents understand reasons for action or for belief. Seen with trust, a situation will contain quite different possibilities for action than will that same situation when viewed through a lens of distrust. Trust makes dependencies that might otherwise look too risky to be good bets come to seem reasonable. In this way, when well placed, trust fosters the goods of cooperation, but when badly placed it fosters exploitation and abuse. The core intuition that has driven all my work on trust to date is that trust is a lens that changes how agents understand their situation and the reasons it affords. It first led me to theorize trust in terms of an affective attitude; namely, an attitude of optimism that the goodwill and competence of the other will extend to cover the domain of our interaction, together with the confident expectation that the one-trusted will be directly and favorably moved by the thought that we are counting on him or her (Jones, 1996). If you think of trust as shaping perception and, in particular, shaping our perception of reasons, whether for action or belief, the move to thinking of it as an affective attitude is pretty obvious. Theorists of the emotions all agree that emotions shape perception and interpretation, including especially our interpretations of the reasons we have, and some (myself included) even think emotions are fully or partly constituted by such patterns of salience and interpretation (de Sousa, 1987). But objections to this move are pretty obvious, too (Blackburn, 1998): while a central range of cases of trust is plausibly characterized by an attitude of optimism about the goodwill of the other, the account appears too narrow to capture trust in all its varieties. While we never trust where we assume ill will, we can trust without assuming goodwill toward us on the part of the onetrusted. Moreover, if trust is an affective attitude that assumes goodwill, it follows that trust cannot be willed, something I once thought a virtue in an account of trust but now think a vice.

16

KAREN JONES

I was persuaded that trust could not be an affective attitude – or at least I said I was (Jones, 2002; 2004). But ghosts of the analysis of trust as an affective attitude continue to haunt work written after the official repudiation, most explicitly in my positing “basal security” – an underlying affective state that affects the salience and valence of risk and so in turn explains patterns in our particular trustings (Jones, 2004). Such refusal to revise even in the light of what I take to be excellent reasons looks like that pathology of excess selftrust, pigheadedness. Or maybe not; maybe the core intuition is right – or at any rate I cannot shake my commitment to it. The not quite so right move is to insist that trust’s close tie with the interpretation of reasons must signal affect. Perhaps sometimes there is an affective component to trust and sometimes not, but trust always works by shaping the agent’s interpretation of reasons. Thus, trust reveals itself in patterns in the agent’s interpretation of reasons – call these “trusting interpretations.” It is through these interpretations that trust is able to do its work, for good or for ill. That is the hypothesis I want to explore. I begin by cataloguing the ways in which trust and distrust shape interpretation in three key areas, two that concern practical rationality and one that concerns theoretical rationality. The areas are: interpersonal trust; trust considered as a character trait; and intellectual trust in oneself. Together they show the inadequacy of a simple cognitive account of trust that identifies trust with a belief or a judgment that the one-trusted is trustworthy. I then turn to the question of how to characterize the relationship between trust and trusting interpretations: does trust give rise to these interpretations, do the interpretations give rise to the trust, or is the relation constitutive? I argue that “trust” is used in both occurrent and dispositional senses, and that trusting interpretations are the occurrent manifestation of dispositional trust. But first a word about the title of this article: it is intentionally ambiguous, though I will only explore one of its meanings here. “Trusting interpretations” refers to the interpretations that, I will argue, are characteristic of trust, but it also invokes the question of the trust that we have in our interpretations. When these interpretations are themselves trusting interpretations, the second sense of the title raises the question of meta-trust, or when we should trust our trust. If we think of trust as primarily a matter of how we interpret reasons, then that is going to affect our assessment of when trust is justified and hence our assessment of appropriate meta-trust. These are important questions, but fall outside the scope of this article, with only an ambiguity in the title remaining as placeholder for them. 1. Trust and Interpretation Our trust explains action and belief by rationalizing them. When asked to explain why we let someone dangerously near something we care about, or why we believed something on the basis of someone’s say-so, we sometimes

Trusting Interpretations

17

say, “Because I trusted her.” From our point of view, our trust makes the action or belief seem reasonable, even though to someone who does not share our trust it might seem plain crazy. These explanations are not merely causal, as when we cite fatigue to explain why, say, we dropped something. Instead, they are explanations designed to reveal the favorable light in which we viewed the action in question, or the grounds that we took ourselves to have for our belief. Where these explanations concern action, a simple belief-desire model might be thought to capture their structure: I have a goal or end, achieving which requires another agent to do Z; I believe that that person is trustworthy with respect to action Z, hence I rely on their performing action Z. Trust, on this simple model, just amounts to the belief that the other is trustworthy with respect to an action and so can be counted on to perform it. Different accounts of trust cash out the content of this belief in different ways. For example, Russell Hardin’s encapsulated interest account cashes it out in terms of the belief that it is in the other’s interest to act in my interests in this matter, typically because they have an interest in maintaining our relationship (Hardin, 2002). Where these explanations concern belief, a simple cognitive model might likewise be thought to capture their structure: I believe that you are cognitively reliable in a domain and that you are honest (here and now) and so I come to form a belief on the basis of your say-so. My trust just is these paired beliefs, which together attribute trustworthiness to you, and which explain and justify the testimony-based belief. In the case of intellectual selftrust, I believe that one of my cognitive mechanisms or belief-forming methods is reliable and so accept its deliverances. My trust just is this belief about reliability. In this section I use case studies of interpersonal trust, trust as character trait, and intellectual self-trust to lend support to the claim that belief accounts are inadequate to explain how trust functions. Our trust can come apart from our beliefs about trustworthiness. But even where trust does not come apart from judgments about trustworthiness, trust operates at a different level from them. It is vulnerable to our interpretive habits, often shaped by our emotions, and sometimes able to be willfully controlled in a way judgment cannot be. Attention to our lived experience of trust reveals a close tie between trust and our interpretation of situations. A. Interpersonal Trust Particular trust relations, which hold between a specified truster and a determinate trustee, are the focus of much of the philosophical literature on trust. Truster and trustee are usually individuals but may be groups of individuals or institutions (the government, the military). The relation appears two-place – “A trusts B” – but the superficial grammar is misleading; there is

18

KAREN JONES

now general agreement that particular trusting relations contain tacit reference to a limit and so are three-place. This three-place relation can be spelled out in different ways: “A trusts B with valued thing C” (Baier, 1986), “A trusts B to do Z” (Holton, 1994) or “A trusts B in domain of interaction D” (Jones, 1996). I will work with the “in a domain” formulation, but nothing much hinges on this. In the standard case of three-place trust, the trust relation is founded on a perhaps explicit but often tacit and unreflective acceptance of the other’s trustworthiness. Consider, for example, the relationship between Othello and Iago (de Sousa, 1987, pp. 195–96; Jones, 1996, pp. 11–12): Iago is able to dupe Othello because he enjoys Othello’s trust. (It is another, and more complicated question, how Othello came to trust so unwisely – for, despite what William Shakespeare says, Othello is not the tale of someone “that lov’d not wisely, but too well” but the tale of someone who failed spectacularly in his trust and distrust.) To the outside observer it is evident that Iago is, or might be, manipulating Othello; we do not need his asides to let us in on the plot. But Othello interprets Iago’s words through the lens of his trust. Where others might see reason for wary suspicion – “Why is he telling me this? What can he hope to gain from coming between Desdemona and me?” – Othello sees only the reluctant admissions of a trustworthy adviser, admissions wrung from him by his loyalty and love for Othello. Othello’s trust in Iago closes down potential lines of inquiry that would have seemed worth investigating by someone similarly situated who had a neutral attitude, and downright compelling to someone who distrusted. What Shakespeare describes is commonplace: an ambiguous action – and are not most actions ambiguous, given the opacity of human motive? – will be given a positive interpretation where there is trust and a negative interpretation where there is distrust. In this way, trust and distrust tend to be self-fulfilling. We see evidence that confirms our trust or our distrust because we see that very evidence through the lens of our trust or our distrust. Trust thus functions to reduce deliberative complexity (Luhman, 1973) because it lets us set aside considerations that might be reasons for action to protect our interest, since it appears we can safely assume they do not obtain. I do not have to consider what to do if others are seeking advantage from me, what to do if they mean me ill, because I assume that they are not. A proponent of the simple belief-desire model might claim that the patterns of interpretation described in the Othello example can be fully explained by the belief that Iago is trustworthy. Grant that it is through its effects on Othello’s perception of reasons that trust works, but those effects are themselves to be explained by the belief, which offers a deeper explanation of the interpretations though which trust works. This reply fails, however, for the belief that someone is trustworthy can be had without it giving rise to trusting interpretations and trusting interpretations can be had without belief in trustworthiness. The belief is neither sufficient nor necessary

Trusting Interpretations

19

for the interpretations. Othello could believe that Iago is trustworthy and yet that belief does not shape perception of reasons in the way the objection supposes it does. If Othello were prone to wary suspicion, he might judge that Iago is trustworthy yet still have thoughts that betray a lack of trust. Despite his belief, he would be able to escape Iago’s clutches as his perceptual habits protected him from his mistaken belief. Our second Othello – an Othello who believes Iago is trustworthy but has perceptual habits characteristic of generalized wary suspicion – will have psychological forces that block the formation of the patterns of interpretation characteristic of trust even where they concern those, like Iago, who he believes trustworthy. By following up on these alternative interpretations, wary-Othello may come to revise his belief in Iago’s trustworthiness, but he need not. He may continue to hold that judgment, but his interactions with Iago will be affected by his generalized wariness. The bare belief in Iago’s trustworthiness is insufficient to explain how Othello came to be so readily duped. The entrusting behavior that is characteristic of trust only occurs when that belief is accompanied by trusting patterns of interpretation. I discuss analogous cases in the next section. Focusing on the role of trusting-interpretations in explaining entrusting behavior has a further advantage: it lets us make sense of therapeutic trust as a genuine kind of trust. Therapeutic trust is trust without an accompanying belief or presupposition that the one-trusted is trustworthy. In therapeutic trust of another we display our trust and signal our vulnerability even while recognizing that it is possible, perhaps even likely, that the other person will take advantage of that vulnerability. We do so in the hope that by having and displaying trust we can eventually bring about trustworthiness on the part of the other (Horsburgh, 1960). Thus, a mother might decide to leave the house in the care of her irresponsible teenager despite being fully aware that the daughter might yet again prove herself untrustworthy. But she hopes a strategy of trusting, systematically pursued, will in the end bear fruit (Jones, 2004). Sometimes we are the targets of our therapy. Consider the “trust game,” a theater exercise designed to build trust in a group (Holton, 1994). In the trust game you fall backwards into the arms of the group who are to catch you before you hurt yourself. At least when starting to play the trust game, the risks are salient to you: you are aware of the possibility that you might be dropped. (The game does not stop just because of a single failure, so you might even be vividly aware of this possibility, having just witnessed it.) You commit to falling nonetheless. In so committing, it appears you decide to trust, just as the mother decides to trust the teenage daughter by leaving the house in her care. The director stages the trust game to elicit trustworthiness from the group, but you play it to change your response to your vulnerability, to learn how to “let go.” Games of “catch me!” are not restricted to children and theater groups; we play variants of them whenever we will ourselves to

20

KAREN JONES

trust despite salient risk. There is dispute over whether therapeutic trust is a form of genuine trust at all – perhaps it is only “as-if” trust (Ruokonen, 2008): therapeutic trusters act as if they trust, but they do not really, and their pretence, if revealed, would be likely to undermine their goal. Only if we think of trust as entrusting – as action – will therapeutic trust and ordinary trust be alike, given the differences in cognitive states between the ordinary truster and the therapeutic truster. But, continues the argument of those who deny therapeutic trust is real trust, there are good reasons for not thinking of trust as the act of entrusting. If trust is the act of entrusting, then it cannot explain why people act as they do, but trust should be able to explain action. Thus, therapeutic trust is no real trust at all. There are two ways of thinking about what happens to our perception of reasons in therapeutic trust. On one model, the outweighing model, the objectors are right: so-called therapeutic trusters and ordinary trusters have only action in common. On the other model, a bracketing model, therapeutic trusters do share something in common with ordinary trusters. They share a “trusting interpretation” in which considerations of risk are not taken to be reason giving. It is because they share this trusting interpretation that both merit being called trust, or so I will argue. On the outweighing model, a therapeutic truster perceives the risk and takes it to be reason-giving, but judges that the risk is outweighed as a reason by the value of the good sought from entrusting (greater openness, trustworthiness on the part of the other). So long as the ratio of risk to possible benefit is such that entrusting brings greater expected utility than not entrusting, entrusting can be rational, even though the therapeutic truster does not believe that the one-trusted is trustworthy as yet, and even though the truster does not expect a favorable outcome immediately. On the bracketing model (McDowell, 1979; Herman, 1991), therapeutic trusters perceive the risk, but set it aside as a reason. Though they accept that in general vulnerability is reason to protect themselves, they embrace (or do not turn from) that vulnerability, here and now. Vulnerability is bracketed as a reason. Their understanding of the reasons that obtain for them in the choice situation thus shares something in common with ordinary trusters who take vulnerability not be a reason against entrusting. The difference is why they take vulnerability not to be a reason, with one group confident that there is no real vulnerability that needs to be taken into account, and the other lacking this confidence but willing to not let vulnerability play its usual role in their deliberation. It might be objected that though the bracketing model makes therapeutic trust have something genuinely in common with ordinary trust, it does so at the cost of making therapeutic trust incomprehensible. Why would anyone ignore the reason-giving force of vulnerability? Are there any real-world cases of therapeutic trust, as I have described it?

Trusting Interpretations

21

The reason we ignore the reason-giving force of vulnerability is that we seek a relationship, or way of being in the world, in which vulnerability is not deliberatively salient. I undertake therapeutic trust to enter into a trusting relationship with the other, or to make myself bolder, more carefree, and less self-protective. But it is not possible to have a trusting relation with another, or to be bold and carefree, while having one eye out for risk. The conflict here is intrinsic and not merely contingent: insofar as I have an eye out for risk, I fail to be carefree; insofar as I have an eye on the harm you might do to me, I fail to trust you. Thus, having noted the risk, I commit to setting it aside as a reason because seeing it as a reason would undermine the very project I am embarked on; namely forging the kind of relationship, or mode of being, that is characterized by vulnerability not having its usual reason-giving force. Do we ever reason like this? I think evidence exists that we do, though I take no stance on how common it is. Let us work through the mother and teenage daughter case in more detail to see what the two models say about it and to see whether we can coherently describe it in terms of the bracketing model. In the weighing model, the mother recognizes that there is a risk the daughter will again be irresponsible and that she may return to a huge cleaning task and complaints from the neighbors. Let us suppose she thinks this outcome is more likely than not. Nevertheless, she thinks that the daughter has the potential to become responsible and can only develop that potential by being given responsibility. Because the mother recognizes vulnerability as a reason, if it were possible to bring about the same outcome while reducing or eliminating that vulnerability, she would choose to do so. Other things being equal, any action option that recognizes the reason-giving force of both the long-term goal and the risk is to be preferred over one that only recognizes the long-term goal. It is for this reason that there can appear to be something devious or unstable about therapeutic trust: what matters is not that the daughter actually be given responsibility, but that she believes that she has been given responsibility. Provided the mother can be certain that there is no way the daughter will learn of it, she would take “side insurance” against the worst of her daughter’s potential irresponsibility by, for example, having someone keep an unobtrustive eye on her. And that is why some think therapeutic trust necessarily deceptive and unstable. Consider, by contrast, the reasoning of a mother who sees her choice situation in terms of the bracketing model: she sets aside the risk that her daughter will be irresponsible and that she may have to clean, or repair, the house and restore relations with the neighbors. She recognizes the risk, but does not take it to be reason-giving because to do so would be incompatible with the relationship that she seeks with her daughter. Even if there were a way of minimizing this risk while still achieving her long-term goal, she would not take it because her deliberative problem is not how to find an action option that recognizes both risk and long-term goal. I think mothers – and others! – do sometimes reason in this way. Nor is there anything deceptive or

22

KAREN JONES

unstable about therapeutic trust, so understood. If these reflections are along the right lines, then three-place trust involves (I deliberately use a vague word) perception of reasons in two ways: first, considerations that without trust might have been salient will not be salient. One will assume that considerations that support self-protective responses are not present in the circumstances; second, though more rarely, one will recognize such considerations but not take them to have their usual reason-giving force. These trusting interpretations rationalize trusting behavior. They are the ground that enables both the goods of cooperation and the harms of the well-executed con. B. Trust as Character Trait We see trusting interpretations at work not only in three-place trust relations, but also in trust considered as a character trait, where it is even clearer that their work cannot be accomplished by belief. Character trust has been largely overlooked in a literature that has been focused on particular trust relations. However, reflection on the aftermath of random violence shows that not all relations that have a claim to be called trust have three-place structure. Threeplace trust cannot capture what we have in mind when we call someone “trusting,” or what goes on when, in the aftermath of random violence, someone says, “my trust in the world was shattered.” In the aftermath of trauma, especially in the aftermath of an attack so random that the victim cannot blame his or her practice of trust for it (it could have happened to anyone, to any woman, black, etc.), the landscape of threeplace trust is radically transformed; where there was once trust, now there is not. These changes cannot always be explained by changes in an agent’s judgment about the objective level of risk he or she faces before and after the trauma. (Though this sometimes happens – after air crashes, murders, or terrorist attacks have been in the news, especially when the coverage is graphic, or after first-hand or close second-hand experience of such things, there can be cognitive distortion in our judgments of risk as salience shifts probability assignments.) Someone can know all there is to know about objective risk and judge that there has been no change in these risks, yet find that he or she is no longer able to trust in the face of a degree of risk that was once of no concern to him or her. This transformation is common among survivors of rape. Before and after an attack, a woman may hold the same set of judgments about the risk of rape, yet her experience of safety and belonging in the world be entirely transformed, and with it her willingness to trust. Willingness to trust need not, and often does not, covary with judgments about degree of risk: someone can willingly trust despite judging the risk is comparatively high, while be unwilling to trust even when judging the risk is low. The observed effects of trauma on three-place trust point to the existence

Trusting Interpretations

23

of what I call “basal security” (Jones, 2004). By basal security I mean an affectively laden, implicit interpretive framework governing how we respond to the vulnerability to harm from other agents that is characteristic of human experience, given we are finite and social beings. When we ascribe an implicit interpretive framework to an agent, we attribute to him or her a set of dispositions of salience, interpretation, motivation, and affect. In the case of basal security, these interpretive frameworks concern vulnerability, safety, and risk, as these relate to human agency. Differences in these frameworks and the dispositions that constitute them, give rise to differences in the way the world is experienced. There is evidence from the study of generalized anxiety disorder and other emotional disorders that interpretive schema operate before the conscious cognitive processing involved in judgment has had a chance to process a stimulus (Öhman, 1993; Mathews and Wells, 1999). They thus affect conscious processing without themselves being conscious and so are able to operate in opposition to conscious reflective judgment. The reason why we need to posit basal security is to explain observed dissonance between intellective judgment regarding the degree of risk present in a situation and a subject’s willingness to trust in the face of such-and-such a degree of risk. This dissonance often plays itself out emotionally, so that we are anxious beyond what we take the situation to merit, or calmly oblivious to risk we know on some level is there. Basal security affects the kind of things agents pay attention to and how they interpret the reasons a situation affords. Consider people with low basal security: they will tend to scan the environment for signs of risk and danger and interpret ambiguous situations as threatening. Post-traumatic stress disorder presents an extreme in low basal security and is characterized by vigilance, heightened startle responses, fear, and anxiety. Low basal security changes the perceived practical significance of a given degree of risk. A given degree of risk might not enter at all into the practical deliberation of someone with high basal security: the risk might not be noticed, or considered important. If risk does enter deliberation it may figure as challenge instead of threat. Someone with low basal security, by contrast, lives in continual awareness of vulnerability. Not only is that vulnerability highly salient, it is strongly negatively valenced, as risk engages emotionally and motivationally. Those with different levels of basal security who are similarly situated perceive their situations differently. The same situation seen with low basal security will appear to afford fewer opportunities for cooperation than it would were it seen with higher basal security. In this way, basal security shapes three-place trust. C. Intellectual Trust in Oneself “Intellectual trust” is trust in one’s own or another person’s judgment, belief, faculties, or methods of inquiry. My focus here is on intellectual self-trust. On

24

KAREN JONES

a simple cognitivist picture, intellectual self-trust is a matter of believing that one is reliable in a domain and so being willing to rely on the deliverances of one’s intellectual faculties. I present two extremes of self-distrust and selftrust to argue that this account is too simple: like trust in particular others and trust-as-character trait, intellectual self-trust works by changing patterns in the agent’s perception of reasons. In intellectual trust, the relevant patterns concern reasons to revisit judgment, withhold it, or to defer to the judgment of others. Intellectual self-trust operates at a different level to judgment about reliability and is shaped by both our emotional state and our habits of perception. The first example shows the shaping role of emotional state; the second, the role of perceptual habit. As an extreme of intellectual self-distrust, consider this everyday bit of neuroses: on the way to the airport I keep checking and rechecking my bag to reassure myself that my tickets and passport are still there. Checking my bag for the fifth time (third time? surely double-checking is ok) shows an extreme pathology of self-trust, a pathology created by anxiety. There has been no intervening reason to revisit the earlier judgment that I have remembered the tickets, a judgment formed, and again confirmed, on the basis of evidence from my senses. Since I last checked for the tickets, I have not been in a crowded place where there might be pickpockets. I am sitting alone in a taxi. It is not that I think that my memory might suddenly have failed or that my eyes are no longer to be trusted. There is no real reason for me to keep revisiting my judgment – and that is my problem. If I doubted for a reason, I could re-examine the basis for my belief and reassure myself that my judgment was correct, that the reason for doubt did not after all undermine my justification. My doubts are not open to rational resolution because they are not formed in response to perceived reasons. If the evidence of my senses and memory will not resolve the matter for me, then appealing to them again through yet more checking cannot resolve it either, so I tell myself sternly, “the tickets are in the bag, stop it!” The neurotic checker and re-checker has lost the capability to filter potential reasons for revisiting a judgment – a bare possibility (what if they are not there?) has come to seem a reason to seek further reassurance, but such reassurance is at best temporary, since that bare possibility will always remain. The neurotic checker sees reasons to revisit judgment where someone with levels of self-trust not distorted by anxiety would not. At the other end of the continuum from the neurotic self-distruster are people pig-headedly committed to their judgment. Where responsible inquirers would see reason to revisit and revise or withhold judgment, or defer to someone else, the overly self-trusting sees no such reason. Other people’s disagreement, their track record of epistemic failure in a domain, all these are not salient, or not taken to be reasons here and now. The problem with the overly self-trusting need not lie in their beliefs about their reliability in a domain; it can just as easily lie in their perceptual

Trusting Interpretations

25

habits. Consider a man who enjoys the privilege of epistemic credibility: he is white, well educated, and articulate. He is accustomed to having his testimony believed and to having others defer to him in their judgment. Suppose that he comes to accept that he has gaps in his epistemic capacities; perhaps, for example he becomes convinced that he is quite frequently mistaken in his assessment of the contributions of his female colleagues. That belief might nevertheless fail to get purchase in routine epistemic practices: he might, caseby-case, fail to see that he has reason to take extra care or to revisit or defer to others in making judgments about female colleagues. Nor need this indicate any straightforward failure of rationality: there is no inconsistency in believing that you are unreliable in a domain, yet not thinking that in this particular circumstance, these particular reasons are grounds to reconsider, defer, or withhold judgment. The tension between judgment and perception emerges not in the particular case, but in the patterns of perception of reasons. One can have a pattern of perception that would be defensible only on the supposition of reliability without having the corresponding belief. The same applies to the overly self-distrusting. Deficiencies in self-trust are often socially generated and those in socially subordinate positions are especially likely to have them. Reminding yourself that you are competent in a domain is often not enough to address self-distrust, for it is possible to be generally reliable, yet here and now be making a mistake. If we think of intellectual self-trust as primarily cognitive – as matter of belief about reliability – we risk overlooking the politics of self-trust. Just as there is a politics of credibility (Jones, 2002), there is a politics of self-trust. The social conditions that create and sustain self-trust can create and sustain pathologies of self-trust and self-distrust. We need to reflect on the ways self-trust and distrust works through shaping the agent’s perception of reasons if we are adequately to understand the pathologies of self-trust and how to rectify them. 2. Trust and Trusting Interpretations I have argued that three central kinds of trust – interpersonal trust, character trust, and intellectual self-trust – work through trusting interpretations. Trust works by changing how we interpret reasons, whether for action or belief, and this in turn explains our willingness to take cognitive or practical risks. The question remains: “What is the relationship between trust and trusting interpretations?” Are trusting interpretations the mechanism by which trust works, or are they somehow fully or partly constitutive of trust? There are at least three prima facie plausible, but apparently mutually incompatible, ways of thinking about the relationship between trust and trusting interpretations. It makes sense to say both: Trust to interpretation (TI): We interpret the world as we do because we trust.

KAREN JONES

26

And Interpretation to trust (IT): We trust because we interpret the world as we do. TI says that our trust gives rise to or explains these trusting interpretations. The interpretations in turn explain why we let ordinarily flawed human beings get near things we care about, why we willingly engage, without demand for a safety net, in all kinds of cooperative activities. We act as we do because we see the world as we do; we see the world as we do because we trust. TI appears supported by many of the examples discussed in Section 1: Othello interprets Iago’s words as he does because he trusts Iago; the rape victim struggling with low basal security in the light of her trauma sees situations as containing threat where once she did not and, as a result, her actions are constrained. Similarly, the overly self-trusting person sees no reason to withhold judgment where others would because he trusts himself; his misplaced self-trust explains why he readily forms false beliefs in the domain in question. The reverse story – IT – appears plausible, too: we act as we do because we trust, and we trust because we see the world as we do. The interpretations give rise to the trust and the trust in turn explains our willingness to engage in cooperation with all its attendant risks. This direction of explanation is supported by the example of mother’s therapeutic trust of her teenage daughter. She trusts because she sets aside concern about risk and so is moved to take no further action to protect her from that risk even if she could do so without jeopardizing her long-term goal of fostering trustworthiness in her daughter. Because she has set aside these considerations, she acts as she does. Iago’s manipulation of Othello appears to work, in part at least, on account of having this structure: Othello has a trusting interpretation of Iago that attributes to him no ulterior motives, which explains why he trusts him and why he falls into the trap. Both TI and IT appear plausible and both seem to ascribe a causal relation between trust and trusting interpretations, with the disagreement being over the direction of causation. Look more closely at the examples, however, and the relationship between trust and trusting interpretations begins to appear constitutive: what makes it true that the mother trusts her daughter, or that the rape survivor no longer trusts men? The (dis)trusting interpretations they have. These interpretations just are the (dis)trust and they explain the subsequent action. Thus, there is a puzzle about how we are to think of the relation between trust and trusting interpretations, with three tempting but apparently conflicting ways of thinking about their connection. I think that there is an interpretation of TI and of IT that makes them both true and that there is also a constitutive relation between trust and

Trusting Interpretations

27

trusting interpretations; that is, we can preserve all three ways of talking about the connection between trust and trusting interpretations. To do this, we need to recognize that “trust” is used in both dispositional and occurrent senses. “Trust” can name both a standing state or attitude and an occurrent state. Trusting interpretations are the central way in which dispositional trust manifests itself occurrently; when “trust” names an occurrent state it names these trusting interpretations. Trusting interpretations are the occurrent face of trust, by definition: were someone not to be disposed to trusting interpretations then he or she could not be said to trust. Thus, a disposition to trusting interpretations is partly constitutive of trust. Dispositional trust is multi-stranded and also manifests itself in reactions (including especially dispositions to reactive attitudes (Holton, 1994)) and feelings, such as security and the absence of anxiety. However, trusting interpretations are the key dispositional strand because it is through them that trust is able to play its explanatory role in action and belief. Trust, understood as a disposition, explains trusting interpretations in the same way that fragility explains why a glass breaks. It is this dispositional sense of “trust” that is invoked in the examples that lend support to the trustto-interpretation reading of the relationship between trust and trusting interpretations (TI). The occurrent versus dispositional distinction also enables us to cash out the reverse interpretation-to-trust reading (IT). It might be objected that it is only because we sometimes use “trust” to name the act of entrusting that this reverse reading makes sense: our trusting interpretations explain why we are willing to entrust. But given I have rejected accounts of trust that equate it with the act of entrusting, I can make no good sense of the claim that trusting interpretations explain trust. In reply, we need to consider the central role of trusting interpretations in trust: they are not only the means by which trust, once activated, explains behavior and belief, they have a central causal role to play in explaining how dispositional trust gets established, maintained, and extended. It is this role that is highlighted in the interpretation-to-trust reading of the relationship between trust and trusting interpretations. Consider the mother engaged in therapeutic trust: she has as yet no standing disposition to trust her daughter. Risk is salient to her, but she elects to set it aside as a reason, and so comes to have a trusting interpretation of the practical reasons her situation affords. This trusting interpretation is what she shares in common with a mother who is otherwise similarly situated but believes her daughter trustworthy, and so they act the same. When we are trusted we seek to live up to that trust: being trusted typically sets in train trust-responsiveness (Pettit, 1995) and that, in turn, establishes or further entrenches dispositional trust. There is a “start-up” problem for trust, but once it gets a foothold it can quickly become entrenched. Trusting interpretations explain why: if we view someone through a lens of trust, we engage in cooperative activities, which, when successful, establish dispositional trust, which in turn further supports trusting

28

KAREN JONES

interpretations and makes us more tolerant of ambiguous actions than we otherwise would be. Any adequate account of trust must recognize the role of trusting interpretations for it is through them that trust is both manifested and entrenched.

WORKS CITED Baier, Annette. (1986). “Trust and Antitrust,” Ethics, 96:2 (January), pp. 231–260. Blackburn, Simon. (1998). Ruling Passions: A Theory of Practical Reason. Oxford, England: Clarendon Press. Dalgleish, Tim and Mick Power, eds. (1999). Handbook of Cognition and Emotion. Chichester, England: Wiley. Greenspan, Patricia. (1988). Emotions and Reasons: An Inquiry into Emotional Justification. New York: Routledge. Hardin, Russell. (2002). Trust and Trustworthiness. New York: Russell Sage Foundation. Herman, Barbara. (1991). “Agency, Attachment, and Difference,” Ethics, 101:4 (July), pp.775–97. Holton, Richard. (1994). “Deciding to Trust, Coming to Believe,” Australasian Journal of Philosophy, 72:1, pp. 63–76. Horsburgh, H.J.N. (1960). “The Ethics of Trust,” Philosophical Quarterly, 10:41 (October), pp. 343–354. Jones, Karen. (1996). “Trust as an Affective Attitude,” Ethics, 107:1 (October), pp. 4– 25. ———. (2002). “Trust: Philosophical Aspects.” In International Encyclopedia of the Social and Behavioral Sciences, eds. Neil Smelser and Paul Bates. Amsterdam, Netherlands: Elsevier Science, pp. 15917–15922. ———. (2004). “Trust and Terror.” In Moral Psychology: Feminist Ethics and Social Theory, eds. Margaret Walker and Peggy DesAutels. New York: Rowman and Littlefield, pp. 3–18. Lewis, Michael and Jeannette Haviland-Jones eds. (1993). Handbook of Emotions. New York: Guilford. Luhman, Nicklas. (1973). Trust and Power. Chichester, England: Wiley. Mathews, Gerald and Adrian Wells. (1999). “The Cognitive Science of Attention and Emotion.” In Handbook of Cognition and Emotion, eds. Tim Dalgleish and Mick Power. Chichester. England: Wiley, pp. 171–192. McDowell, John. (1979). “Virtue and Reason,” The Monist, 62:3 (July), pp. 331–50. Öhman, Arne. (1993). “Fear and Anxiety as Emotional Phenomenon.” In Handbook of Emotions, eds. Michael Lewis and Jeannette Haviland-Jones. New York: Guilford, pp. 709–729. Pettit, Philip. (1995). “The Cunning of Trust,” Philosophy and Public Affairs, 24:3 (Summer), pp. 202–225. Ruokonen, Floora. (2008). “Building Trust: A Fairly Honourable Defeat,” Sats – Nordic Journal of Philosophy 9:1, pp. 46–68.

Trusting Interpretations

29

Smelser, Neil and Paul Bates, eds. (2002). International Encyclopedia of the Social and Behavioral Sciences. Amsterdam, Netherlands: Elsevier Science. de Sousa, Ronald. (1987). The Rationality of Emotion. Cambridge: MIT Press. Walker, Margaret and Peggy DesAutels, eds. (2004). Moral Psychology: Feminist Ethics and Social Theory. New York: Rowman and Littlefield.

Three TRUST IN WITTGENSTEIN Olli Lagerspetz and Lars Hertzberg Introduction There are important reasons why the work of Ludwig Wittgenstein ought to be central to any philosophical study of trust. This is partly because of what he explicitly says about this and related issues (such as knowledge and certainty), and partly because of the applicability of his more general insights. Wittgenstein’s different contributions to philosophy are, generally speaking, related to certain basic insights about the role of language. He thinks we should not presuppose that the meaning of a given expression is established by identifying its referents, the phenomena that it stands for. Instead, we should look at the uses of the expression in different contexts of human life; in particular, its different uses. For our present purposes, it is profitable to look at Wittgenstein’s discussions of psychological expressions. He encourages the reader not to engage in introspection in order to capture the essence of given psychological states. Applying this to the present case: What we need to realize here is that, by invoking the language of trust and betrayal, we do not simply identify facts out there. Instead we invoke a certain perspective. We are invited to see someone’s behavior in a certain light. Thus the question to ask is not, “what are the proper criteria for applying the word ‘trust’ to psychological or behavioral phenomena?” but instead, “what is the role, in human interaction, of speaking about trust?” The aim of the present paper is to present relevant strands of Wittgenstein’s work and to relate them to the contemporary discussion on trust. Three main Wittgensteinian themes will be identified: (1) a general methodological point about how to investigate psychological expressions; (2) a critique of the idea that our grasp on reality can ultimately be spelled out by means of factual assertions; and (3) Wittgenstein’s own emphasis, especially in On Certainty, on the logical and epistemic role, in learning and judging, of our dependence on other people. However, in our discussion we will be going beyond what Wittgenstein said. We hope to show that his methods and insights may be fruitfully used to shed light also on themes that he did not address directly.

32

OLLI LAGERSPETZ AND LARS HERTZBERG 1. On Psychological Expressions A. Is Trust a Psychological State?

Grammatically, trust is a state that an individual is in, a state somehow relating to another individual (as in “X has trust in Y”). However, so far this description does not specify what kind of a state trust is. By “state”, we may mean a person’s current state of mind or body, but also dispositional states, even such “states” as that of being rich or poor. The point of speaking of a state in all these cases is that some lasting state of affairs pertaining to an individual is used as a point of reference when something about the individual’s life is explained or judged. Thus the fact that I rely on someone’s promise may, in some context, be explained and judged in terms of the fact that I trust her. But is trust a psychological state? For instance, when, if ever, can a feeling of trust be identified? (Those who speak of trust as a feeling include Giddens, 1991, p. 36; Sellerberg, 1982, pp. 40, 45, 46; Govier, 1993a, p. 156.) According to Annette Baier, trust “has a special ‘feel,’ most easily acknowledged when it is missed, say, when one moves from a friendly ‘safe’ neighborhood to a tense insecure one” (Baier, 1994, p. 132). However, an obvious problem with this description is that it does not specify a feeling of trust. Baier is, instead, describing a case of distrust. This highlights a general difficulty about pinpointing specific feelings or thoughts that might constitute trust as a psychological state. Often trust is, on the contrary, characterized by the absence of certain feelings and thoughts such as suspicion and fear. Also, when I trust someone it is not, for the most part, something I think about; rather the opposite, as we will see. Nor do I mostly do so because I have, at some point in time, formed the judgment that the person in question is trustworthy. If I am short-changed at the grocery store, but only notice it afterwards, it will be correct to say I had trusted the shop attendant to hand me the correct change. However, perhaps I had never addressed the question in my mind. And I need not have harbored specific feelings about the shop attendant either. In his Philosophical Investigations, Wittgenstein takes up an analogous question. He discusses the “feeling of familiarity” that may strike us, for instance, when we enter the same old room after a long absence. He points out that, apart from such situations, familiar objects in our surroundings do not typically give us a feeling of familiarity. It is easier to get at a feeling of strangeness at the sight of unfamiliar objects, even though we do not have that feeling whenever we see something unfamiliar. (Wittgenstein, 1953, I: § 596.) Thus even if feelings of familiarity do exist, their presence or absence is not typically the criterion of what is familiar to us. This is part of a more general theme addressed in Zettel. In that work, Wittgenstein distinguishes between cases of believing, feeling, thinking, and

Trust in Wittgenstein

33

so on, that involve “genuine duration” and others that do not. (Wittgenstein, 1967, §§ 45, 78, 81; see also §§ 46–47, 50, 76–77, 82–83, 85.) A psychological state has genuine duration if it makes sense to ask whether the state endures from one moment to the next. We can imagine asking “do you still feel the pain?” and a few minutes later, “what about now?” Also, if a state has genuine duration, we can imagine it being interrupted. Being in pain is a clear case of genuine duration. On the other hand, suppose I ask you “are you still a Republican?” and again, after five minutes of silence, “what about now?” This could hardly be understood as anything but a joke unless you have received some quite extraordinary revelation in the meantime. Nor could we imagine someone being interrupted in being a Republican. The state of being a Republican, then, does not have genuine duration (For a discussion of genuine duration, see Malcolm, 1984, pp. 1–101; esp. p. 79ff). Anything that can be measured with a stopwatch would be a clear case of genuine duration. This may not always be possible, though, contrary to what Wittgenstein suggests, since the beginning or end of the state may be diffuse, and thus it may not be clockable. If the answer to the question “Do you still feel it?” is “I’m not sure” that does not necessarily rule out its being a case of genuine duration. For instance, sleep, physical pain, intense thinking, and intense expectation have genuine duration. Knowledge, ability, understanding, and intention, on the other hand, do not. (Wittgenstein, 1967, §§ 82, 45, 50.) They are not interrupted when a person is asleep or in a faint. One can be interrupted in thinking or planning, but not in intending (Wittgenstein, 1967, §§ 50). An important connecting theme in the Investigations is Wittgenstein’s emphasis on the fact that in many cases, the intelligibility of attributing a feeling or attitude to someone is dependent on that person’s overall situation. Thus we can meaningfully attribute love, or hope (Wittgenstein, 1953, I: § 583), or grief (Wittgenstein, 1953, p. 174) to a person only because that person’s situation as a whole makes these descriptions applicable. We cannot correctly and intelligibly attribute feelings of grief to a person unless he or she has, for instance, recently lost a loved one, regardless of what else may be true about that person’s states of mind at a given point in time. The presence or absence of a mental state at a given point in time does not settle the issue on its own. For instance, someone at a cinema may feel like crying at the death of the film’s (fictitious) main character, but this will not qualify as genuine grief unless some quite peculiar explanation is produced. Let us here interject a point of methodology. The foregoing remark is not an attempt to exclude unexpected cases by setting up ordinary language as a norm for language use, but a reminder of the fact that intelligible uses of the word “grief” presuppose some plausible anchoring in the person’s situation. Thus we do not suppose that limits of intelligibility can be laid down in the abstract. In suggesting that it might not be intelligible to say certain things in a certain type of situation, we are appealing to the readers’ ability to

34

OLLI LAGERSPETZ AND LARS HERTZBERG

imagine themselves confused if someone were to utter certain words in those kinds of circumstances. This does not entail that their confusion would have to be irremediable. For instance, they might come to understand the words as a joke, a metaphor, a sign of poor command of English, or the like. Still, the fact that our initial response would be one of confusion may help draw our attention to certain features of the use of the word. Turning to trust in the light of these considerations, it appears clear that the state of trusting does not have genuine duration. In other words, trust is not for the most part manifested as a particular state that occupies one’s mind. The presence of trust must instead be established by looking for an overall pattern in a person’s thinking and acting: a pattern in the weave of life, to apply what Wittgenstein said about grief (Wittgenstein, 1953, p. 174). Someone’s trust in another may show itself in him or her being relaxed in the other’s company, as well as in things he or she does not do, such as not taking certain precautions. It may show in the person’s thinking about the future, perhaps plans that rely on information from the trusted person; or perhaps in the complete absence of a plan where one might be expected. But it does not need to involve specific feelings toward the trusted one or explicit thoughts about him or her. Again we do not deny that, in the right circumstances, it may be natural to describe someone’s state of mind as involving feelings of trust. But the presence or absence of certain psychological states does not settle the issue on its own. On the contrary, the feelings can plausibly be described as ones of trust only if the appropriate context is in place. B. Is Trust a Disposition? At this juncture, someone might suggest a way out of the problem. The reader might think that trust is, or involves, a behavioral disposition instead of (merely) a psychological state. Views along these lines are mostly adopted in current literature. Trust is defined in terms of expectations of a cooperative disposition in others; expectations that may or may not be accompanied by specific feelings. On that view, my trust in a person involves my explicit or implicit expectation that the person will not take advantage of me but will, instead, behave in ways that are beneficial or at least not harmful to me. (Baier, 1986, p. 235; Gambetta, 1988, p. 217, Govier, 1992, p. 17; 1993a, p. 157; Jones, 2005, p. 253.) The presence of such positive expectations (or the absence of negative ones) is, then, seen in the fact that I am prepared to enter into some kind of cooperative relation with the person in question. Some degree of cooperation is involved in almost all human pursuits. Thus, if the above definition were taken literally, almost all normal human activities would involve an element of trust. Quite several writers have been driven to that extreme position. Baier claims, “[w]e inhabit a climate of trust as we inhabit an atmosphere and notice it as we notice air, only when it

Trust in Wittgenstein

35

becomes scarce or polluted” (Baier, 1986, p. 234). For instance, she claims, in a library we trust our fellow library users to be looking for books and not victims between the library stacks (Baier, 1986, p. 234). Still according to Baier, the wife that goes to bed next to her husband trusts that a brain disease will not unexpectedly turn him into a mad aggressor (Baier, 1994, p. 159). (One additional problem of the example is that if the husband did become violent through a sudden onset of brain disease, she probably would not consider that a betrayal of her trust.) Niklas Luhmann claims, “[o]ne who goes unarmed among his fellow men puts trust in them” (Luhman, 1973, p. 25). Without such background trust, Lorraine Code sums up, society would “simply fall apart” (Code, 1987, p. 377; Govier, 1992). In most of these cases, “trust” just appears to be equated with the absence of outright distrust. But if this is done categorically and regardless of the circumstances, trust is turned into a trivial element of almost all interaction, of many cases where people do nothing more than keep out of each other’s way. I have not formed any specific judgment (good or bad) about most of the individual men and women who constitute humankind. Should we say I trust them all since I go unarmed among them? But the fact remains that people make a distinction between different cases of human coordination. More remains to be said about what is expressed when such a relation is described as one of trust. Suppose I go to the library and do not think of my fellow library users at all, and suppose that this library is not known to be badly crime infested. And suppose just nothing out of the ordinary happens. In what way will it be helpful now to say I trusted the other patrons, as opposed to saying I was hardly aware of them at all? Perhaps the suggestion is that my ordinary life in a society will involve generalized trust in most people around me. But again, we would need an illuminating description of the difference between saying this, and just saying that I typically do not distrust other people unless I have some reason to. Here someone might think that the definition of trust as a cooperative disposition should be supplemented by additional criteria. For instance, Diego Gambetta suggests that my favorable expectation from another qualifies as trust if the perceived probability of cooperation is >0.50 (Gambetta, 1988a, p. 218). One difficulty with such specifications is that they risk ruling out some cases that might plausibly be described as ones of trust, while still including implausible ones. The quoted definition would rule out the (numerous) cases where we are either completely unaware of our trust or have no precise idea of the probabilities. On the other hand, it is applicable to a manipulative attitude that might more properly be described as suspicious. However, there is a more fundamental objection. It appears to us that, in looking for additional criteria for a proper definition of “trust,” one inevitably runs into problems similar to the ones indicated in Section 1.A. The definition of trust as a psychological state and the definition of trust in terms of

36

OLLI LAGERSPETZ AND LARS HERTZBERG

expectations involve the same problematic assumption: namely, the idea that the presence of the relevant state, disposition, or behavior could be established neutrally, regardless of context. At this juncture, our suggestion is not, say, “Trust is the expectation of good will plus something else.” What we need is not a more precise classification of different types of human attitudes. We instead need to see why and when, in human interaction, relationships are described as ones of trust. Thus the focus is moved from the classification of different cases to an inquiry into the way the concept itself enters our lives. For instance, it appears to us that the question to be asked is not, “Do we, or do we not, trust our fellow members of society, our fellow library users, and so on?” but instead, “What (if anything) are we saying if we answer this question one way or the other?” In other words, what can be achieved by such talk; in what kinds of contexts may it be illuminating to talk in these ways about our relations to other people? To suggest a possible case, a generalized claim about trust in society might have a point in making a contrast between different societies, say, one in which people always lock their front doors, and one in which they do not. If such talk is intelligible, it will be so because a specific contrast is invoked. Trust is a pattern in the weave of life. To call something a pattern is to see it under the aspect of meaningfulness and purpose, not as a haphazard combination of elements. We propose the question: what are we doing when we recognize such a pattern and point it out to others? 2. The Use of “Trust” A. Trust and Betrayal The difficulty of giving general criteria for the state, belief, disposition, or attitude that might be involved in trusting is importantly connected with one of Wittgenstein’s central methodological insights. His discussion of the meaning of psychological expressions is closely related to his general critique of what is known as the Augustinian picture of language, presented in the opening pages of the Investigations: the picture where language at bottom consists of names that have referents. Wittgenstein wants us to get away from the idea that the most basic use of language consists in simply reporting how things are. We are encouraged to regard language as a set of tools, used in a variety of ways to cope with different situations. (Wittgenstein, 1953, I: §§ 11–12, 14, 17, 23.) A similar point holds for both mental and behavioral phenomena. The roles of what in some respect may count as instances of the same attitude or behavior may be quite different depending on the situation. This is an instance of Wittgenstein’s insight that, if we want to get clear about the meaning of a word, it is of little help simply to focus on some entity that the word stands

Trust in Wittgenstein

37

for. In many cases there is no such common entity, and even when there is, there remains the question under what aspect the entity is being referred to when the word is used. Instead, we should look at the role of the word in human interaction. Then how is the word “trust” used? In what kinds of situation do we intelligibly ascribe trust to others or to ourselves? Probably the most natural situation where we speak of trusting is one where someone’s trust has been betrayed, or perhaps kept in the face of severe temptation. Thus the ideas of trust and betrayal belong together. This is not to say that trust, at bottom, includes a deposit of suspicion. Yet the point is that, in order for an attribution of trust to be intelligible, one must be able to invoke some imaginable disappointment. The disappointment may be due to active betrayal, but also to incompetence or thoughtlessness. (In many cases, incompetence and thoughtlessness will also involve betrayal. This is true when you have accepted a task though you should have realized you were not qualified to perform it.) Thus by saying that I trust my friend I vouch for him against a possible suspicion. This is not simply a neutral description of my thinking or behavior. Instead, I am justifying my thinking and behavior (as well as his) against a suggestion of disappointment. The issue, then, is not just that of finding a fitting description of the situation. By choosing to describe the situation in a certain way I may in fact change it. Suppose I invite a good friend for dinner. Do I also trust that he is not going to pocket the family silver when I am not looking? The reader might reply, “Of course!” But if you were to ask this question in a real life situation, I would probably not answer “Of course!” but “What do you mean?” In other words, what makes you ask this silly question? Is there something I should know? To say I trust my friend not to steal from me is to imply that he might do it. On the surface, this point is reminiscent of Baier’s influential definition of trust. She describes trust as “accepted vulnerability”: When I trust another, I depend on her good will toward me. (. . .) Where one depends on another’s good will, one is necessarily vulnerable to the limits of that good will. One leaves others an opportunity to harm one when one trusts, and also shows confidence that they will not take it. (. . .) Trust then, on this first approximation, is accepted vulnerability to another's possible but not expected ill will (or lack of good will) toward one. (Baier, 1986, p. 235, emphasis added.) Trust, then, according to Baier is, “awareness of risk along with confidence that it is a good risk” (Baier, 1986, p. 236). What is right in this description is the emphasis on the conceptual relation between trust and the risk of betrayal. On the other hand, it involves a troubling ambiguity. It is not clear from

38

OLLI LAGERSPETZ AND LARS HERTZBERG

whose perspective Baier is viewing the situation. If I trust my friend, from my own point of view I have not made myself vulnerable by inviting him. I am not leaving my friend an opportunity to harm me, any more than I would if my friend was a locksmith and I failed to have safety locks installed to keep him out. To speak of opportunity here implies a certain idea of what he wants to do. And if in the end he does steal from me, from my reaction it will be obvious that I do not regard his behavior as something I had accepted. On the other hand, the police officer who investigates the case will note that I turned my back and gave my (ex-)friend the chance to line his pockets with valuables. This brings up the fact that in quite many cases it is such a divergence between perspectives (between those of different individuals, or those of one individual at different points in time) that provides a room for talk about trust. We will return to this point shortly, but first we need to address an obvious objection. Granted that I do not think of my friend’s behavior in terms of vulnerability, might one not say that I objectively make myself vulnerable to harm? After all, I am not taking measures to prevent him. In answer to this, one needs to point out that risks, probabilities, and possibilities are not entities that exist out there on their own. They are, instead, something we invoke in the context of our practical reasoning. We hope to get things done (or undone), we fear other things, and we realize that certain obstacles and prohibitions stand in the way of our action. Possibilities are not invoked haphazardly but in the light of human aspirations and expectations. To speak of some risk as “objective” is to claim, for instance, that I ought to consider a scenario to which I have not paid sufficient attention. In one sense it will be “possible” for my friend to steal from me, just as it will be “possible” for him, in the abstract, to do some other completely unexpected things. Saying “Joe stole from his friend” will not involve a logical problem (the way that saying “Joe stole from himself” might), nor is there anything physically to stop him. But no one could challenge a person’s trustworthiness merely by bringing up considerations of what is logically conceivable or physically feasible. Any such discussion could be relevant in the first place only if there were reasons for suspicion. The fact that a scenario is imaginable in some sense, then, does not necessarily make it possible in the sense relevant to the present example. The meaningfulness of speaking of vulnerability depends, not on what might be construed as imaginable in some abstract sense, but on what is the point that might be made by invoking a scenario in a context. (For a discussion of the notion of what can be imagined, see Hertzberg, 1994a.) Thus in a sense we do not distrust people because we consider certain things possible for them. On the contrary, the fact that we consider certain things possible for them is an

Trust in Wittgenstein

39

expression of our distrust. By saying that it is possible for my friend to steal from me we would normally be voicing our suspicion that he might want to do it. To say that it is possible for him not to do it sounds even odder. This last point may sound surprising since, according to a widely held philosophical view, whatever is the case must by definition also be possible. But possibilities are not objective entities that can meaningfully be said to exist regardless of our reasons for invoking them. To describe someone’s attitude as trusting implies a risk of betrayal. But unless my friend tries to steal from me, there will be no obvious reason why I should agree that there ever was a risk. (On the other hand, a spectator may, with or without good reason, harbor suspicions against my friend that are alien to me.) For these reasons, it would typically not be meaningful for me to say, out of the blue, that I trust my friend not to pocket valuables from the house. I could only say it meaningfully as a reply to what I can recognize as an intelligible expression of suspicion. I would find it silly to say, of any of my friends individually, that I trust them not to steal from me. But on some occasions, perhaps, there still is something to be said for the general claim that, as a rule, we trust our friends not to do such things. It may, for instance, be a way to admonish someone who has tried to steal from a friend. The meaningfulness of such descriptions will depend on how credible they are in establishing the new perspective. Thus by invoking the language of trust and betrayal, we do not simply identify facts, possibilities, or risks that exist out there. Instead we take up a certain perspective. In the sequel, it will be argued that this perspective is an ethical one. B. Trust and Risk Taking In much the same way as the connected word “betrayal,” our talk of trust invokes an ethical perspective on human action. To describe a relation as one of trust is already to claim that the breach of that relation would constitute betrayal. However, we are struck by the almost programmatic avoidance of that very perspective in current literature. There the assumption is, instead, that philosophers should define “trust” in an ethically neutral way and perhaps subsequently address the question whether it might be wrong to betray someone’s trust. In that literature, trust is usually described as a form of risk taking or risk management. Thus, according to Gambetta, [w]hen we say we trust someone or that someone is trustworthy, we implicitly mean that the probability that he will perform an action that is beneficial or at least not detrimental to us is high enough for us to

40

OLLI LAGERSPETZ AND LARS HERTZBERG consider engaging in some form of cooperation with him (Gambetta, 1988a, p. 217).

Baier’s notion of accepted vulnerability is similar to this view. Trudy Govier’s account is also closely related: Trusting another, we are willing to go ahead without a guarantee. We feel that we can rely or depend on the other, even though there is always some possibility that he or she will act in unexpected ways, or even betray us. (Govier, 1997, p. 4.) Given the risks connected with “go[ing] ahead without a guarantee” the reader may ask why we should trust others at all. A conclusion might be: if we are to trust, the rationally required amount of good grounds should, in each case, be proportionate to the value of what is at risk. Govier reaches this conclusion, only adding that the calculus should “in some cases” be tempered by “ethical or prudential considerations,” which are somehow superimposed (Govier, 1993a, pp. 167–168). Trust can be slight, moderate, or complete (Govier, 1993a, p. 157). For instance, “[t]o accept a man’s help carrying packages across a busy street, a woman needs to trust him, but slight trust will be enough—unless the packages contain exceedingly valuable items” (Govier, 1993a, p. 167). Thus it is a curious fact that several writers, in trying to give an account of trust, apparently end up describing what is more correctly characterized as a kind of cheerful suspiciousness. On one natural way of reading them, important cases of real, genuine trust simply fall out of the picture. (D. Z. Phillips has noted a similar tendency among philosophers of religion trying to give an account of religious trust in Phillips, 2002) As was argued in the previous section, trust is typically characterized by the fact that we do not consider the possibility that we might be let down. In contrast, the definitions quoted above would be applicable to a sort of cynical calculus characteristic of coercive or manipulative relations. These are relations where trust is manifestly missing. This criticism applies to much of the theoretical work on trust carried out within Game Theory but also, in less formal terms, to other contributions to what may safely be described as the current mainstream (A position similar to Gambetta’s is explicitly stated by Dasgupta, p. 51; Good, p. 33; Hart, p. 186– 187; Pagden, p. 129; Williams, p. 8; see also Govier, 1993, p. 15; Johnson, 1993, p. 15). According to Gambetta, to quote an extreme example, a slaveowner’s trust in his slaves may simply amount to trust in the fact that the slaves are not going to rob him of their work force by committing mass suicide (Gambetta, 1988a, p. 219). This description strikes us as nothing short of (an inadvertent) reductio ad absurdum of the position that many authors are advancing.

Trust in Wittgenstein

41

C. Two Approaches to Human Interaction What attracts philosophers to such paradoxical views? One possible explanation is that there is a clash here between two ways of adjudicating what is at stake. There are two ways to address the question what should count as justified trust. First, beliefs may be evaluated from an instrumental point of view. On this view they guide us in our attempts to cope with reality. Holding a specific belief is a good thing for us if it helps us achieve our goals and stay unharmed. The beliefs we hold about other human beings can also be considered from this perspective. In this perspective, other people constitute one subset of the different entities that we have to deal with in our efforts to get by in the world; entities that are, to be sure, especially intricate and that may have an uncommonly powerful influence on our chances of success. When someone’s belief in the favorable intentions of another is being considered from this perspective, this belief may or may not be a good thing. Whether it is depends on the outcome in each particular case. What we expect from others is here simply seen as a matter of the projections we make on the basis of their past behavior. For this reason, any blame for possible disappointments lies squarely with us as believers. If our erroneous prediction about someone’s behavior was one that we could have avoided, it shows deficient judgment on our part. This means that, from this perspective, no room exists for a notion of trust as an ethical relation, irreducibly involving two individuals. When regarded under the aspect of trust, however, disappointments may be judged differently. This is connected with the internal relation between trust and betrayal. If A’s relation to B is rightfully to be called a case of trust (that is, if A had a right to trust B) this means that certain ways of behaving on the part of B will constitute a betrayal. If A is disappointed in B, we may judge that B had betrayed A’s trust, thus implying that B, not A, is the one to blame. The question whether what B did was a betrayal of trust can only be answered by considering the relation between A and B. For instance, we may need to ask whether B was A’s parent, teacher or friend, or whether, through promises or through past behavior, B had permitted A to count on him or her, or, for instance, whether what A was counting on was something that anyone could count on from another in the circumstances. Thus, people would normally take it to be self-evident that a stranger offering directions will not willfully mislead, or that a passer-by helping you climb out of a well will not suddenly let go. (It should be clear that for something to constitute a betrayal does not depend on there having been any explicit undertaking.) My faith that you will not betray me goes beyond expecting you to do, or refrain from doing, certain specific things; it involves resting assured that you will be mindful of my well-being. Trust, in this sense, is an open-ended

42

OLLI LAGERSPETZ AND LARS HERTZBERG

relation between two individuals. On the other hand, the present analysis will not apply to cases where a person’s invocation of trust is considered groundless. If a kidnapper claims to “trust” the abducted child’s parents not to contact the police after its release, the kidnapper will not be in a position to accuse them of betrayal if they do. She or he has no right to expect them to keep their promise. In the end, what would be called for in adjudicating the issue are neither epistemological nor strategic considerations, but a moral judgment. If B betrays A’s trust, B will not be absolved by the argument that A should have seen it coming, any more than the fact that I have been a scoundrel in the past gives me a license to go on being a scoundrel. From this perspective, the question whether a person’s trust was misplaced is primarily a judgment about the person trusted and not about the person having the trust. Here someone might object: this may be true as far as it goes, but it does not change the fact that, if A had been less trusting, things would not have turned out as badly as they did. However, for A to be less trusting would simply mean that A did not trust B. But the fact is that A did trust. For A, there was no room for plausible suspicion: that is what it means to trust someone. True, it is probably unwise to rely on people whom we know to be untrustworthy. But the fact that our trust is betrayed will not as such show that we acted unwisely, given the extent of our knowledge at the time. There is simply no account of rationality that guarantees immunity against misfortune and ill will. The question here is, instead, how we should judge B’s performance given the fact that A trusted B. Out of these two approaches, mainstream theorists invariably adopt the instrumental perspective. This may be surprising, as it looks unpromising as a way to approach moral relations. However, it is in accordance with a rationalist self-understanding where philosophy is perceived as a disinterested and in some sense scientific pursuit. This perspective gives the impression of offering a neutral vantage point from which to assess human behavior. In contrast, accounts given from the perspective of trust may give rise to the suspicion that normative views are being proposed under the guise of conceptual analysis. It looks as if they committed us to the view that trusting is a priori always a good thing. Thus, both Baier and Govier have reacted to what they perceive as a Wittgensteinian normative argument in favor of less reflection and more trust (Baier, 1997, p. 121; Govier, 1993, pp. 24–25, 32– 33). In contrast with what she sees as idealizing descriptions, Baier claims that “[o]nly if we had reason to believe that the most familiar types of trust relationships were morally sound would breaking trust be any more prima facie wrong than breaking silence” (Baier, 1986, p. 253, emphasis added). This is to misconstrue the analysis just proposed. As is clear from our example (where A is betrayed by B), we are not proposing that trust can never be misplaced. Nor are we claiming that each and every relation that someone, for whatever reason, chooses to call trust is beyond moral criticism. People

Trust in Wittgenstein

43

simply do distinguish between trust on the one hand and, for instance, naïveté or manipulation on the other; and by calling a person’s expectations from another “trust” the speaker commits him- or herself to the view that to let that person down would constitute a betrayal. Thus, normativity is not brought in by philosophical theory; it is already there when someone chooses to describe a relationship as one of trust. Baier has suggested that Olli Lagerspetz’s account of trust (Lagerspetz, 1997; see also Lagerspetz, 1998) is expressive of a cultural bias, due to his lack of exposure to life in large American cities (Baier, 1997, p. 121), and Govier has made an analogous suggestion invoking Lars Hertzberg’s North European value background (Govier, 1993a, p. 172; Govier is referring to Hertzberg, 1988). But it must be possible to judge a philosophical attempt to clarify what it means to speak about trust independently of the writer’s cultural environment. Whether to recommend that people on the whole be more trustful or more suspicious is, we suppose, a matter of world-view. This is bound to reflect one’s experiences in life. (Although it is not obvious what conclusions one must be drawing from experience. Suppose someone refuses to give up on people despite having been deceived time and again. Some of us might consider this stupid or self-deceived, while others would find the attitude an admirable sign of strength. Both responses, we would argue, may be intelligible.) But such recommendations should be kept separate from the attempt to get clear about the sense of the remarks that people may make about trust. Calling something a betrayal of trust is a negative remark, whether it is made in Åbo, Watts, or in the streets of Baghdad. 3. Certainty and Evidence A. Introduction In her essay, “Trust as an Affective Attitude,” Karen Jones describes trust as an attitude of optimism about the goodwill and competence of the other (Jones, 2005, p. 253 and passim). Her description agrees with other mainstream accounts in representing trust as an essentially risky undertaking that stands in need of justification. In the present context, however, what interests us is her emphasis on the relation between trust and the notion of evidence. She writes: Trust restricts the interpretations we will consider as possibly applying to the words and actions of another. When we can—and sometimes even if doing so requires ingenuity—we will give such words and actions a favorable interpretation as consistent with the goodwill of the other. Trusting thus functions analogously to blinkered vision: it shields from view a whole range of interpretations about the motives of another and restricts the inferences we will make about the likely actions of another.

44

OLLI LAGERSPETZ AND LARS HERTZBERG Trusting thus opens one up to harm, for it gives rise to selective interpretation, which means that one may be fooled, that the truth might lie, as it were, outside one’s gaze. (Jones, 2005, p. 262 and passim.)

As the one trusted “is viewed through the affective lens of trust,” the person who trusts him (the victim of this distortion) will be prepared to risk dependence on him “often on the basis of the smallest evidence” (Jones, 2005, p. 263 and passim). Given this characterization of trust as a distortion of available data, the author maintains that we need to find some justification for the fact that we trust others at all. It is not obvious that such a need exists. Jones’s description of trust as a case of blinkered vision will primarily be applicable to cases of ill-advised trusting (bad judgment, naïveté, self-deception, recklessness, or the like). But these are not cases that anyone should wish to justify. On the other hand, the description is not going to cut ice with persons who do not already think there is a problem about their trust. They will simply not think of themselves as overly optimistic about their trusted friends. If their friends are trustworthy, trust will be a matter of realism, not of optimism. Distrust, on the other hand, will come out as pessimism; or perhaps, as optimism regarding one’s superior skills in detecting falsity. Suppose the cited passage is travestied as follows: Distrust restricts the interpretations we will consider as possibly applying to the words and actions of another. When we can—and sometimes even if doing so requires ingenuity—we will give such words and actions an unfavorable interpretation as consistent with the ill will of the other. Distrust thus functions analogously to blinkered vision: it shields from view a whole range of interpretations about the motives of another and restricts the inferences we will make about the likely actions of another. Distrust thus opens one up to harm, for it gives rise to selective interpretation, which means that one may be fooled, that the truth might lie, as it were, outside one’s gaze. This description strikes us as just as applicable as the previous one. Generally speaking, it is hardly reasonable to suggest that anyone should try to settle, in the abstract and regardless of the situation at hand, whether trust or distrust is the more rational attitude. However, Jones’s description does highlight the intimate relation between the notions of reasoning and evidence on the one hand, and the notions of trust and distrust on the other. Only her account turns the relation upside down. Jones offers us an account where the starting point is a recognizable standard of good reasoning based on sound evidence. Trust enters the picture as a disturbing element. (Jones concedes that distrust too sometimes disturbs reasoning in an analogous way in Jones, 2005, p. 267 and passim.) Thus in this picture, the fully rational person attends to the evidence

Trust in Wittgenstein

45

(which is already there) and, uninfluenced by others, takes it for what it is worth, neither more nor less. In contrast, we would maintain that the ability to trust is constitutive of there being any such thing as valid reasoning and sound evidence in the first place. B. “The Substratum of All my Enquiring and Asserting” This theme is something of a Leitmotif in Wittgenstein’s On Certainty. That work starts off with a discussion of G. E. Moore’s attempted refutations of skepticism (Moore, 1993, pp. 147–170). However, Wittgenstein does not really focus on skepticism but, instead, on the remarkable fact that we would find it impossible to doubt certain statements about the world and our lives in it. I cannot seriously doubt that the world has existed for a long time before my birth or that all human beings have intestines. I cannot doubt that I have two hands or that I have never been to China. These statements do not qualify as logical truths in a conventional sense, yet they are not straightforwardly empirical either. As Moore, too, points out, in cases of this kind I could not point to specific evidence that has settled the question for me (Moore, 1925, p. 118). I could not cite a letter by Napoleon to prove that the world is more than two hundred years old. I can take the letter to be authentic only if I already accept that the world existed at the time. “The belief that the world is more than two hundred years old” is then not a historical belief in any normal sense. If there are to be historical inquiries at all it is, instead, taken for granted that we do not doubt such things. (Wittgenstein, 1972, § 188; see also §§ 163, 182–192.) As Wittgenstein notes, “I have a world-picture. Is it true or false? Above all it is the substratum of all my enquiring and asserting.” (Wittgenstein, 1972, § 162.) He adds, “I say world-picture and not hypothesis, because it is the matter-of-course foundation for [the scientist’s] research and as such goes unmentioned” (Wittgenstein, 1972, § 167). An analogous point is possibly developed in R. G. Collingwood’s Essay on Metaphysics, where he puts forward a conception of metaphysics as the study of the “absolute presuppositions” of thought (Collingwood, 1979). According to Wittgenstein, my convictions form “a system, a structure” (Wittgenstein, 1972, § 103). The system “is not so much the point of departure, as the element in which arguments have their life” (Wittgenstein, 1972, § 105). I do not explicitly learn such convictions, but I can later recognize that they are implicit in the things I do learn, “like the axis around which a body rotates.” Like the axis of a globe floating in space, it does not support the body but instead “the movement around it determines its immobility.” (Wittgenstein, 1972, § 152.) The upshot is that it is misleading to think that we might find evidence for something completely obvious. By “evidence,” we mean facts that rightly

46

OLLI LAGERSPETZ AND LARS HERTZBERG

incline us to settle an open question one way or another. This means, for instance, that a historian today could not seriously treat newspapers from 1944 as evidence for the fact that the Second World War took place. It would be easy enough to locate written material that implies there was a great war at the time. But to call it evidence would imply that serious disagreement exists about the matter. For anyone with an ordinary Western education such historical facts will count as at least as obvious as is the idea that old newspapers could be employed as historical sources in the first place. This is not to say that no one could ever doubt that the Second World War happened. What we cannot doubt today perhaps some future generation will (Wittgenstein, 1972, § 152; see also §§ 96–99). (For related reasons, Collingwood describes the study of the absolute presuppositions of thinking as an essentially historical investigation, since those presuppositions will be different during different historical periods.) Nor is it to say that there is nothing the historian could say to the professed skeptic. But insofar as such a thing as historical inquiry will exist, some facts of this general kind will be commonly treated as unquestionable. Our ability to distinguish between serious and spurious disagreement is something we acquire by being brought up in a culture. We generally take as true what we find in textbooks. If in doubt, we consult other written sources or ask an expert. Thus “I learned an enormous amount and accepted it on human authority, and then I found some things confirmed or disconfirmed by my own experience” (Wittgenstein, 1972, § 161, emphases added). The child begins its learning process by believing the adults around it (Wittgenstein, 1972, §§ 160, 310). Without such obvious trust, the child could never develop the critical faculties that make genuine doubt possible. Reasoned doubt will always presuppose trust in other directions. But trust in the testimony and sound judgment of others is not only characteristic of children. In adults, it is not a residual “childlike thoughtlessness, innocence, and powerlessness” (Cf. Baier, 1997, p. 121). On the contrary, as we grow older we learn to make more use of the information and expertise available around us. Our ability to distinguish between serious research and nonsense is based on our ability to trust some sources. Our intellectual life as a whole is characterized by what one may call an epistemological division of labor. In sum, our faith in the judgment and testimonies of others “is not, so to speak, hasty but excusable: it is part of judging” (Wittgenstein, 1972, § 150, emphasis added). While trust may involve intellectual distortion in some cases, the general fact that we tend to trust others is, on the contrary, constitutive of sound judgment. It is not a weakness to be excused or justified. On the whole, On Certainty, like the Investigations, shows Wittgenstein as a “social” thinker. Both works describe meaningful thought as an activity that arises and makes sense in the context of human interaction and interdependence.

Trust in Wittgenstein

47

C. The Pathology of Distrust We see that there is an important asymmetry between the place of trust and distrust in human growth. No critical faculties will develop unless the child starts by not doubting. (Analogously, there is an asymmetry between pretense and genuine communication. Wittgenstein points out that a child must learn a great deal before it can pretend, presumably including the use of the corresponding genuine expression. Wittgenstein, 1953, pp. 228–229.) Does this imply that normal children trust their parents from the very start? Or only that they do not distrust them? Both answers might be right depending on the context of the question. As argued earlier, the point of using the language of trust is not primarily to arrive at an accurate classification of behavior or mental states but, instead, to invoke an ethical perspective. To say that a child trusts its parents is, for instance, to criticize parents who ignore its needs or talk ironically to it. Our present point is that normal human lives will, from the very start, involve relations of mutual dependence that are not questioned. In a small child, the opposite of trust is typically not called doubt, distrust or suspicion but, perhaps, fear. An infant who did not start life with a basically trustful attitude toward those around (that is, an infant whose attitude is an overwhelmingly fearful one) would be considered abnormal. It is doubtful whether one could coherently describe such an infant as suspicious at all. Something analogous is true of adults. If each time we go to a restaurant our friend brought along a chemistry set to make sure the food is safe to eat, or always looked under the table to see if it were bugged, then, unless given specific reasons, we would not just describe our friend as unusually suspicious. We would soon start worrying about our friend’s sanity. (see Gaita, 1991, p. 314.) In contrast, we will not consider people pathologically careless or gullible just because they never check for poison or bugs except in extraordinary circumstances. We normally go about our daily business without giving much thought to the strangers around us. Yet as soon as the need for interaction arises, most of the time we will incline in the direction of trust unless we have grounds for acting otherwise. If we would not ask a stranger in the streets of New York or Moscow to keep an eye on our camera, it is because we are conscious of the ubiquity of graspingness in big city cultures. Nevertheless we might ask that same stranger to help us call for an ambulance. On Certainty and Philosophical Investigations do not contain sustained discussion of the ethical character of our mutual dependence. This is perhaps part of Wittgenstein’s general reluctance to address ethical questions in his properly philosophical work. However, some such discussion is included in the notebook remarks that appear in the posthumous collection Culture and Value. Wittgenstein

48

OLLI LAGERSPETZ AND LARS HERTZBERG

comments on the problems of distrust. He points to the kinship between suspiciousness and insanity: Madness doesn’t have to be regarded as an illness. Why not as a sudden—more or less sudden—change of character? Everybody is (or most are) mistrustful, & perhaps more so toward their relations, than toward others. Is there any reason for mistrust? Yes & no. Reasons can be given for it, but they are not compelling. Why shouldn’t someone suddenly become much more mistrustful of people? Why not much more withdrawn? or devoid of love? Don’t people get like this even in the ordinary course of events? (Wittgenstein, 1998, p. 62e.) In an earlier notebook, Wittgenstein writes about our inclination to shut ourselves off from others. He describes it as a moral shortcoming: Someone who (. . .) opens his heart to God in remorseful confession opens it for others too. He thereby loses his dignity as someone special [alternative translation: outstanding] & so becomes like a child. That means without office, dignity & aloofness from others. You can open yourself to others only out of a particular kind of love. Which acknowledges as it were that we are all wicked children? It might be said: hate between human beings comes from our cutting ourselves off from each other. Because we don’t want anyone else to see inside us, since it’s not a pretty sight in there. You must continue to feel ashamed of what’s within you, but not ashamed of yourself before your fellow human beings. (Wittgenstein, 1998, pp. 52e–53e.) Shutting oneself off from others, and, accordingly, hate between human beings, is closely connected with lack of trust. On Wittgenstein’s view, then, the significance of trust did not simply lie in its role in human thought and judgment, but many forms of human conflict had their source in, or were aggravated by, a lack of trust. We have earlier referred to the grammatical connection between trust and betrayal. This connection highlights the responsibility of anyone rightfully considered an object of trust. In his notebooks, Wittgenstein draws attention to the other side of the relation. He suggests that there is an obligation to trust, or, instead, not to be distrustful. However, he is not offering this as the result of a grammatical investigation. It is expressive of Wittgenstein’s understanding of his life, and of what is important in human affairs.

Trust in Wittgenstein

49

4. Summary We have discussed three themes in Wittgenstein’s work particularly relevant for understanding the concept of trust. (1) Wittgenstein’s discussion of language and of psychological concepts in particular suggests that psychological expressions (such as “trust”) are not primarily employed to refer to independently identifiable psychological states. This concept should instead be seen as part of an interaction where moral relations come into play. In particular, it is connected with the attribution of blame. Thus we need not expect the presence of a specific state of mind or a specific behavior pattern every time the word “trust” is applicable. Conversely, the fact that the word is truthfully used to describe a given case of human interaction does not imply that all similar cases may be so described regardless of the moral relations involved. (The main relations to consider are that between the truster and the trustee as well as that of the observer to them both.) This implies, in particular, a criticism of the idea (by Luhmann, Baier, and others) that a generalized form of trust is necessary in order for social life to be possible. (2) If philosophers assume that our grasp on reality is ultimately to be spelled out in terms of factual assertions, they will come to see trust as a matter of holding certain beliefs about the person trusted; beliefs that, in turn, will be taken to justify our confidence in what the person tells us. This widely held view has largely come to shape current accounts of trust. No room exists in them for the idea of genuine trust as a moral relation irreducibly involving two individuals. According to Wittgenstein, the starting point for this line of thought needs to be drawn into question. The sense of an assertion, he claimed, is dependent on the way it enters into a context of life. Hence the idea that our relation to reality is ultimately constituted by factual assertions is confused. This realization opens up for a non-reductive understanding of trust. In many cases the trust we have for another individual will be basic to the beliefs we come to form, not the other way round. (3) In On Certainty, Wittgenstein is exploring the fact that we take some ways of thinking as self-evident without asking for further evidence. Skepticism is only possible against the background of massive general agreement. A certain dependence on knowledge claims and judgments by others is not only a practical necessity but part of what it means to make reasoned judgments. Our general situation may be described as logical and epistemic division of labor. It is not illuminating to describe all instances of such division of labor as trust regardless of the contrasts one wants to make; nevertheless, ascriptions of trust as well as distrust make sense against this background of mutual dependence.

50

OLLI LAGERSPETZ AND LARS HERTZBERG

WORKS CITED Alanen, Lilli, Sara Heinämaa and Thomas Wallgren, eds. (1997). Commonality and Particularity in Ethics. Basingstoke, England: Macmillan. Armstrong, D. M. and Norman Malcolm, eds. (1984). Consciousness and Causality. Oxford, England: Basil Blackwell. Baier, Annette. (1986). “Trust and Antitrust,” Ethics, 96:2 (January), pp. 231–260. ———. (1994). Moral Prejudices. Cambridge, Mass.: Harvard University Press. ———. (1997). “Reply to Olli Lagerspetz.” In Commonality and Particularity in Ethics, eds. Lilli Alanen, Sara Heinämaa and Thomas Wallgren. Basingstoke, England: Macmillan, pp. 118–122. Code, Lorraine. (1987). “Second Persons.” In Science, Morality, and Feminist Theory, eds. Marsha Hanen & Kai Nielsen. Calgary, Alberta: University of Calgary Press. Collingwood, R. G. (1979). An Essay on Metaphysics. Oxford, England: Oxford University Press. Dasgupta, Partha. (1988). “Trust as a Commodity.” In Trust: Making and Breaking Cooperative Relations, ed. Diego Gambetta. Oxford, England: Blackwell, pp. 49–72. Gaita, Raimond. (1991). Good and Evil: An Absolute Conception. Basingstoke, England: Macmillan. Gambetta, Diego. (1988a). “Can We Trust Trust?” In Trust: Making and Breaking Cooperative Relations, ed. Diego Gambetta. Oxford, England: Blackwell, pp. 213–245. ———, ed. (1988b). Trust: Making and Breaking Cooperative Relations. Oxford, England: Blackwell. Giddens, Anthony. (1991). Modernity and Self-Identity. Cambridge, England: Polity Press. Good, David (1988). Individuals, Interpersonal Relations, and Trust.” In Trust: Making and Breaking Cooperative Relations, ed. Diego Gambetta. Oxford, England: Blackwell, pp. 31–48. Govier, Trudy. (1992). “Trust, Distrust, and Feminist Theory,” Hypatia, 7:1 (Winter), pp. 16–33. ———. (1993a). “An Epistemology of Trust,” International Journal of Moral and Social Studies, 8:2 (Summer), pp. 155–174. ———. (1993b). “Trust and Testimony: Nine Arguments on Testimonial Knowledge,” International Journal of Moral and Social Studies, 8:1 (Spring), pp. 21–39. ———. (1997). Social Trust and Human Communities. Montreal & Kingston: McGillQueens University Press. Hanen, Marsha and Kai Nielsen, eds. (1987). Science, Morality, and Feminist Theory. Calgary, Alberta: University of Calgary Press. Hart, Keith. (1988). “Kinship, Contract, and Trust.” In Trust: Making and Breaking Cooperative Relations, ed. Diego Gambetta. Oxford, England: Blackwell, pp. 176–193. Hertzberg, Lars. (1988): On the Attitude of Trust,” Inquiry, 31:3, pp. 307–322. ———. (1994a): “Imagination and the Sense of Identity.” In The Limits of Experience. Helsinki: Acta Philosophica Fennica, pp. 96–112.

Trust in Wittgenstein

51

———. (1994b): The Limits of Experience. Helsinki: Acta Philosophica Fennica. Johnson, Peter. (1993). Frames of Deceit: A Study of the Loss and Recovery of Public and Private Trust. Cambridge, England: Cambridge University Press. Jones, Karen. (2005). “Trust as an Affective Attitude.” In Virtues: Introductory Essays, ed. Clifford Williams. Basingstoke, England: Palgrave Macmillan, pp. 253–257. Lagerspetz, Olli. (1997). “The Notion of Trust in Philosophical Psychology.” In Commonality and Particularity in Ethics, eds. Lilli Alanen, Sara Heinämaa and Thomas Wallgren. Basingstoke, England: Macmillan, pp. 95–117 ———. (1998). Trust: The Tacit Demand. Dordrecht, South Holland: Kluwer. Luhman, Nicklas. (1973). Trust and Power. Chichester, England: Wiley. Malcolm, Norman. (1984). “Consciousness and Causality.” In Consciousness and Causality, eds. D. M. Armstrong and Norman Malcolm. Oxford, England: Basil Blackwell, pp. 1–101. Moore, G. E. (1925). “A Defence of Common Sense.” In Selected Writings. ed. Thomas Baldwin. London: Routledge, pp. 106–133. ———. (1939). “Proof of an External World.” In Selected Writings. ed. Thomas Baldwin. London: Routledge, pp. 147–170. ———. (1993). Selected Writings. ed. Thomas Baldwin. London: Routledge. Pagden, Anthony. (1988). “Trust in Eighteenth-century Naples.” In Trust: Making and Breaking Cooperative Relations, ed. Diego Gambetta. Oxford, England: Blackwell, pp. 126–141. Phillips, D. Z. (2002). “On Trusting Intellectuals on Trust,” Philosophical Investigations, 25:1 (January), pp. 33–53. Sellerberg, Ann-Mari. (1982). “On Modern Confidence,” Acta Sociologica, 25:1, pp. 39–48. Williams, Bernard (1988). “Formal Structures and Social Reality.” In Trust: Making and Breaking Cooperative Relations, ed. Diego Gambetta. Oxford, England: Blackwell, pp. 3–13. Williams, Clifford, ed. (2005). Virtues: Introductory Essays. Basingstoke, England: Palgrave Macmillan Wittgenstein, Ludwig. (1953). Philosophical Investigations. Oxford: Blackwell. ———. (1967). Zettel. Berkeley: University of California Press. ———. (1972). On Certainty. New York: Harper & Row. ———. (1998). Culture and Value. Oxford, England: Basil Blackwell.

Four A TRUST-BASED ARGUMENT AGAINST PATERNALISM Simon Clarke Introduction This essay addresses the role of trust in political philosophy. In particular, it examines the idea that trust is necessary for a particular type of government action — paternalistic action — to be justified. Liberal theory and liberal democratic practice are characterized by a large degree of anti-paternalism, understanding paternalism to be the restriction of individual liberty for a person’s good, instead of to protect or benefit others. It would be a mistake to think that liberal democracies have no paternalism; seatbelt, motorcycle helmet, and drug prohibition laws, for example, are probably at least partly motivated by paternalistic reasons. But it is easy to imagine more pervasive paternalism. Society could, and does in some cultures, restrict people’s choices of occupation, marriage partners, and where to live, with the rationale that these restrictions are for people’s good. Many people believe that the liberal position is the correct one that more pervasive paternalism would be unjustified, but what is the philosophical justification for anti-paternalism? One way of arguing against paternalism is to deny the basis for it. Perhaps no reason exists to think that people’s choices are mistaken. John Stuart Mill, for example, argued that people are the best judges of their interests. Others are likely to get it wrong and hence self-regarding decisions should be left to individuals to make. (see chapter 4 in Mill, 1859.) A different way of arguing against paternalism is to say that even if people are mistaken about some matters and intervention could do some good, there are reasons not to interfere. What might these reasons be? Perhaps the most common answer is to appeal to individual autonomy. In The Morality of Freedom (Raz, 1986) Joseph Raz argues that the moral basis of freedom is an ideal of personal autonomy. According to Raz, personal autonomy is the idea that people should make their own lives. The autonomous person is a (part) author of his life. The ideal of personal autonomy is the vision of people controlling, to some degree, their own destiny, fashioning it through successive decisions throughout their lives. (Raz, 1986, p. 369.)

54

SIMON CLARKE

Autonomy requires, amongst other things, freedom from coercion and manipulation and an adequate range of options (for Raz’s enlightening discussion of these conditions see chapter 14 in Raz, 1986). If people should be autonomous, then much paternalism would be unjustified since, even though it might advance people’s welfare, it violates autonomy. It is hardly new to think that the ideal of personal autonomy is the foundation of liberty, but Raz’s discussion of autonomy is pioneering for its attempt to explain just what autonomy is, why it is valuable, and what its implications are for political arrangements. However, the argument for liberty from the ideal of autonomy has faced criticisms, such as that autonomy is of value only in particular types of societies. (This and other objections to Raz’s ideas on autonomy can be found in McCabe, 2001; Waldron, 1989; Norman, 1989; Caney, 1994.) There are replies to these criticisms of the autonomy argument (see especially Raz, 1989). But even if these criticisms succeed, Raz has another, autonomy-independent, argument against paternalism. While he holds that the value of living an autonomous life is a justification for liberty, it is not the only one, for the value of autonomy is “no more than part of the reason for our concern for liberty” (Raz, 1996, p. 113–129. Henceforth, this article will be referred as “LT”). Even if autonomy were not valuable, liberty “would still have been justified as it is by the non-autonomy-related considerations which contribute to its justification.” (LT, p. 116) This essay examines one of these considerations, one that centers on the notion of trust. I will call it “the trust argument.” The trust argument claims that paternalism is justified only if a relationship of trust exists between the paternalistic actor and the target of paternalism, and that such a relationship will seldom be the case. More precisely, the argument takes the form of a catch-22: paternalism is justified only if the person being targeted reasonably trusts the paternalist but the very fact that someone acts paternalistically toward you is a reason not to trust him or her, and so paternalism is not justified. Section one below sets out the argument more fully. Section two will focus on the first claim of the argument, namely that trust is a requirement for justified paternalism. Raz provides an argument for this “trust condition” but I will try to show the weakness of his argument. I will argue that an alternative argument provides some support for the trust condition. The other claim in the argument is the catch-22 that behavior being paternalistic provides good reason not to trust the paternalist. Section three tries to provide some reasons to reject Raz’s argument for this claim; the conclusion is that the claim should be accepted but with considerable qualification. So in short, the trust argument against paternalism succeeds but only in a qualified sense and for different reasons than Raz provides.

A Trust-Based Argument Against Paternalism

55

1. The Trust Argument Paternalism is the restriction of individual liberty for a person’s own good and coercive paternalism occurs when threats of harm are used to bring about such restriction. Laws that prevent people from engaging in certain occupations, which outlaw religions and that prohibit activities, insofar as they are intended for the good of those whose freedom is restricted, are examples of coercive paternalism. Raz claims there are two conditions needed to justify coercive paternalism: first, it is undertaken for a good reason, one sufficient to make reasonable a partial loss of independence. Secondly, barring emergencies, it comes from the hands of someone reasonably trusted by the coerced. (LT, p. 122, footnote omitted.) I will refer to the second as “the trust condition.” Raz intends it as a necessary condition for justified coercive paternalism. He writes that “[o]nly those trusted by the coerced can have authority to use paternalistic coercion” (LT, p. 123) and that “[t]rust in the government is a condition of its right to apply coercion to people for their own good, a condition for the legitimacy of paternalistic coercion.” (LT, pp. 126–7) The next step of the argument is a catch-22. The very fact that government is acting paternalistically toward you means that you should not trust it: If [government] pursues coercive moral paternalism against me it will, by definition, be preventing me from following my way of life (. . . .) It undercuts my trust that I and my interests are seriously being taken into account in deciding public action. (. . .) [G]overnments cannot resort to moral paternalism for by doing so they undercut their right to do so because they lose the trust of those against whom the coercion is used. (LT, pp. 127–8, emphasis omitted) Raz says that coercive “moral” paternalism occurs when a person is prevented from following his or her way of life, for example laws and policies intended to discourage homosexuality. What then would count as coercive “non-moral” paternalism? The idea here is to draw a distinction between coercive paternalism that imposes values on a person who does not accept them (moral paternalism), and coercive paternalism that forces people to do things that they themselves would recognize as good (non-moral paternalism). An example of the second could be laws forcing people to wear seatbelts in cars. Most (though not all) people recognize the good of not being thrown through car windscreens, so forcing them to wear seatbelts is coercive non-moral paternalism. Raz’s argument is not aimed at this type of paternalism. (Henceforth, I will refer to coercive moral paternalism as “CMP.”)

56

SIMON CLARKE

Raz does not explicitly define “trust.” I take it as implying two main beliefs: a belief in the good intentions and a belief in the ability of the trustee. Belief in the ability but not good intentions is reliance not trust (I take this point from Baier, 1995a, pp. 98–99): If I believe that a bus driver is competent at getting me to my destination on time but does not have my welfare at heart, then I may be said to rely upon instead of to trust the driver. On the other hand, if I believe a person has good intentions toward me but lacks the ability to carry them through, then again I do not trust that person. That Raz has something like this definition of “trust” in mind is shown in several of his remarks. As we will see, a central part of his argument is that targets of moral paternalism will always lack a belief in the good intentions of the paternalist and therefore lack trust. The trust condition is intended to be of independent moral force as an argument against paternalism. That is, it is intended to render unjustified, paternalism that might otherwise be justified. Recall the first of Raz’s two conditions for justified coercive paternalism: that it is done for good reason. The trust condition is intended to add to instead of merely reflect this. As Raz notes: Even where the cogency of the paternalistic reasons for coercion is not in doubt one may well object if a stranger, let alone a potentially hostile stranger, takes it upon himself to coerce one for one’s own good. The point reflects the nature of trust (. . .) (LT, p. 122) However, the trust condition requires that the paternalist must be reasonably trusted by the coerced according to Raz. The adjective makes the condition semi-objective instead of purely subjective. This says Raz means that it “turns not simply on how people actually feel but how it is reasonable for them to feel.” (LT, p. 124) So in some cases trust exists but it is not reasonable: Some people are deluded into believing that the government takes their interests fully into consideration [and consequently trust the government], when in fact the government and the law are systematically biased against them. (LT, p. 124) Trust in such a case does no justificatory work since it is not reasonable trust. Conversely, there could be cases where no trust exists but this lack is unreasonable. (LT, p. 124) This raises the question whether intervention in such cases would be justified, since the lack of trust is unreasonable. Raz suggests that it would not be, and I will assume the same in the remainder of the essay. Trust must be present for intervention to be justified, but it must reasonable trust. The reasonableness requirement risks robbing the trust argument of independent force. Consider a case of paternalism the targets of which trust

A Trust-Based Argument Against Paternalism

57

the paternalist. If the trust condition required mere trust, whether reasonable or unreasonable, the presence of trust in this case would make the intervention justified (so long as other required conditions are met). However, with the requirement being “reasonable” trust, we have to ask whether the trust is reasonable, and if it is not, intervention is unjustified. But there is a temptation to interpret the reasonableness of the trust or lack thereof as turning entirely on whether there are good reasons for the intervention. Raz is careful to ward off this implication. Requiring trust to be reasonable only makes the trust condition semi-objective. As Raz says, the condition has a subjective component. It turns not on how governments actually treat people but on what they feel about their treatment, so long as their feeling is not unreasonable. This subjective element gives [it] an independent status as a constraint on governmental actions. Had the condition (. . .) been simply that one is treated by the government as one ought to be treated it would have had no independent role in guiding governmental action. (. . .) However, given that the condition is partly subjective it has an independent normative force: governments should not only act justly, they should also be seen to act so, at least by all those whose beliefs on the matter are not unreasonable. (LT, pp. 124–5) In short, the requirement of reasonable trust attempts to veer between two extremes. If it was purely subjective it would simply be a matter of consulting people and asking them whether they trust a paternalistic government, but this would allow cases where people are deluded into trusting. If it was purely objective it would be a matter of whether the paternalism is supported by good reasons, and the trust requirement would be doing no work. It is important to keep this in mind, especially in section 4 when we examine whether it is reasonable to trust a paternalistic government. Any affirmative argument for this must not simply refer to the good reasons that might support the paternalism and thus render the trust requirement redundant. (Christopher Wolfe makes this mistake in Wolfe, 1996.) The trust argument can be formally summarized in the following way: P1: Coercive paternalism is justified only if the paternalist is reasonably trusted by the coerced. P2: A person subjected to CMP cannot reasonably trust the government. Therefore, C: CMP is not justified. The conclusion follows validly from the two premises so I turn to examination of those premises, starting with P1.

58

SIMON CLARKE 2. Justifying the Trust Condition

Why is trust required for legitimate intervention? Why is it not sufficient that action just be good intervention, regardless of whether it is trusted? Imagine that I am correct in thinking that you are mistaken, that the mistake will significantly undermine your well-being, and that I am in a position to prevent you from harming yourself. On the trust argument, even if there are good grounds for paternalism, a lack of (reasonable) trust means intervention would be unjustified. Trust might make it more likely that intervention would succeed, instead of backfiring by creating resistance and resentment to the intervention. These are reasons to think that trust is useful, but hardly reasons to make trust always necessary before intervention is legitimate. In some cases intervention without trust could succeed. Why should I need your trust before I intervene? A. Coercion and Whole-heartedness Raz’s defense of the premise is that it is an implication of what is wrong with coercion generally. Coercion is the use of threats to force a person to act or refrain from acting some way. (This is a rough definition of coercion. Raz does not define it in LT but this definition is in keeping with what he says about coercion there, and with the definition he does give in Raz, 1986, pp. 148–149.) Legal constraints are usually coercive since they usually threaten some form of punishment if not complied with. Coercion is presumptively wrong for several reasons. One is that it may interfere with autonomy (though Raz says that coercion may also sometimes promote autonomy by providing opportunities and access to them LT, p. 120). But as I suggested in section one, we are putting autonomy to one side — Raz intends the trust argument to be an independent argument against paternalism. Another reason that coercion is wrong, Raz argues, is due its effect on personal well-being. Well-being, according to Raz, consists in the whole-hearted and successful engagement in valuable activities and relationships. (This conception of well-being is given at LT p. 113. Elsewhere Raz gives other formulations which differ in subtle ways but which point to the same general conception: LT p. 121; Raz, 1986, p. 308; 1995, p. 3.) The value of a person’s life is undermined by failure on any one of those counts: if a person does not whole-heartedly engage in it; is unsuccessful in achieving his or her goals (for example, of two people who both aim to be great novelists, the one who produces novels of great insight lives a better life than the one who produces shallow mediocrity); or a person’s goals are worthless instead of valuable (for example, those who devote their lives to counting blades of grass live a worse life, all things equal, than doctors or librarians). Raz claims that coercion is incompatible with the whole-heartedness aspect of well-being (LT, p. 121). A life that is coerced is one that has its value undermined (even disregarding the value of autonomy)

A Trust-Based Argument Against Paternalism

59

because people who are coerced cannot wholeheartedly pursue their goals. This argument about the nature of coercion is the first step in supporting P1 of the trust argument. According to the argument, people are worse off when coerced. This does not mean, says Raz, that its use is always unjustified. Laws that prevent people from killing or assaulting others, from violating their rights, involve coercion, but they are justified uses of coercion even though the coerced are harmed, justified by the need to protect others and their rights. However, it will be difficult for coercive paternalism to be justified, because the coercive aspect of the intervention undermines that person’s well-being when the intervention is supposed to be for his or her good. Forcing people into art galleries for example will undermine their whole-hearted appreciation of art. But in some cases, says Raz, coercion may not undermine well-being, such as when used by friends or others whose good intentions are not in doubt (LT, p. 122). When this condition holds, the whole-heartedness aspect of wellbeing is not undermined. If a person does not doubt the good intentions of her coercer, then she may remain whole-heartedly committed to her goals even though coerced into them. (Raz does not explicitly state this step of the argument, but it appears to me the most plausible way to do so.) This, says Raz, reflects the nature of trust. When trust occurs, the normal prohibition on coercion may be relaxed (LT, p. 122). Being forced by government into art galleries undermines my whole-hearted appreciation of art, but being so forced by a close friend who I trust does not. The argument for P1, in short, is that coercive paternalism is unjustified because it harms instead of benefits the target, but that this harm may be avoided only in those cases where trust exists. There are several points at which this argument can be questioned. First is the claim that coercion undermines the whole-hearted pursuit of goals, and thereby well-being. Is it not possible to whole-heartedly pursue something I have been coerced into, even if I do not trust the good intentions of the coercer? Someone might be kidnapped and press-ganged into a ship’s crew by unscrupulous merchant-seamen, but still come to find value in a life on the oceans. The denial of such possibilities is an empirical claim, but Raz appears to hold that it is a necessary truth. However, even if we accept it as a necessary truth or a strong contingent one (in the sense that it almost always holds) this still does not mean that the trust condition should be accepted. Either of those would show that paternalistic coercion without trust undermines well-being. Even if so, it is possible that the reduction in wellbeing brought about by coercion without trust could be outweighed by the benefit to well-being that the paternalist could bring about. Coercive paternalism without trust may undermine the whole-heartedness aspect of well-being, but perhaps that loss can be outweighed by a gain in well-being if the person chooses something more valuable instead and engages in it with some lesser degree of whole-heartedness. (T. M. Wilkinson makes this point in Wilkinson, 1999, p. 136.) Maybe the choices are only these two:

60

SIMON CLARKE

government forcing people into art galleries or those people never discovering galleries for themselves and hence never experiencing great art. In this case, it is possible that the loss in well-being due to less whole-heartedness could be outweighed by the gain in well-being due to engaging in more valuable activities. For Raz’s premise to be true, it must be the case that the reduction in well-being brought about by coercion without trust always outweighs any other possible gain in well-being that might be brought about by intervention. And that is implausible. .

B. A Positive Attitude Argument To rescue the argument we need some other reasons for thinking that trust is a necessary condition for justified paternalism. Some of Raz’s comments suggest such an alternative argument for P1. Raz criticizes the common thought that government is legitimate only if it has the consent of those governed. He argues that “the ideal of universal consent, however attenuated, is unrealizable.” (LT, p. 121) But he does suggest that consent has normative significance and that this significance is related to the notion of trust. After setting out his trust condition he claims that it “explains the way consent can be relevant to the justification of coercion. It can express or establish a relationship of trust.” (LT, p. 122) Ulysses’ temporary loss of autonomy when he is tied to the mast of his ship so as not to succumb to the lure of the Sirens, says Raz, was justified. He consented to it, but “had he been tied to the mast by his best friend” (LT, p. 121) without consent, it would also have been a justified act due to the relationship of trust. So consent can justify paternalism because it expresses or establishes trust. This is not an argument for the trust condition because it presupposes that condition and then claims that consent is a way of showing trust. But perhaps a different way of conceiving the relation between consent and trust is possible. The view that consent is necessary for legitimate paternalism is an attractive one that has its defenders (see for example VanDeVeer, 1986). But it faces the objection that there are some cases, such as intervention with a person about to unknowingly cross a dangerous bridge, where paternalism without consent appears justified. To avoid this difficulty defenders of the consent view sometimes appeal to hypothetical consent; the would-be bridge crosser would, it is argued, consent to intervention if he were not ignorant about the dangerous bridge. But hypothetical consent arguments also face objections. Perhaps instead, some other notion such as trust can capture the desirable features of the consent account, while avoiding its problems. Consent and trust are both expressions of positive attitudes. Other such attitudes could be agreement, gratitude, approval, and endorsement. The argument I am suggesting is this: if a positive attitude from the subject of paternalism is a necessary condition for justified paternalism then it is trust, not consent, that is the appropriate positive attitude. This bare sketch of an

A Trust-Based Argument Against Paternalism

61

argument needs to be filled out in several areas. Even if a positive attitude of some sort is necessary perhaps it is consent, not trust, since although I mentioned above that consent accounts face difficulties, perhaps there are responses to these. And even if a positive attitude is necessary and it is not consent, perhaps it is not trust either but something else such as gratitude or endorsement. (Dworkin, 2000, pp. 267–274.) Even if answers to these questions are forthcoming, something needs to be said about why any positive attitude is needed. As I asked at the beginning of section 2, if I can prevent you from making a mistaken decision that will significantly undermine your well-being, why should I need a positive attitude from you before I go ahead? Perhaps requiring such an attitude is a deontological requirement for respecting persons. But what is it about persons that respecting them require a positive attitude on their part before it is justified to treat them in particular ways? When we focus on consent, the usual answer given is that consent is required for justified paternalism in order to respect a person’s capacity for autonomy. (chapter 2 in VanDeVeer, 1986.) But as was pointed out in section 1, we are seeking a rationale against paternalism that is independent of appeals to autonomy, so we cannot ground the positive attitude in autonomy. Some attempts have been made to ground consent on considerations other than respecting autonomy, but whether these grounds could also support the trust requirement is unclear. So we still lack independent grounds for the trust condition. The argument being considered tries to turn the appeal of consent in favor of the trust condition, but in doing so leaves unanswered the question of the grounds for requiring a trust relationship in order to justify paternalism. Until we have some autonomyindependent grounds for the requirement of consent and can show that these grounds also or instead support the trust condition, then the argument is unsatisfactorily incomplete. C. A Lockean Argument The concept of trust plays an explicit role in the political philosophy of John Locke. According to Locke, government is legitimate when and only when individuals in the state of nature enter a two-stage social contract. In the first stage, people agree to unite themselves, forming a civil society in order to avoid the inconveniences of the state of nature. In the second stage this society entrusts political power to a government. If government exceeds the limits of its rightful power, it has betrayed the trust given it by the people, is therefore illegitimate, and rightful power returns to society, which may remove that government and set up another in its place. (Locke, 1689, pp. 126ff. Laslett’s introduction is an interpretation of Locke that emphasizes the notion of trust.) On this argument, trust is a necessary condition for legitimate government and therefore a necessary condition for legitimate government paternalism. While this is narrower than P1, which requires trust for all

62

SIMON CLARKE

coercive paternalism whether from government or others (I owe this point to Diane Proudfoot), it provides a premise that would result in a conclusion that would rule out CMP by government, which is Raz’s main concern. However, the centrality of the notion of trust in this argument is illusory. According to Locke, the legitimate power that government is entrusted with is defined by the protection of the natural rights to life, liberty, and property that individuals have in the state of nature. It is due to the uncertain protection of these rights that people agree to form civil society and it is for the protection of these rights that people agree to form civil society and entrust their protection to government. Trust itself is doing no work in the argument; legitimacy of government turns more fundamentally on whether it is protecting and enforcing people’s natural rights. It might be thought that this is not a significant difficulty because even understood in this way, Lockean theory still provides an argument against CMP — such paternalism would violate people’s natural right to liberty. But even if such an argument were sound, it is not the type of argument that Raz seeks and that we are investigating. Recall that the trust argument is intended to show how trust is of independent moral force as an argument against paternalism. It is due to what he sees as the weaknesses of other views of the foundations of liberty, including natural rights views (examination of these and other foundations of liberty is the main theme of The Morality of Freedom), that Raz turns to the trust argument. The Lockean argument fails to show how trust is independently necessary for legitimate paternalism. D. A Functional Argument Perhaps trust is necessary for legitimate government because without trust, government could not even function. If people do not trust government, it will be unable to command obedience, will not be able to make its policies effective, and so on. Tom Tyler has argued that voluntary compliance is needed for government to be effective and that trust is crucial in motivating voluntary compliance. (Tyler, 1998. The argument that government needs trust to be effective goes back at least to Confucius. In the Analects, Book 12 verse 7, he states that without trust, government cannot stand. I thank Shui Chuen Lee of National Central University, Taiwan for this reference.) Against this, Russell Hardin maintains that governments of many countries, such as “Naziruled Czechoslovakia, (. . .) Spanish rule of southern Italy, medieval rule of randomly conquered regions, and the rule of different Chinese empires and many colonial governments” have functioned through the use of coercion instead of voluntary compliance. (Hardin, 1998, p. 10.) The claim that effective government requires citizen trust, Hardin concludes, is conspicuously false. But three weaker claims about the importance of voluntary compliance and trust in government could be made, which although weaker, may be

A Trust-Based Argument Against Paternalism

63

sufficient to rescue P1 of the trust argument: (1) Citizens must not have extreme distrust of government if it is to be effective (Hardin, 1998, p. 10). (2) Trust is an important good for effective government (Hardin, 1998, pp. 10–11). (3) Trust is necessary in some areas of government action (Hardin, 2002, p. 152). Hardin does not elaborate, but the thought behind the first claim would appear to be that while it is possible to govern effectively using coercion instead of voluntariness, if the lack of trust is extreme, there will be such disobedience that even coercive government will not function. The second claim concedes that trust is not necessary for effective government but maintains that it is highly important — government that enjoys voluntary compliance is able to implement its decisions more easily and can use its scarce resources that would otherwise be spent on securing compliance, for other purposes. The third claim says that some government actions do require voluntary compliance — Hardin mentions military drafts and payment of taxes. Another example that is more one of paternalism could be enforced religious conversion. As Locke argued, forcing someone to pray do not convert that person since what is sought is genuine belief. Policies for religious conversion hence require some voluntary compliance to be effective. (Locke, 1689.) Each of these claims is open to much discussion regarding their truth and implications. But here we want to know their relevance for supporting P1 of the trust argument. If they are all true, that would show (1) that for paternalism to succeed it must not be accompanied by extreme distrust of the paternalistic government, (2) that trust would be an important good for paternalistic government, and (3) that trust may be necessary if paternalism involves areas of government action that require voluntary compliance. Taken together these claims do not add up to the blanket prohibition against CMPwithout-trust that P1 of the trust argument would establish. That condition forbids a government from proceeding with CMP without trust, but on these three conditions it may proceed so long as (1) the lack of trust is not extreme distrust, (2) the gain in welfare to the target of paternalism is enough to outweigh the opportunity cost of whatever else government could do with the resources it has to use to enforce the paternalism, and (3) the paternalistic action does not regard an action that requires voluntary compliance. Given the difficulties government faces if it is to make fine-grained policies, these three claims may provide sufficient reason for it to adopt the more general rule against CMP that P1 sets. Governments are fallible and if they try to engage in CMP while respecting each of the three claims above, they will sometimes be paternalistic when paternalism is unsuccessful — that is, when intervention leaves the targets worse off — and will sometimes not

64

SIMON CLARKE

be paternalistic when paternalism would have been successful — that is, would have benefited the targets. How to weigh up these gains and costs is a difficult matter. (For discussion see Arneson, 1989.) The thought that government could coerce people for their good but end up making them worse off is a scary one, but people leading bad lives when they could lead better are also consequences that must be taken into account. If the costs of unsuccessful paternalism would outweigh the gains of successful paternalism then government would have a rule-consequentialist rationale for the trust condition P1. However, if on the other hand government is extremely competent it will be able to engage in CMP while respecting the three claims above. It would not have to abide by the trust condition. As we will see in the next section, the competence of government will affect not only whether it ought to be bound by the trust condition but also how reasonable it is to trust it when it engages in CMP. For the moment I conclude that some reason exists to accept P1 of the trust argument though for different reasons than Raz suggests. Because of the three claims above, government might sometimes accept P1 when this would have better consequences than having a more fine-grained policy of trying to work around the three claims. This conclusion is qualified because there may be circumstances in which government could do a better job by working around the three claims instead of accepting P1. So P1 can be taken as true only in particular conditions instead of generally so.

3. Is it Reasonable to Trust Paternalistic Government? I turn to the second premise: a person subjected to CMP cannot reasonably trust the government. Why is this true? Imagine that government is genuinely concerned with my interests and pursues what it takes to be well-justified paternalistic policies and that I know this. Trust in this case would appear to be reasonable. However, according to Raz, even in such circumstances, such trust would be unreasonable. The argument, in short, is that by preventing me from following my way of life, the state denies the validity of my fundamental beliefs concerning how to live my life, which thereby makes it impossible for me to think that the government is seriously taking my interests into account when making decisions, and that this is sufficient to make trust unreasonable. To elaborate, Raz claims that before a person may reasonably trust the government, she or he must enjoy full citizenship. Enjoying full citizenship means she or he feels (and is not unreasonable to so feel) that the government recognizes her or him as a person who matters in her own right, that her or his fate is a matter of intrinsic value in the eyes of the state (LT, p. 124). She or he reasonably believes that her or his interests are given due consideration by the government when it makes its decisions. The condition of citizenship contains a subjective element. It depends not only on how the government treats a

A Trust-Based Argument Against Paternalism

65

person but also on how she or he feels she or he is being treated, so long as that feeling is not unreasonable. According to Raz, citizenship is a precondition of trust. He writes that “[t]hose who find it impossible to regard the government as their government, and to regard themselves as full citizens alongside everyone else, are unable to trust the government.” (LT, p. 127) Next, Raz claims that most paternalism undermines the existence of full citizenship. He says that if the government pursues coercive moral paternalism against me it will, by definition, be preventing me from following my way of life, and it denies, in a purported exercise of its authority, the validity of propositions I hold true and which underpin my way of life. If it does so, however, it denies me full citizenship. (LT, p. 127, emphasis omitted.) In doing so it “undercuts my trust that I and my interests are seriously being taken into account in deciding public action.” (LT, p. 127) Someone whose fundamental values and beliefs are rejected as worthless by the government will not trust the government as acting in her or his interests. CMP may be motivated by concern for the well-being of the subjects of the coercion. The government may view them as of intrinsic worth in their own right. But the crucial question, on Raz’s argument, is whether the subjects of paternalism will see it this way, and would not be unreasonable to do so. Raz argues that the subjects would not feel the same way as the paternalist. He argues that even if the government is acting out of the best intentions, it is bound to seem to them as failing through moral blindness to take them and their well-being seriously, lacking the ability to give their interests the weight they deserve. The people who feel like that may be wrong. Their beliefs and way of life may indeed be worthless. But it is reasonable for them to deny that and to feel that the government has forfeited their trust. (LT, p. 127) So we have two steps of the argument for P2: X. Government is reasonably trusted by a person only if she reasonably feels that the government recognizes her as someone who matters intrinsically. Y. A person cannot reasonably feel that the government recognizes her as someone who matters intrinsically if it coercively prevents her from following her way of life, which is what CMP does (by definition). Raz’s argument for X is that a person may reasonably trust government only if she has full citizenship and that a person has full citizenship only if she reasonably feels that the government recognizes her as someone whose interests matter intrinsically. I have no quarrel with these claims, although we might question whether this is an accurate account of the meaning of citizenship. Be that as it may, we should accept X as true. It is implausible that

SIMON CLARKE

66

we could reasonably trust someone that we believe does not take our interests as mattering in their own right. Some belief in good intentions is a necessary condition for reasonable trust. But what about Y? It is true that CMP denies the validity of the propositions that underpin my way of life, and may condemn these fundamental beliefs as base and worthless. Paternalism is motivated by the belief that the target’s views are mistaken. But is this a sufficient reason to think the government does not take my interests seriously? It appears plausible to think so, but I will argue that it is not. Crucial to my argument is a distinction between content-dependent and content-independent reasons for trusting or not trusting. The first refers to whether or not the subject of paternalism agrees with rationale for the paternalism, whereas contentindependent reasons make the reasonableness of trust depend on factors other than the substantive content of the decision. An underlying assumption of claim Y is that reasonable trust turns only upon or strongly upon contentdependent reasons. Raz’s argument is that it is unreasonable to trust government when a person disagrees with the content of a decision it makes, especially where the disagreement is over something as fundamental as one’s way of life. It assumes this to be so even though there may be strong contentindependent reasons for trusting. To challenge this assumption I will outline several areas of content-independent reasons for trusting by drawing on some of the literature on trust and government. These reasons appeal to procedural fairness, competence, trust-responsiveness, and encapsulated interest. My claim is that when there are sufficient content-independent reasons for trusting, anyone who trusts a government engaging in CMP is showing reasonable trust, even though government is denying his or her fundamental beliefs. A. Procedural Fairness Imagine I know that the government has met the following conditions: x

x

x x

Participation: citizens have opportunity to have input into the policy via submissions to government committees, committee meetings being open to the public, etc. Considered judgment: the government has thought long and deeply about the issue when arriving at its decision, considering all the arguments or all that are available on either side of the issue. Publicity: government takes adequate steps to explain the rationale behind the paternalism, to explain how it thinks people are mistaken. Consistency: the paternalistic policy is applied to all those who meet the relevant criteria, not arbitrarily applied to a select few.

A Trust-Based Argument Against Paternalism

67

If I know that these conditions are met, then I would have reason to believe that the government is seriously taking my interests into account in its decision-making. These conditions of procedural fairness are contentindependent reasons for trusting in that their being met is compatible with a wide range of substantive outcomes. Claim Y assumes that even if these conditions are satisfied, any trust that someone has toward a government that is engaging in CMP is unreasonable trust. But people could, I suggest, reasonably trust a government that makes decisions they disagree with so long as those decisions are made according to legitimate processes. Government having fair procedures signals that it takes peoples interests seriously. Empirical evidence shows that citizens do trust government when they believe procedures are fair even though the outcomes are unfavorable to them. (Levi, 1998, pp. 88, 91. Levi references for the evidence in Frohlich and Oppenheimer, 1992; Tyler, 1990.) A survey of why United States citizens trusted or distrusted government showed that substantive policies were of little importance. More significant were the caliber of personnel, accountability, transparency, honesty, and lack of corruption. (Jennings, 1998, see especially the tables on pp. 233–238.) Raz could deny that this empirical evidence has any bearing on whether trust is reasonable. However, once the conditions are spelt out it is intuitively plausible that it is reasonable to trust under such circumstances. B. Competence Imagine that government also meets the following conditions: x x

x x x

Success: government succeeds in forcing me into the way of life it believes is valuable. Localized intervention: it does its utmost to ensure that the coercion is as localized as possible so that it does not interfere with my other options. Significance: the mistake is of such magnitude that my life will be significantly worse. No alternatives: no less intrusive measures would achieve the same outcome. Trust transference: if government acts competently in nonpaternalistic spheres, such as in its protection of property rights and provision of public goods, the trust it builds in those spheres makes it more reasonable to trust government in its paternalistic sphere. (Levi, 1998, pp. 84–5. But there are counter-examples: “Citizens in the Anglo-Saxon democracies, at least at the start of World War I, seemed to trust their governments to enforce property rights but were not so willing to trust them with the power to conscript.” Levi, 1998,

SIMON CLARKE

68 p. 85.)

Once these conditions are met, something more than good intentions has been demonstrated. The government has also demonstrated willingness and an ability to put its good intentions into action. This willingness and ability make it reasonable for me to think the government believes my interests count, even if I am right and it is wrong about what is in my interests. When these conditions are met, someone subject to CMP could reasonably trust government. C. Trust-Responsiveness A further consideration lies in the self-fulfilling nature of trust. It is more reasonable to trust government if it is trust-responsive. Trust-responsiveness occurs when the person trusted takes the trust shown him or her to be an additional motivation for being trustworthy. (Pettit, 1998.) To illustrate, say that someone I know has reasons for behaving toward me in a particular way. If I trust that person to so behave, then if she or he is trust-responsive, she or he will take that trust to be another reason, above and beyond the reasons she or he already has, to behave in that way. And that fact — the fact that she or he is trust-responsive — gives me an additional reason to trust her or him. If citizens know that in addition to the reasons government has for acting the way it does, government will take the trust shown it to be an additional reason, then they are more likely to trust government. Here is how Philip Pettit puts it: The appearance of personal trust among the citizens, then, can actually increase the grounds that people have for feeling trust [if government is trust-responsive]. For when citizens trust government agents to do that which the citizens apparently only have reasons of trustworthiness to expect, then in reality there are also other reasons for expecting those agents to comply. (Pettit, 1998, p. 307.) If government is trust-responsive in this way, then it is more reasonable for citizens to trust it, even if it is engaging in CMP. Pettit argues that its desire for good opinion will provide government with sufficient reason to be and appear to be trust-responsive. Perhaps, but there may be other reasons. In particular in the present context, government will have reason to be trustresponsive because — given the trust condition — it knows that for its paternalism to be justified it must have the trust of people. If government wants to advance the welfare of its citizens through paternalistic policies, but it is bound by the trust condition that says that paternalism must be accompanied by trust, then government will want to do all it can to ensure that it is trusted. And one thing it can do is take a position of trust-responsiveness.

A Trust-Based Argument Against Paternalism

69

Doing so will make it more likely that it is trusted, and thereby improve the chances of being able to engage in justified CMP. Here again is another content-independent reason for trusting paternalistic government. So long government is trust-responsive in this way, which it has strong motivation to be if it wants to engage in legitimate CMP, then people have reason to trust it even if it is acting paternalistically. D. Encapsulated Interest Trust is more reasonable if it is in the trustee’s interest to do what is in the trustor’s interest (Hardin, 1998, pp. 12–13; Hardin, 2002, chapter 1). It is more reasonable to trust a government engaging in CMP if you know it is in the government’s interest to act in your interests. The typical way in which citizens’ interests are encapsulated with government’s interests is via democratic elections. Elections make it more likely that government will act in people’s interests because if it does not, it will be elected out of office. An obvious difficulty in using this mechanism of encapsulating interests to make paternalistic government trustworthier is that it will only work if citizens know and vote according to their best interests, but the types of policies we are discussing — paternalistic ones — are precisely those in which people do not know their best interests. Elections will provide government with an incentive to refrain from paternalism when it may make them unpopular even though that paternalism would be good for people. One way around this difficulty is to seek a mechanism other than elections, some institutional arrangement that will provide a self-interested incentive for government to make sure its paternalistic policies are trustworthy. For example, an independent branch of government such as the judiciary could be given the task of examining CMP to ensure it is justified. Courts in many countries already have the power to overturn government policies and legislation, and governments have interests in avoiding decisions that may be overturned. The possibility of the courts finding government CMP unjustified would give government an incentive to make sure that its policies are sound. Under such arrangements, people would have reason to trust the government because they know that courts will ensure that it is in government’s interest to further people’s interests. (But this still leaves the problem of whether and how the courts have an encapsulated interest in protecting people’s interests.) E. Two Scales of Reasonable Trust Summarizing thus far, I have been attempting to show that contrary to claim Y of Raz’s argument, it can be reasonable to trust a government engaging in CMP. If such government conforms to fair procedures, is competent in paternalistic and non-paternalistic spheres, is trust-responsive, and has an interest in furthering its citizens’ interests (say due to possible chastisement

70

SIMON CLARKE

from the judiciary), it is reasonable to trust it. When the government takes such steps, it would be difficult to think that it is not acting in order to further what it sees as in people’s interests. For brevity, I will refer to these criteria collectively as CIRTs — Content-Independent Reasons for Trusting paternalistic government. CIRTs provide some reason for doubting P2, but they should not lead us to reject it completely. They may well ameliorate the distrust that CMP would create: if all CIRTs are met, there may be less distrust than if they are absent. But even if they are all met, still the suspicion exists that someone forced into a way of life she or he disagrees with could not reasonably trust the government: despite all the procedural fairness, government competence, trust-responsiveness, and encapsulated interest, this person is still being forced into a lifestyle that — in her or his view — has no value. Our question is whether someone subjected to CMP can reasonably trust the government, but we still have no clear answer, just reasons on each side. A way forward, I suggest, is to think of the issue as involving two sliding scales. The first scale is that of the CIRTs. The reasonableness of trusting government increases the more that the CIRTs are satisfied, and decreases the less they are met. This assumes that the degree to which the CIRTs are met can be measured. Each of the criteria in the CIRTs can be met to a greater or lesser extent: each of the fair procedures can be less-than-fully satisfied; the competence of government admits of degrees; and similarly for how much trust-responsiveness and encapsulated interest exists. With suitable weightings, these factors can be combined to give an overall measure of how far the CIRTs are met. The second scale measures the extent of CMP. The claim that it is reasonable not to trust CMP is also, I suggest, a matter of degree. Taking the example of homosexuality that Raz uses, imagine a society in which such conduct is criminalized, homosexuals are discriminated against, and portrayals of homosexuality in the media are forbidden (LT 126). In such circumstances there would be strong reason for gay people not to trust government, and for the reason Raz gives: government denies the validity of the beliefs and fundamental values that underpin their way of life, the essential aspects of their way of life are rejected as base and worthless. But there could be other forms of CMP that do these things to a lesser degree. There are at least two dimensions along which cases of CMP could vary. First they could vary in how fundamental the beliefs and values that are denied are. The homosexuality example is at an extreme end on this dimension since sexuality is so central to most people’s identity. But other parts of one’s way of life could be less fundamental. Say government prohibits a form of literature that I enjoy to such an extent that it forms part of my way of life, though not a crucial part. This is still a case of CMP but one that denies beliefs and values that are less fundamental than those concerning my sexual orientation. Other things equal, it is less reasonable to trust government that

A Trust-Based Argument Against Paternalism

71

prevents me following my sexual orientation than one that bans my favorite type of novels. The second dimension of the CMP scale is in how much denial exists. Consider again the homosexuality example. What occurs in the circumstances Raz imagines — criminalization, discrimination, etc. — amount to a strong denial of that way of life. But there could be less strong denials. Some forms of CMP would amount to government saying that the activity it is not worthless, in the sense of no value at all, but worth less in the sense as not as valuable as other activities. Consider two sets of policies a government could have. In the first, it tells people that homosexuality is wicked, immoral, base, and depraved. In the second it tells people that it is not wicked etc but not as virtuous as heterosexuality. The first involves a greater degree of denial than the second. Other things equal, it is less reasonable to trust the first types of CMP than the second, since it is less reasonable to trust government that condemns your sexual orientation completely than government that views it as merely not as worthwhile as other orientations. Putting these considerations together, we reach a conclusion about the reasonableness of trusting CMP. Assume we can measure how far the CIRTs are met on a scale with high numbers for a high degree of satisfaction (fair procedures, competent government, etc) and low numbers for a low degree (unfair procedures, incompetent government, etc). Some way of measuring these criteria and combining them to give an overall score is needed. But without working that out here, I will assume that some way exists. Call the resulting scale the CIRT index. Assume we can also combine the two dimensions of CMP outlined above — how fundamental are the beliefs denied and how strongly they are denied — and also measure them on a scale whereby the high numbers mean the beliefs denied are fundamental and/or strongly denied and the low numbers mean the reverse. Call this the CMP index. The two scales, CIRTs and CMP, together determine the reasonableness of trusting paternalistic government. Strong forms of CMP can be reasonably trusted, and therefore justified, only if they are accompanied by the CIRTs being met to a sufficiently high degree. The less the CIRTs are met, the less reasonable trust will be and the less justified CMP is. For any given level of CMP it is reasonable to trust government only if there are adequate CIRTs. Figure 1 represents these ideas graphically.

SIMON CLARKE

72

Figure 1

1 Line X

CIRT

Bo A o

0 CMP

Line X is similar to indifference curves used in economics. At each point on the line it is equally reasonable to trust government engaged in CMP. With relatively mild forms of CMP, it is reasonable to trust government if and only if it achieves a relatively low score on the CIRT index (point A). With stronger forms of CMP, it is as reasonable to trust if and only if government meets the CIRTs to an even higher degree (point B). As government’s CMP gets stronger the CIRTs must rise in order to maintain the same reasonableness of trusting it. Other lines of reasonableness could be drawn parallel to Line X. Lines above and to the left would represent more reasonable trust — for a given CMP, the CIRTs would be greater than with line X. Lines below and to the right of Line X would represent less reasonable trust, since CMP would be stronger for a given level of CIRTs. I will assume that Line X is the level of reasonableness just sufficient to justify CMP. Several questions are left unanswered here. First, how steep is Line X and other lines? To answer this, we would have to know whether the increase in CIRTs has to be large or small for a given increase in CMP. Second, line X is shown as a straight line, but perhaps it ought to be curved upwards. This would be plausible if, as CMP takes stronger forms, CIRTs must be increased at an increasing rate in order to maintain the reasonableness of trusting. Third, does Line X intersect at the point of origin or does it connect with the vertical axis at some level above zero? The second would indicate that some positive level of CIRTs is needed before government can even begin to justifiably use CMP. Fourth and finally, is there some point where Line X stops/ becomes vertical or does it continue indefinitely? The first would be the case if there

A Trust-Based Argument Against Paternalism

73

were a type of CMP at which no amount of CIRTs can be enough to make it reasonable to trust government. Perhaps some forms of CMP are so strong that trust is never reasonable, even if CIRTs are high. (I thank Lesley Knight for raising some of these questions.) Setting aside these questions for now, we still have a rough answer to whether it is reasonable to trust CMP. It is reasonable if and only if the corresponding level of CIRTs are met. 4. Conclusion Recall the original argument: P1: Coercive paternalism is justified only if the paternalist is reasonably trusted by the coerced. P2: A person subjected to CMP cannot reasonably trust the government. Therefore, C: CMP is not justified. I have questioned both P1 and P2. Raz’s argument for P1 is weak, and I examined other arguments for it. The most promising was the functional argument that emphasizes the importance of trust for effective government. This argument supports P1 but only if government lacks competence and only if the consequences of unsuccessful paternalism would be sufficiently bad. Without these assumptions the argument at best supports three ways in which government should be cautious in engaging in paternalism (see section 2 D). P2 was also questioned. If my arguments are correct then we must conclude that the reasonableness of trusting government engaged in CMP is not as straightforward as P2 suggests. It may or may not be reasonable to trust government engaged in CMP, depending on how strong that CMP is and how well-arranged the political institutions are that provide CIRTs. So the trust argument could be restated thus, replacing P2 with P2*: P1. Coercive paternalism is justified only if the paternalist is reasonably trusted by the coerced. P2*. It is reasonable to trust government engaged in CMP if and only if that CMP is in proportion to the content-independent reasons for trusting government. Therefore, C. CMP may or may not be justified, depending on the correlation between a government’s CMP and its CIRTs. This is not an exciting conclusion I admit, but sometimes the truth is not exciting. However, perhaps a less vague conclusion can be drawn. For the stronger forms of CMP to be justified they must be accompanied by political arrangements that rate highly on the CIRT index. But high levels of CIRTs are

74

SIMON CLARKE

far removed from reality. Most governments do not achieve procedural fairness, competence, trust-responsiveness, and encapsulated interest to a high degree. No government meets them perfectly and most would score lowly. It is quite plausible to withhold trust from government that wants to engage in stronger forms of CMP if the CIRTs are not met. Since the CIRTs do not hold highly in the real world then stronger forms of CMP cannot be reasonably trusted. This gives us support for a different version of P2, P2**: that we should not trust government if it engages in strong CMP (at least until the CIRTs are better met). So we can restate the argument again, this time with P2** for P2 and P2*: P1. Coercive paternalism is justified only if the paternalist is reasonably trusted by the coerced. P2.** A person subjected to strong CMP cannot reasonably trust the government (at least in real world conditions). Therefore, C. Strong CMP is not justified (at least in real world conditions). While in principle coercive moral paternalism could be justified if government meets particular conditions to a high degree, until that time strong forms of such paternalism are not justified. I thank the participants of the trust workshop at Macquarie University in July 2005, an audience at the Value Inquiry Conference 2005 at Louisiana State University and a seminar audience at the University of Canterbury for their helpful comments.

WORKS CITED Arneson, Richard. (1989). “Paternalism, Utility, and Fairness,” Review of International Philosophy, 43:170, pp. 409–437. Baier, Annette. (1995a). “Trust and Anti-Trust,” Moral Prejudices: Essays on Ethics, Cambridge, Mass.: Harvard University Press. ———. (1995b). Moral Prejudices: Essays on Ethics. Cambridge, Mass.: Harvard University Press. Braithwaite, Valerie and Margaret Levi, eds. (1998). Trust and Governance. New York: Russell Sage Foundation Caney, Simon. (1994). Liberal perfectionism. Oxford, England: University of Oxford, D.Phil. thesis. Cahn, Steven M., ed. (1997). Classics of Modern Political Theory. New York: Oxford University Press. Dworkin, Ronald. (2000). Sovereign Virtue: The Theory and Practice of Equality. Cambridge, Mass.: Harvard University Press.

A Trust-Based Argument Against Paternalism

75

Frohlich, Norman and Joe A. Oppenheimer. (1990). Choosing Justice. Berkeley, California: University of California Press. George, Robert P., ed. (1996). Natural Law, Liberalism, and Morality. Oxford, England: Clarendon Press Hardin, Russell. (1998). “Trust in Government.” In Trust and Governance, eds. Valerie Braithwaite and Margaret Levi. New York: Russell Sage Foundation, pp. 9–27. ———. (2002). Trust and Trustworthiness. New York: Russell Sage Foundation. Jennings, M. Kent. (1998). “Political Trust and the Roots of Devolution.” In Trust and Governance, eds. Valerie Braithwaite and Margaret Levi. New York: Russell Sage Foundation, pp. 218–244. Levi, Margaret. (1998). “A State of Trust.” In Trust and Governance, eds. Valerie Braithwaite and Margaret Levi. New York: Russell Sage Foundation, pp. 77– 101. Locke, John. (1689a). Two Treatises of Government, ed. Peter Laslett, 1965, New York: Mentor Books. ———. (1689b). Letter Concerning Toleration. In Classics of Modern Political Theory, ed. Steven M. Cahn, 1997. New York: Oxford University Press. McCabe, David. (2001). “Joseph Raz and the Contextual Argument for Liberal Perfectionism,” Ethics, 111:3 (April), pp. 493–522. Mill, John Stuart. (1859). On Liberty, ed. David Spitz, 1975, New York: Norton and Co. Norman, W. J. (1989). “The Autonomy-Based Liberalism of Joseph Raz,” Canadian Journal of Law and Jurisprudence, 2:2, pp. 151–162 Pettit, Philip. (1998). “Republican Theory and Political Trust,” In Trust and Governance, eds. Valerie Braithwaite and Margaret Levi. New York: Russell Sage Foundation, pp. 295–314. Raz, Joseph. (1986). The Morality of Freedom. Oxford, England: Clarendon Press. ———. (1989). “Facing Up: A Reply,” Southern California Law Review, 62:3–4, pp. 1153–1235. ———. (1995). “Duties of Well-Being.” In Ethics in the Public Domain. Oxford, England: Clarendon Press. ———. (1996). “Liberty and Trust.” In Natural Law, Liberalism, and Morality, ed. Robert P. George. Oxford, England: Clarendon Press, pp. 113–129. Tyler, Tom. (1990). Why People Obey the Law. New Haven, Connecticut: Yale University Press. ———. (1998). “Trust and Democratic Governance.” In Trust and Governance, eds. Valerie Braithwaite and Margaret Levi. New York: Russell Sage Foundation, pp. 269–314. VanDeVeer, Donald. (1986). Paternalistic Intervention: The Moral Bounds on Benevolence, Princeton, New Jersey: Princeton University Press. Waldron, Jeremy. (1989). “Autonomy and Perfectionism in Raz’s Morality of Freedom,” Southern California Law Review, 62:3–4, pp. 1097–1152. Wilkinson, T. M. (1999). “Review of Natural Law, Liberalism, and Morality,” Utilitas, 11:1 (March), pp. 134–137. Wolfe, Christopher. (1996). “Being Worthy of Trust: a Response to Joseph Raz.” In Natural Law, Liberalism, and Morality, ed. Robert P. George. Oxford, England: Clarendon Press.

Five TRUST AND RISKY SECRETS Timo Airaksinen 1. Dangerous Mistrust Katherine Mansfield’s tragic short story The Stranger tells a story about risky secrets and the lack of trust (Mansfield, 1946). A husband waits in a harbor to meet his wife who is coming back to New Zealand from Europe. The ship takes a long time to moor. Finally they are able to meet and talk, but the husband becomes suspicious, as if he did not trust her: “she was keeping something from him (. . .) He thought he’d noticed just something.” He thinks she has a secret, and it is clear he does not truVWKHUZKDWHYHUWKHLVVXHLVíKH does not know yet. He cannot allow her to have secrets now when she is finally back. “You are thinking of something else,” he says as if he were interrogating her. Later she tells him in an innocent and trusting manner of how a young man, a perfect stranger died in her arms on the ship. He had a weak heart and the doctor was late. The husband is shocked, and the story ends when he realizes the chilling truth during an intimate moment: “They would never be alone together again.” The stranger’s ghost will always be there between them harassing him. What has happened? When they talk, after a moment’s hesitation she trusts him and thinks that her story can and should be shared. She does not think it is necessary and perhaps not even appropriate to keep it secret. Yet he cannot handle the crucial piece of information, obviously because he does not trust her anyway (in the marital, shady sexual sense of trust). From his male point of view, the stranger’s and the wife’s intertwined story should have been kept as a secret. He need not know it, and he should not have heard about it. But all this implies his lack of trust. Yet she trusts him and therefore she tells him her story. Alas, her trust is misplaced in a tragic manner. Now her miserable husband knows her story, and perhaps even the truth, which creates an intolerable contradiction and psychological conflict. He does not trust her yet he knows about the stranger. The consequences are severe. His inability to trust brings about severe mental pain to him. He is no longer able to think that his wife unconditionally belongs only to him. He does not trust her and therefore he sees the stranger taking a position between him and his wife. Love without trust means control, possessiveness, and jealousy and this is what bothers the husband. Could he still be able to possess his wife fully, as he obviously used to be? The answer is in the negative.

78

TIMO AIRAKSINEN

Mansfield’s short story allows us to see a couple of important points about trust and secrets. First, to allow another person to keep a secret requires trust. Secrets may be dangerous to those who do not share them. And you should not tell a secret to a person you do not trust. If you tell, bad things can happen. You may get hurt. Such a person may disclose your secret, but she may also harm you in some other manner using her newly acquired information as a weapon. The husband in Mansfield’s story will hurt his wife before or later, that conclusion is unavoidable to the reader. It is interesting to notice how and why mistrust may be dangerous. Sometimes it is better to trust and let other people keep their secrets than to make them reveal them. If you do not trust, you should not know too much; but because you do not trust you want to know everything. Only the strongest can afford their mistrust because what they want to know may hurt them. The inquisitive weak destroy themselves, like Mansfield’s husband. In many cases benevolent persons keep information secret precisely because they do not want to take the risk of hurting another person. This is exactly what the wife in Mansfield’s story does not understand. The suspicious husband wants to know and she tells him. You should not want to know everything and therefore secrets protect you. You just need to trust others and believe that they are benevolent toward you when they do not tell, even if they had something to tell. If you fail to follow this rule, your mistrust may prove costly. You should not ask if you cannot cope with the consequences of the answer. In that case, it is better to trust. 2. The Risks of Shared Secrets The concept of secret means, first, that someone has important information he does not want to share with others. Second, this information can be willingly and freely shared but it is still a secret if one does not want others to know it. Sometimes we know that a person has a secret although we do not know what the secret is. But sometimes it is a secret that I have secrets, sometimes it is not. (Bok, 1982.) Two principles should be kept in mind and they form the logical core of this paper: (1) If you are allowed to keep a secret, unchallenged, you are trusted. (2) If you freely and willingly reveal your secret, you are in a trusting position. Two different relations exist between secrets and trust. You only want to share a secret with a person or institution you trust. Otherwise, the situation is too risky. I mean by risk the uncertainty of the outcome. (Gigerenzer, 2002, p. 26.) But a second case also exists: you allow a person to keep a secret from you because you trust the person, or if you cannot do anything to change the situation. Any undisclosed information may harm you. You just do not know.

Trust and Risky Secrets

79

Often we need to trust persons because we cannot make them reveal their secrets. We do not have an alternative. But the obvious question is whether this constitutes trust in any realistic sense of the word, or is it merely as-if trust? I want to argue that this may be proper trust, but only under special conditions. What these conditions are is one of the main points of this paper. Hence the question is this: Can we infer that A trusts B when B allows A to keep his secrets? The answer is given in (2) by means of a reference to the idea of freely and willingly shared secrets. If you reveal your secrets, say, under the threat of torture, you are not in a trusting position. Let us next look at the case in which A does not trust B who is known to have a secret. The undisclosed secret may be of negative expected utility to A in the sense that A does not want it to become publicly known. Thus its disclosure is as such undesirable from A’s point of view regardless of the future consequences of the act. What should A do? The general rule is to make B reveal the secret to A alone. Or if this is not possible, try to create so much trust between A and B that the disclosure of the secret becomes sufficiently improbable. The reason behind the first strategy from A’s point of view is that then he or she at least knows what the issue is and can take countermeasures. If the secret stays as a secret, A is helpless. Therefore, always try to know another person’s secrets, if you can. But a caveat must be kept in mind, the fate of Mansfield’s poor husband who could not cope with what he came to know. A narrative illustrates the case. An old priest tells a dramatic story in a party: “Can you believe that my first penitent confessed a murder?” Next a middle-aged man enters the room and says: “Can you believe that this good father was once my confessor and I was his first penitent?” The truth has been revealed. Both the priest and the young man are now in serious trouble. Obviously to share the secret with the confessor was risky, even if the priest was supposed to be a fully trustworthy agent. And the confessor’s knowledge was dangerous to himself as well. He knew too much for his good. Perhaps we should accept the principle No secrets without a risk. This appears to be intuitive as the story of the confessional shows. Why should we otherwise create and keep secrets? Secrets logically entail a risk. They may be dangerous to all the parties involved, also to the person who only shares the secret (like the priest) but does not own it (like the murderer). Owners have a reason to want to keep a fact to themselves but they may freely and willingly share it with other persons. Then they become trustors. It is still a secret, if the relevant group of persons is limited. Curiously, also public secrets exist, that is, those known by all but not discussed by any because of (say) the generalized shame or danger it causes. All the citizens of Wonderland may know that their President is a crook and a pervert, but they never want to mention it publicly. This is a public secret which has its own obviously dangerous aspects as well. That is why it becomes unmentionable. Many types of risk are connected to secrets. There are conventional

80

TIMO AIRAKSINEN

risks: I do not want other people to know them, as if my secrets had intrinsic value. Perhaps this is somehow related to privacy. Sometimes we want to keep our secrets simply because they are secrets. We need some secrets to make our lives and personalities unique, valuable, and protected. My secrets are my own, and you are allowed to share them on the condition that you respect my ownership. You should not mishandle my secrets, if you still respect me. We even create new secrets to make it certain that we are full persons whose existence is meaningful. In other words, a socially fully transparent person lacks existence. Secrets make us more solid. We can then ask whether this refutes my earlier claim that there is no secret without risk. The answer must be in the negative because if we create personal secrets we must also create a corresponding idea of a risk, however trivial. This makes the difference between merely undisclosed information and a true secret. If it does not matter whether other people know it, it is not a secret. But it must be admitted that this is a limit case. Genuine harm (think of the cruel fate of the young murderer in the story of the confessional above) is also possible, as well as nuisance and shame (perhaps the most common and obvious case, compare the priest). We notice that the risks are different to the owner and the sharer. Sometimes the sharer faces no risk, but the owner appears always to be in a risky position. The owner’s secrets may always be mishandled so that conventional risks are realized. The temptation is not to share any secrets, but this is an anomalous case. Secrets are meant to be shared with trusted persons. Otherwise, they may become a meaningless burden. This is a paradox in the core of secrets: they are always risky but they are meant to be shared with trustees. Human beings as social creatures find it difficult to live with their undisclosed secrets for too long periods of time. Secrets are risky, and not to share a risk looks difficult. Only solipsistic narcissists keep all their secrets to themselves. On the other hand, all people have a bunch of secrets they are extremely reluctant to share. Normally this is because of shame. But such secrets are nothing but a burden. We would instead live without them. Sharable secrets are different. They are a source of power, joy, and wonder. Fully unsharable secrets diminish a person but shared secrets empower people. However, when secrets are revealed other secrets are created to replace them. In this way we can have our secrets and also share them. As I said, human beings as social beings need shared secrets, but we also want to be solid individuals who have our undisclosed secrets. The needs of sociability and individuality must be satisfied alike. Yet there may be a special class of unsharable secrets which are nothing but a source of pain. We can also say that shared secrets entail trust. If my secret is shared with a person whom I do not trust, the situation is anomalous. It is no longer a secret, or at least I should not regard it as such. This is what the lack of trust does. Trust and secrets are so intimately related that the case where I share a secret without trust is problematic. I mean a case where the sharer of the

Trust and Risky Secrets

81

secret is not a trustee. Yet someone may say that it is not necessary that secrets without trust are no secrets. If this is so, they may still be called secrets although they are now unstable in the sense that the owner of the secrets does not expect them to stay unviolated (or to be kept and not to be used against him or her). Their status as secrets is now unstable in that special sense. It becomes impossible to rely on such secrets, and in this they become idle and meaningless. We can just wait what will happen. The person is left in the hands of enemies. And then secrets are just a matter of hope. But does this not mean that secrets have vanished, when shared with untrusted persons? I suggested that this is what happens. If trust fails, shared secrets vanish. And they vanish even before the untrusted sharer has done anything. An unsafe secret is no longer a secret. In the case of sharing my secrets with a trusted person, I can enjoy my secrets. I play with them, and I can use them in different ways. My friends know that I am a recent Lotto winner, and this shared secret provides me many novel opportunities. But I do not want to spread the knowledge, so I need to trust my friends. Other people do not know how rich I am, and this is both useful and amusing. But a person I do not trust can spoil the fun, and I expect him to do so. In this sense my secret becomes idle. The person who merely shares the secret may be in danger too, as the priest learned. His career was in ruins. Obviously his mistake was not misplaced trust but a kind of logical blunder. But the same kind of story can be told in terms of misplaced trust on the owner of the secret. Suppose X is a petty thief and she or he tells me the truth. Should I fail the trust and tell the Police or remain quiet and keep her dirty little secret? Both alternatives are equally bothersome. There are secrets we prefer not to know, but if we do, at least the owner of the secret must be trustworthy. To know someone’s secrets may be dangerous, and thus a hearer of a secret needs to trust the owner. The thief should not tell anybody that I know. To know is to be responsible, and this entails accountability which entails a risk. If I know that you are a thief I should tell the authorities; if I do not, I am partially responsible for the losses other people suffer because of your criminal actions; but whatever I do, tell or remain silent, I face a typical risk. In many cases it is better not to know. 3. The Risks of Trust Secrets, trust, and risk are connected in a complicated manner. For instance, secrets cannot be shared with untrusted persons because sharing a secret always entails a twofold risk. First, there is the risk of the secret becoming public knowledge and, second, the unfolding of the secret may bring about harm to you. Hence you share your secrets only with the persons you trust. However, most people seem to think that trust as such is always risky. Does this mean that you can never share a secret? You trust somebody and she or he may always fail your trust, and as we

82

TIMO AIRAKSINEN

know, this implies harm. But in the Ideal World it is possible to trust without risk. Thus, risk and trust are not logically connected unlike risk and secrets. For instance, I place my trust in God’s wisdom, and I always get what I deserve. There is no risk. Can I share my secrets with God without risk, at least if I confess them to Him without a middleman? If this is so, the logical link between secrets and risk is broken, and my theory fails. I think we must say that because there is no risk it does not make sense to speak about secrets when we discuss the special us/God relationship. No secrets ever existed and the reason is not only the fact that we cannot hide anything from him but also the fact that we face no risk. People deny the existence of risk when then they talk about trust in God. Because God is all-good, no risk ensues because of our trust. It is difficult to know how to understand such a view. Do we talk about real trust or about some kind of secondary case, a kind of metaphoric trust which need not be taken literally? On the other hand, some religious people say they really trust God and do so in a most literal sense. Perhaps we can conclude that there is no logical connection between risk and trust, although normally trust brings about risk. If the trustee is virtuously trustworthy or otherwise fully trusted the risks of trust are minimized, but never eliminated. But these are only empirical considerations. The fact is that in this imperfect world trust can always be failed. It is possible to trust with confidence, and then risks are not considered. This appears to require a special attitude. The argument that no confidence is at all justified in this context and thus all trust is risky is erroneous in the same way as epistemic skepticism. Our beliefs may always be wrong (skepticism), but this need not worry us if we do not have a special reason for epistemic doubt. In the same way trust is called risky only if we have a reason to think so – and we do not always or necessarily have such a reason. In fact to say that we trust is to say that no checks are needed against the risks inherent in the case. For example, when I say I trust you to keep my secret, I am saying that I am prepared not to consider the risks of the secrets being disclosed. The function of the concept of trust is to eliminate the considerations of risks in the given case. It is a mistake to say that the function is to eliminate the risks themselves. Sometimes we bypass the risks, and this is the case when we trust. In the same way we bypass the consideration of epistemic risks when we present a knowledge claim. Nevertheless, the typical risks are always there. We may say equally well “I know I can trust you” and “I know you will keep my secret.” Hence we can say, “I know I can trust you to keep my secret.” This entails full confidence. As-if trust is a consequence of the experience of real life risks. But sometimes we trust in the full and real sense of the word. The distinction between as-if trust and full trust indicates that sometimes risks need to be considered, sometimes not. Trust may be risk-relevant or not. Accordingly it is sometimes needless to mention risks, although they always exist. (Jones,

Trust and Risky Secrets

83

1996.) In other words, sometimes the risks of trust are not reasonable and we may skip them. In real life you may resort to as-if trust, which allows for degrees, because you often have various different reasons to be careful and not to trust unconditionally. As-if trust can be understood in different ways. Here is one: trust as if your attitude were full trust and take all the necessary precautions against failure and disappointment. There is also a special instrumental and strategic case of as-if trust: A tells B a secret when A expects B to tell this to C; A wants C to come to know it, but B does not know about this intention. A pretends to trust B. Thus, as-if trust can be honest or dishonest. The honest case is understood easily in reference to secrets: a trustor tells the trustee only partial and camouflaged truths and keeps an escape road open. A may be vague and ambiguous. He or she tests B by telling meaningless secrets first and observing the behavior. But A need not make B think that he or she fully trusts. The instrumental case is dishonest. There is also manifest as-if trust, namely, when I say “trust me.” To say that means that the trust is not to be taken in its full sense. Such words are pleas for the would-be truster to trust at will, which may be impossible anyway. The plea is for an impossible achievement, but it is still a plea for trust, as-if trust. When one decides to trust, one trusts only in the as-if -sense of the term. (Baker, 1987, p. 2.) Perhaps this is controversial, but I cannot deal with the problem in any comprehensive manner here. However, the speaker claims that she is trustworthy, but the minimalist speech act falls short of establishing such a demanding and complicated social relation as full trust. And if the speech act does not work, the would-be trustor’s only possibility is to believe at will, which cannot be done. (Holton, 1994.) So it is a plea for asif trust. The plea may achieve something and therefore it is an exaggeration to say that it never works. 4. Undisclosed Secrets and Social Power When we discuss trust, it is useful to consider secrets as well. In fact secrets form a special field where trust plays a major role, especially when secrets are shared with others. I tried to justify this proposition above. The existence of secrets shows how trust is distributed in social life, that is, who is supposed to trust and whom. Discussions concerning secrets are especially useful in the context of trust because we can then specify a special subject matter of trust, secrets. To discuss trust without specifying what we are talking about is often too vague and general. I already mentioned some important aspects of secrets in relation to trust and now we move forward to discuss trust in the context of social power and institutions. Two intuitions are relevant in such a context: First, if A trusts B, A allows B to have secrets. This is obviously true but it is more important to ask whether the fact that A allows B to have secrets implies that A trusts B? Second, A’s demand that B have no secrets implies no trust. These principles need an explanation. A understands that the secrets

84

TIMO AIRAKSINEN

possessed by B create a threat and risk to him or her. Secrets are risky in this very sense, namely, what A does not know may be used against him or her. B possesses knowledge which A does not know, and this is a potential risk to A. Two different cases exist: A may know or not know that B has such knowledge. Anyway, if A is more powerful than B, A is able to make an effort to control the relevant effects of B’s secrets, and has a motive to do so. A wants to eliminate the risks which he or she is or may be facing. A possible risk is a risk too. Therefore A does not allow for B’s secrets, if A does not trust B. If A trusts B, secrets are to be accepted even when A is in a position of power. Power is not relevant when A trusts B. Such a common sense account of trust, secrets, and social power has its problems, namely, it does not track the logic of power faithfully enough. The critical Hobbesian question is both simple and compelling: why would powerful A tolerate the risks associated with trust in B who has undisclosed and unshared secrets? My approach here can be called Hobbesian. (Airaksinen, 2008.) It is simpler to erase trust and insist that B cannot have any secrets. This move eliminates all the risks, unlike the benevolent attitude of allowing for B’s secrets accompanied with trust. A need not trust B because A has a power advantage over B. For A it is prudentially rational not to trust at all. And if A does not trust, she or he cannot allow for B’s secrets. A must come to know them. Thus powerful agents do not allow for the secrets of their subordinates (no trust). Power advantage does not allow for (rational) trust, if we discuss secrets. In a cynical way we may even suggest that the need for trust is a consequence of a lack of power. Hence you trust because that is your only viable alternative. What happens if A’s cost of becoming to know B’s secrets is quite costly and the probability of the secrets being trivial is high? The answer is, if A cannot afford to know B’s secrets A’s social power is lacking. Often this is the case and thus A must trust B. A’s attempt to know B’s secrets is a neverending project. We are discussing here a process whose aim is never completely achieved. Subordinate agents must allow for the secrets of the powerful. God has many secrets of his own. He is a mystery to His followers but His creations have no secrets. Therefore, they trust Him. B has no secrets, unlike A, who possesses them and shares only those secrets he or she wants to share. B as a less powerful agent cannot do anything to change the situation. B must allow for A’s secrets, however threatening they may seem. Such secrets are risky from B’s point of view but the lack of power means inability to change the situation. Does this mean that B trusts A? Obviously the answer is, prima facie, in the negative. We must immediately comment on such a question because it may suggest a potentially implausible answer. Does the fact that B admits that A may have and keep secrets entail that B also trusts A? We need not trust an enemy simply because we cannot unravel her or his secrets, and the validity of

Trust and Risky Secrets

85

this observation is not dependent on the power of the enemy. B may trust A in a weak as-if sense of trust (here is a novel sense of as-if trust) but does B also trust in some stronger sense of the word? B does not trust A in the sense of having a trusting positive emotional attitude toward A. I admit that this is so. But in another sense B trusts A. B knows that the secrets are risky to him or her but cannot do anything to change the situation. In this sense it is useless to resort to mere as-if trust. Such an attitude is inefficient, meaningless, and void. B cannot take many precautions against A simply because of not being allowed any secrets of his or her own. Meaningful as-if trust is not available. Normally the luxury position of as-if trust belongs to the more powerful agent, and here B is the less powerful one. As-if trust entails precautions and checks, but such measures do not apply here. A is not interested in B’s trust and so no display of trust may advance B’s cause. By means of standard as-if trust B is merely self-deceived. But in the new weak sense of as-if trust, B displays as-if trust, that is, a trusting attitude implying a typical negative emotional component which resembles regret. This new type of as-if trust is a psychologically truncated form of full trust but it does not involve the typical precautions and tests which belong to normal as-if trust either. In the typical sense I tell myself that if I cannot trust I want to do something about it; in this new power-relevant case I tell myself that I must trust because I cannot do anything about it. This fact brings about regret. If this is so, B may as well resort to a stronger version of trust than mere weak as-if trust. In fact B benefits from trust, in the sense that if B must allow for A’s secrets and this entails a trust-like social situation, he or she may as well go all the way, that is, accept the logical demands of the situation and trust A. I am not saying that this is B’s conscious decision based on an act of will. B cannot decide to trust. Yet in this context the trusting attitude emerges as a response to the logic of the situation. So, B can do nothing else but trust A and hope that A’s secrets are not dangerous. But if he or she trusts A, such worries may be forgotten. This is a major benefit. Trust implies confidence that the risks are not imminent. It is good to trust. Trust in the context of social power is different from trust in the context of equality, but it is also true that social power influences all social relations. We can call B’s trust on A as a special type of weak as-if trust, but this does not help much. Why not call this a special type of full trust, without adding as-if to it ad hoc? I illustrate these points by discussing authority. I suggest that in the case where B comes to trust A, A must become an authority to B. B trusts A because A is an authority to B. In other words, if A is an authority, B’s trust is rational; otherwise, it is not. If it is not, it is based on self-deception and entails a lack of personal integrity. A. Trust in an Authority (Standard Case) When we discuss authority, the above ideas of power, trust, and secrets start

86

TIMO AIRAKSINEN

looking more plausible. We can see why B trusts A in way a stronger way than any of the as-if senses. An authority in the sense discussed here is an institutional agent whose relevant opinions, rules, or policies replace those of the subordinate agent. The subordinate agent suspends his or her judgment and assumes those of the authority as replacements. (Raz, 1985; Airaksinen, 1988, pp. 93–170.) The agent has initially accepted this authority for different reasons ranging from tradition and faith to rational calculations. These reasons are not relevant to us now. But we should notice that the benevolence of an authority toward subordinates need not be mentioned in the definition. Some authorities are excessively demanding and some are repulsive or hostile. Yet an authority is a position of institutional power whose wielder is somebody to be trusted in a normal sense. I may have a personal reason not to obey an authority, but because it is an authority I suspend my own judgment as required and act accordingly. An element of trust is involved here. For instance, the State wants to draft you to the army; you hate the idea; but you know that you should go because it is for the good of the nation, as the authority (the State) says. To suspend our judgment in favor of that of the authority implies trust. When A tells us to move and we moves because we are told, not because we are threatened or coerced, we suspend our judgment concerning the reasons for moving and follow those of A. This requires trust, and an authority is necessarily trusted; if an agent or agency is not trusted, its directions and commands are coercive and based on threats. If the authority is benevolent to B, that is, if its description mentions benevolence, A has a good motive to obey B. But this is not always so. Benevolence may not be mentioned and thus it cannot form the crucial motive to obey an authority, A. An authoritative command or judgment is motivating simply because it is issued by an authority. Hence, an authority is a special type of power, namely, a power to be trusted; yet it is power and B is in a subordinate position relative to A. Two types of authority must be distinguished: an authority and authorityto. What is called coercive authority is typically authority-to, that is, the police is authorized to use violent threats in order to coerce people. Whether an authority can be genuinely coercive is an open question. Let us then return to secrets. We can say that to be in a position of an authority is to be able to guarantee that subordinate persons have no secrets and that only the authority has them, if it is a rule issued by the authority not to keep secrets. Some philosophers would disagree with this and say that state authority today is transparent (O’Neil, 2002, p. 65). Anyway, authority A states that no secrets are to be kept and the subordinate B has thus a reason not to keep secrets, or has no available reason which would back a decision not to disclose them, or to resist the surveillance of all relevant personal information by A. Therefore, the subordinate person B trusts A, but the authority A does not trust B, or at least need not do so. It is interesting to recall here that to come to know a secret may be

Trust and Risky Secrets

87

dangerous to the recipient of the information. Does this point apply to an authority as well as to an agent who is equal to B? In other words, does the power advantage of A eliminate the dangers of coming to know the secrets of B? One must answer in the affirmative. No information or secrets disclosed by B can harm A, because A is an authority and has the relevant power advantage. The basic point is this. Whatever information B has against A, it is not relevant as B’s judgment will be replaced by that of A. The State tortures its prisoners and B knows this; if State is an authority to B, B accepts the State’s judgment of the treatment of the prisoners. Hence B’s ideas are irrelevant and not dangerous to A. I refer also to my example about the use of surveillance systems below. This forms a crucially important example of secrets and their distribution between parties of unequal power. Nothing we know can hurt an institutional authority and thus they are free to investigate us. This is a depressing intuition of the 1984 type. Think of the omnipresent modern 1984 type institutional surveillance systems. They know all about us citizens. (Lyon, 2001; Whitaker, 1999.) Whatever they know cannot harm them. They are invulnerable. We know something about them, in fact what they want us to know, but not everything. We are supposed to know that they are watching us. If we knew their secrets, it would be dangerous to us. No police or secret service allows the citizens to know too much about them. They have their secrets and to come to know them to means a risk to us. They never trust us, but we need to trust them in the sense that we believe them to be benevolent and interested in the common good. Because these facts are not part of the definition of authority, the trust in their benevolence must come from an independent source. They say they are benevolent, and because they are an authority, we believe, suspending our earlier judgment about these matters. They may be benevolent, that is true, but it does not matter. We trust them anyway. They do not trust us, that is why they survey us and that is why they do not tell us about what they do. They keep their secrets, which we do not want to know. The important point is that we accept the surveillance and think it is for our good. This is trust. We believe that we need and deserve it. We must trust them, otherwise we would rebel. This is the crucial point: Because we do not rebel there is an authority, and we trust. And notice that the definition of an authority is such that it does not make sense of an act of rebellion. We accept authority and suspend our judgments – this is authority and it leaves no room for rebellious feelings and actions, except as a form of entertainment. We do not rebel and we hardly even protest. This entails trust in a quite strong sense of the word. Their institutional secrets are “justified,” ours are not. B. The Demands of Care Another example of trust and secrets in the context of institutional power deals with doctors and their patients. Patients, understood as the recipients of

88

TIMO AIRAKSINEN

care, are allowed by the rules of the institution to keep their secrets. It is their right to do so. They hide their own folk medication when they are in a cancer ward. This is a well-known problem in hospitals. When a patient dies her night desk drawer is full of vitamin pills and health magazine clippings explaining their beneficial effects. The pills are not allowed at the hospital but no one has the right to search the patients’ private realm. This is how it is in Finland, and I suppose in other liberal societies as well. In prisons they do such searches but then this institution is a coercive one. An authority may be allowed to search as well, but the methods and their justification cannot be the same as in prisons. Prisoners are not expected to accept what is done to them, and this means that the searching agent acts as a coercive one, or an agent who is authorized to coerce. In hospitals the search must be licensed by the patient. The doctors are supposed to trust their patients in the sense that they must allow for their secrets. This is institutional trust which may not correspond to the feelings and ideas of an individual doctor in a given situation. Again, this institutional fact means that doctors are bound by the rules and conventions of the medical institution. When we discuss doctors and patients here, we need to keep it in mind that the relationship is one between an institution and a private person. Secrets are sometimes harmful to the patient and thus indirectly to the provider of care whose responsibility it is to guarantee the health and safety of the patient. Such a situation indicates the stronger demands of the autonomy of patients and their diminishing trust in the mainstream medical authority. Medical authority is a vanishing breed which makes more room for patients’ autonomy. (Lantos, 1997.) Patients gather all the information available to them in order to make their own decision as if they were capable of evaluation and judging their own case. They want to make their own decisions but at the same time they are searching for alternative health advisors. This is not as inconsistent as it may sound. The patients pay attention to their autonomy and therefore they want medical advice from new providers whose claim to authority does not expand beyond purely cognitive limits. The patients want good advice and that is all. They trust novel advice but only within quite narrow boundaries. This is an example of care, but in this paper the concept of care must be understood as broadly as possible and independently of an authority. The patients take and leave information as they themselves like. The carers, such as doctors and nurses, are not allowed to conceal their secrets, which they hide from the receivers of care, such as patients. This is one of the main ethical norms of medical care, and of all care for that matter. According to the accepted norms of care, everything which concerns the patient, even indirectly, must be released to her or him, even if she or he would not like to hear it. It is an inalienable right to get all the relevant information. We may also call it immunity to undisclosed information. The truth may be painful to hear, but at the same time patients are afraid of the

Trust and Risky Secrets

89

secrets of the carers, which again implies a lack of trust. Such a fear is not the only reason for insisting on full information but it is part of it. However, if this is so the cared for patients are in a position of power. This is not authority. The patients’ judgments do not replace those of the carers. The carers’ judgments which actually replace those of the patients tend to cover a quite narrow band of medical advice, and obviously even a narrower band of information in the future (see below). The patients have power and thus the carer must trust the cared for persons, but not the other way round. Perhaps an ideal situation of care implies mutual trust, but it is difficult to see why the cared for persons should consent and share their secrets with the carers. One relevant reason may be the scale of distribution of secrets between carers and their patients. If the carers know all of the secrets of the cared for persons, those secrets are automatically distributed far and wide, at least in the imagination of the patients. And the secrets in question tend to be personal in nature. The secrets the patients come to know are, on the contrary, institutional, impersonal, and concern facts and values which are relevant to them and their well being. The patients also want to know something personal about the carers, for instance their use of recreational drugs. But even this information is relevant to their well-being. The nature of available information is not uniform and the demands to know are asymmetric. This is not always so: for instance, parents know more about their children than the other way round. But we can dismiss this example here because we will deal with it in the next section which discusses paternalism. Let us say that the typical distribution of secrets in institutional care presupposes an antipaternalistic social context. Thus we can say that the patients do not trust their carers even if they themselves expect to be trusted and cared for. All this may sound counterintuitive from the point of view of the traditional rhetoric of care, but the world is changing. The difference between authority and care is interesting because the standard view is that, say, doctors are in an authoritative position with respect to their patients. However, this implies that doctors have secrets unlike the patients. This is not the case, and thus we can say that the doctors’ institutional position is not that of an authority. The only sense in which the doctors still are an institutional authority is in a purely epistemic sense. They know more and their revealed medical opinions should replace those of the patients. In this limited sense doctors are still authorities. However, a time existed when doctors were also non-epistemic authorities. Plato talks about acceptable lies, including the doctor’s lie, which is one of the noble lies told in the name of the good of the patient. Such a case implies undisclosed secrets and turn doctors into real authorities and power wielders. Lies and secrets, however noble they are, are no longer possible. The balance of power has shifted from the doctor to the patient, with far reaching consequences concerning the professional ethics of doctors. Doctors do not command anymore, they only advise. A doctor knows how to heal a patient and cannot have any good reason

90

TIMO AIRAKSINEN

to conceal this knowledge. On the other hand, it is the patients’ own private wish to become and stay healthy. No authoritative agent can have a say on the strength, persistence, and rationality of such a wish. No authority extends that far. The need for security, unlike health, allows for an authority. The main reason for this is that security is our common business, when health is private. I am healthy and we are safe. It does not make sense to say that we are healthy and I am safe. Relevant counterexamples are epidemic diseases, like smallpox, because to fight against them requires social measures and thus introduce genuine medical authorities into the context. They may well have their secrets. They need not tell us everything when they are fighting epidemics like AIDS. The relevant institutions tell us only what we need to know about AIDS and campaigns against it. It appears possible to argue against my view above by saying simply that an individual doctor does not trust the patient. The doctor only dutifully respects the patient’s right to privacy and thus we may say that the institution is the locus of trust, not the person. This may well be true. Why this is still inaccurate, depends on the doctor’s institutionally normal benevolent attitude as a care giver toward the patient and her or his condition. If this crucial attitude is missing, the relationship in question is no longer that of care. Only if there is benevolence can we say that the caregiver trusts the recipient of the care when they allow for personal secrets in the context of their co-operation. Such an institution is characterized by its benevolence. If the doctors cannot trust the patients they lack benevolence and miss the fundamental idea of care. The carers think that they and the patients co-operate and thus it would be unreasonable not to trust them. The patients may have secrets. Benevolence and trust are closely connected so that benevolence facilitates trust and creates a necessary condition to something better than mere as-if trust. Yet trust remains one sided. The patients are individual persons whose goals are their own. They may rely on a doctor, but that is all. In sum: Modern doctor-patient relationships must be generalized to apply to all situations of care which contain no successful claims to authority and which are free from coercion. It appears to me that it can be done. Carers are no longer fully trusted, even if they are often relied on, and thus they cannot conceal their secrets; the receivers of care are trusted and they can keep secrets. The explanation for this is that all care is for the good of the receivers. Their personal well-being is all that matters in this institutional context of benevolence. Accordingly they have power, but this power cannot be called authority. The carers need not suspend their judgment because the patients express their own. The carers must recognize the judgments of the patient but this does not change the mind of the carers. It is quite consistent to say: I do as you wish but I do not think your view is justifiable. In the case of an authority this is not possible. Recognized authorities make the subordinate persons change their mind set. The receivers of care can never achieve this in their present role.

Trust and Risky Secrets

91

C. Problems of Paternalism I mean by institutional paternalism measures of benevolent good care independently of and even against the will of the patient (Häyry, 1991; VanDeVeer, 1986). The key point is that paternalistic agents have a reason not to trust the patients. Paternalistic institutions are not trusting agents (Baier, 1986, p. 225f). This is because the patients have a reason to struggle against the paternalistic carers, if we suppose that the patients still want to assert their own will, which is not always the case. For instance the carer is an authority to the patient. But if the patients do not suspend their judgment and conform to an authority, the patients still tend to assert their judgment and will and therefore a conflict ensues. The carers do not recognize the patients’ will anyway. In this case also the patients lose the ability to trust the other agent, however benevolent this actually is. Thus both parties keep their secrets which entails mutual lack of trust. This is because paternalistic care is independent of the will of the patients who do not care of or even understand their best interests, as defined by the paternalistic agent. Therefore the patients who follow their will struggle against the paternalistic agent, and therefore also want to protect their secrets. The agent cannot trust the patients and thus does not want to reveal secrets to the patients. This is a conflict situation which entails mutual distrust. Is our description of paternalism a problem for my present theory of power, trust, and secrets? It may be because the institutional paternalistic agent is a power wielder and the patient a subordinate agent. This fact should entail a new distribution of secrets, which in power contexts is typically asymmetric, but which is now symmetric as both parties deny their opposition access to their secrets. Should we say that a paternalistic context displays only a conflict and therefore we cannot talk about trust at all within its limits? But this sounds strained because the paternalistic agent wants the good of the patient. How can it then be a mere conflict? The source of the apparent conflict is therefore the patient’s unwillingness or incapability to recognize and accept this truth, along with its practical consequences. This means resistance against a would-be authority. An apparent conflict of interest exists. This is true. The paternalistic agents are still in a power position in the sense that their will dominates in the situation. If we accept the idea that power in conflict situations does not allow for trust at all, we can also say that any talk about the distribution of trust is needless. Then it follows that my theory of trust and secrets applies only to consensual applications of power, such as authority. But it is not restricted merely to authority. When patients keep their secrets from their carers, they do not turn into authorities by doing so. They still stay as receivers of care, although they have power. But they are not power wielders because they cannot impose their will on the context of care, except in one strictly limited way: they keep their secrets. Thus, my theory is not limited to authority. But it cannot say much about power in conflict

92

TIMO AIRAKSINEN

situations because trust is always misplaced or superficially as-if in such a context. It appears that the only possible solution is that the paternalistic agent becomes an authority to the patient. Then no conflict exists and the carers can keep their secrets which are needed for paternalistic purposes. Paternalism without authority is a problem, paternalistic authority is not. Often authority and paternalism are mixed. The carers make the patient believe that their domination is justified and the patient must reveal his or her secrets in a one-sided manner. Paternalistic authority is common even in a modern welfare state. Such a mixed mode of power and care re-institutes trust to its realm, as it is easy to see. The patient surrenders her or his person to the care of the paternalistic agents and the agents are able to realize their plans of care without opposition. Now the receiver of paternalistic care trusts the caregivers. Notice that the crucial difference between paternalistic and benevolent authority is that the first acts for the good of the patient regardless of his or her will unlike the second. The paternalistic authority may try to change the will of the patient but would act regardless of such success. An authority is never fully efficient and therefore there is always room for paternalism too. Paternalistic care is often camouflaged as real care, but in this case the patient’s will cannot dominate, if the context is still going to be paternalistic. In the ideal world of care the cared for person is the one to be trusted, yet the paternalistic agents cannot handle such a requirement without losing their status as paternalistic agents. They need their secrets because they are going to override the will and desires of the patient anyway. A fully transparent agent cannot do that. They may pretend strategically that they trust the patient by allowing him or her to keep secrets, and they say they reveal their own. But the real situation must be different, if the context is still paternalistic. For instance, the patient’s secrets cannot be allowed because they may facilitate a struggle against the carer. I conclude that the case of the paternalistic authority is still paternalism. 5. Lost Secrets: A Pro-Surveillance Argument as a Test Case A common argument can be stated as follows: I have nothing to hide, therefore you may put me under surveillance, with my implicit permission. This is to say: I have no secrets, or I have full confidence in you. I trust you even if you did not trust me – why would you otherwise put me under surveillance? I do not even want to know whether you trust me or not because it is totally irrelevant to my decision to trust you. An example of a standard popular argument (Folk Philosophy) follows: Henry Porter, like all rights activists, spouts rubbish. Why is he paranoid about the state holding information about him? What does he have to hide? If he has nothing to hide, then he has nothing to fear. (Williams-

Trust and Risky Secrets

93

Key, 2007.) This shows perfectly how authority and authoritarian attitudes work: Alan, who wrote the letter, trusts the State and this means that he has nothing to fear even if the State possesses his secrets which are highly relevant to his wellbeing. Henry does not trust the State, which means that the State is not an authority to him, but when it is spying instead a coercive agency. Henry wants to keep his secrets from the State. The main point is that the citizens cannot know what information the State holds about them. The State always has its secrets. Perhaps Georg Wilhelm Friedrich Hegel is right when he calls the State God, or the Divine Idea on Earth. Thomas Hobbes calls it a Mortal God. The State has its secrets, and we are invited to trust our State. However, a fully transparent democratic State (and a Utopia or Paradise) would have no secrets. Perhaps it would then no longer be a State, but instead some kind of an anarchistic construction. Yet, we the citizens may still like to keep our secrets, which would lead to care-giving situations in which the State has the role of the provider of care, a kind of super welfare state. This happens in an ideal welfare state, which provides care without surveillance on the basis of full trust. Such a state trusts its citizens and is itself transparent. The citizens, on the contrary may keep their secrets so that the welfare benefits and other types of care are given them on the basis of trust. This does not mean only on the basis of trust but at least partly. The citizens’ privacy is respected. Perhaps this is only a utopia and perhaps also welfare states are based on authority that follows its typical norms of the distribution of trust and secrets. However, the case in which neither the State nor its citizens possess secrets implies no power but, on the contrary, free full trust and cooperation in Utopia. Surprisingly such a Utopia cannot rely on the relations of care. I am grateful to Mr. Pekka A. Mäkelä (University of Helsinki) both for his critical comments on the first draft of this paper, and to Ms. Eija Prossor and Professor Manfred Holler (University of Hamburg) for many informative discussions on secrets.

WORKS CITED Airaksinen, Timo. (1988). Ethics of Coercion and Authority. Pittsburgh, Pennsylvania: Pittsburgh University Press. ———. (2008). “Hobbes on Trust.” In Perspectivas latinoamecanas sobre Hobbes, ed. Maria de Stier. Buenos Aires: Educa, pp. 133–145. Baier, Annette. (1986). “Trust and Antitrust,” Ethics, 96:2 (January), pp. 231–260. Baker, Judith. (1987). “Trust and Rationality,” Pacific Philosophical Quarterly, 68, pp. 1–3.

94

TIMO AIRAKSINEN

Bok, Sissela. (1982). Secrets. New York: Pantheon Gigerenzer, Gerd. (2002). Reckoning with Risk. London: Penguin. Holton, Richard. (1994). “Deciding to Trust, Coming to Believe,” Australasian Journal of Philosophy, 72:1, pp. 63–76. Häyry (Gylling), Heta (1991). The Limits of Medical Paternalism. London: Routledge. Jones, Karen. (1996). “Trust as an Affective Attitude,” Ethics, 107:1 (October), pp. 4– 25. Lantos, John. (1997). Do We Still Need Medical Doctors? New York: Routledge. Lyon, David. (2001). Surveillance Society. Buckingham, England: Open University Press Mansfield, Katherine. (1946). Stories. Cleveland, Ohio: World Publishing. O’Neil, Onora. (2002). A Question of Trust. Cambridge, England: Cambridge University Press. Raz, Joseph. (1985). “Authority and Justification,” Philosophy and Public Affairs, 14:1 (Winter), pp. 3–29 de Stier, Maria, ed. (2008). Perspectivas latinoamecanas sobre Hobbes. Buenos Aires: Educa. VanDeVeer, Donald. (1986). Paternalistic Interventions. Princeton, New Jersey: Princeton University Press. Whitaker, Reg. (1999). End of Privacy. New York: New Press. Williams-Key, Alan. (2007). “Reply,” The Guardian Weekly, 24.8.2007.

Six PUBLIC TRUST Cynthia Townley and Jay L. Garfield When the Dao is lost, it is replaced with “humanity” and “morality.” When such practical wisdom emerges so does great artifice. When the feeling of kinship is lost, it is replaced with “filial piety” and “affection.” When communities and relationships are lost, they are replaced with “loyal ministers.” Daodejing 18 Introduction We often think of trust as an interpersonal relation, and of the distinction between trust and reliance as a distinction between kinds of interpersonal relations. I may trust one colleague but not find her reliable; rely on another but find him untrustworthy; both trust and rely on my best friend; neither trust nor rely on my dean. One of us has discussed the nature of such relations and distinctions at length. But trust is not only an interpersonal matter. Human society comprises not only individuals but also institutions, and individuals who occupy crucial institutional roles. We often individually or collectively trust, distrust, rely on, or fail or refuse to rely on such institutions or institutional players qua players. In this essay we explore the structure, scope, and value of such institutional trust and distrust, which we call “public trust.” Public trust, we will argue, plays a central role in constituting the public sphere and in structuring societies in which it is worth living. We first characterize trust and distinguish it from mere reliance. This terminology is not consistent within the literature: it is possible to discuss trust without drawing a distinction between it and reliance, but instead making one a part of the other. For example, Philip Petit characterizes trust as a form of reliance (Pettit, 1995). Lawrence Becker conversely defines three forms of trust: credulity, or “a disposition to believe what another person says and to banish skeptical thoughts”; reliance, or “a disposition to depend upon other people in some respect”; and security, or “a disposition to have confidence about other people’s motives, to banish suspicious thoughts about them.” (Becker, 1996, pp. 45–46.) Here, all dependence is a form of trust, but the forms of trust he identifies can be independent. Thus, according to Becker, I can believe without feeling secure about a person’s motives, or depending on

96

CYNTHIA TOWNLEY AND JAY L. GARFIELD

them in any obvious way. Conversely, I can depend on someone without understanding what they say, so I am not in a position to believe. Each of these, according to Becker, is a form of trust. However, instead of identifying credulity, reliance and security as different kinds of trust, we retain a distinction between trust and reliance, using dependence as the umbrella term. The powerful normative connotations of terms like “trust” and “trustworthy” appear applicable to a specific range of relationships and instances, so it is preferable to restrict this term. Trust involves forms of engagement that differ from those involved in reliance. We apply a distinction between trust and reliance to institutions, arguing that the relations between individual or social clients and social institutions can be understood in these terms just as can those between individuals. With this account in hand, we turn to the importance of public trust in our collective life, to the appropriate limits of public trust and the circumstances in which reliance might be a preferable relation to public institutions, and finally to the impact of modernity and mass society on public trust and to appropriate responses to this impact. 1. Trust, Reliance and Reactive Attitudes Trust and reliance are distinct forms of dependence that are most familiar as relationships between persons in their private or public roles. Trust enables what Peter F. Strawson (Strawson, 1974) called “reactive attitudes,” emotions that involve moral judgments specific to the action descriptions appropriate to the actions that inspire them. When we trust someone we become open to the possibility of our trust being rewarded or betrayed. In the first case, we may extend further trust; in the second, we experience betrayal. In contrast, our expedient reliance can only be disappointed. Although any kind of dependence makes us vulnerable, not all forms of dependence potentiate the same moral emotions. This is not to say that affective attitudes arise only when trust has been breached. People are often disgruntled, dismayed, frustrated, and disappointed at failures of reliability, and these responses may manifest in strong emotional reactions. They may motivate withdrawal from the relationship. If Mary forgets one more time to put the cat out, despite promising to do so, John may move out, and take the cat. But it would be bizarre if he were to justify ending the relationship on the grounds that in being so unreliable, Mary betrayed him. But the reactive attitudes appropriate to betrayal form a distinct range of moral emotions, which arise when the relationship has been qualitatively undermined, not just contingently disrupted. If, on the other hand, Mary disclosed details of John’s intimate confessions to his coworkers at a party, despite having promised not to do so, a feeling of betrayal might be more appropriate. One party to a relationship might be unreliable because she is

Public Trust

97

overwhelmed by work, or is unwell, or is grieving, such that a change in circumstances might motivate a reversion to and remedy of the dependence relation. Even irremediable unreliability due to cognitive infirmity need not undermine an entire interpersonal relationship. This is not the case with trust breaches. “I am less busy now, so you needn’t worry about my divulging confidences again” does not make sense in the way that “I am not so busy now, so I won’t forget to let the cat out again” does make sense. That trust is moral and reliance pragmatic explains why the goods of trust and the harms of betrayal, unlike the goods and harms pertaining to reliance, are not fungible. A failure of mere reliability can be offset by a compensatory or symmetrical benefit. If the pizza is later than promised, I do not have to pay for it. Mary might even agree to clean the bathroom for a month to make up for the inconvenience she caused by dereliction of feline duties. If trust has been breached, however, it makes no sense to compensate. What could John ask Mary to do to compensate for divulging confidences? What possible use would that be in repairing their relationship? (This does not mean there should be no effort at compensation for harm done, but that this cannot be about fixing or making up for the betrayal per se). The reestablishment of trust amounts to reconstructing a relationship from scratch, and cannot be achieved by providing an opportunity to even the score by doing to the betrayer what she has done to you. A tit for tat strategy does not work for trust. If a relationship has been breached by betrayal, not only will it not be remedied if it is breached in a similar fashion by the other party, but remedying the material harm, even if this were possible, will do nothing to repair the situation. Trust, as Annette Baier (Baier, 1986) and Karen Jones (Jones, 1996; this volume) have emphasized, is not a simple two-place relation between a truster and a trustee. It is always a three-place relation. One person trusts another within a specific domain, or, on some accounts, with a specific good. On the other hand, as Jones argues persuasively, this three term relation is intimately bound up—fostering and depending on—a fundamental character trait, that of being trusting. As Jones argues, and as we will argue below, this is a desirable and essential attribute of a flourishing citizen. Trust and its betrayal, as well as reliance and its disappointment, are not restricted to the private sphere. They have public dimensions as well. We rely on or trust public institutions and holders of public office; we expect to be trusted or to be relied upon by such institutions and officeholders, and we expect such relations to obtain between them. Such characteristically human relations as trust bring humanity to our public life as well as to our private life. In what follows, we choose the kind of institution we know best, the academy, as a case study in order to shed light on the structure and value of public trust and the dangers of betrayal of such trust. Education is a public good enabled by public institutions (or private institutions operating squarely within the public sphere) within which

98

CYNTHIA TOWNLEY AND JAY L. GARFIELD

dependence relations are constituted, including relations of trust and of reliance. We collectively trust teachers to discharge their responsibilities with integrity and rely on students to be honest. Teachers who offer an essay writing service to students betray the public trust in a way that the studentpurchasers, who act corruptly, do not, however much they may betray private trust. This section continues by using the example of a teacher and a learner to illustrate and make precise the distinction between trust and reliance. In the following section, we extend the discussion to trust in and reliance on the institution itself. Consider a teacher, Terry, and her student, Lia. Trust and reliance may be present in such a relationship, and either can be breached. Lia, like her classmates, depends on Terry to be competent, professional and responsible. The students rely on their teacher to keep track of assignments, to show up for class or meetings, to provide suitable course content and feedback and so on. Terry is also in a position of trust insofar she is expected to maintain the conditions for effective teaching and learning, and, for example, not exploit her position for reasons of personal advantage. These expectations are not only those of her students, but also of her colleagues, her institution/employer and from the public at large. Terry is a trustworthy teacher, seeking to uphold the educational values with which she is entrusted. But she might be unreliable. For example, she might misplace Lia’s assignment, or be egregiously late, and disorganized. Lia might become exasperated with Terry’s unreliability, and change classes, thereby ceasing to be reliant on this particular teacher. Nevertheless, she might still properly regard her as trustworthy. (Likewise, Terry may become exasperated with Lia’s lateness, but need not for that reason distrust her.) When Terry has lost an assignment, or been late in returning work, she might seek to make it up somehow, say, by making extra office hours available, or allowing extra time for her students to complete the following assignment. A student’s lateness in submitting the assignment is met with a penalty such as loss of marks. This balancing of costs and benefits is possible because the reliance aspect of the relationship is instrumental and pragmatic— within it, the parties can try to establish other conditions that make up for lack of reliability. The goods of reliability and the harms of unreliability are, as we put it earlier, fungible. Unlike Terry, Arthur is reliable, but untrustworthy. Arthur never loses track of an assignment, and is never late. He does, however, knowingly misappropriate student work, publishing some of the good stuff under his name. He is a plagiarist. He also, periodically, sells essays to students for use in other classes. Losing an essay is less than ideal, but plagiarism or selling assignments to students is not a simple mistake for which a teacher might make amends in ways consistent with the preservation of the teacher-student relationship, or with the maintenance of public trust in the educational system. The very possibility of teaching and learning is undermined by such betrayal.

Public Trust

99

Such acts destroy constitutive elements of a teaching/learning relationship, such as respect for intellectual production, academic integrity and the like. And a teacher’s appropriation of a student’s work is therefore worse than “mere” plagiarism by a student, since particular responsibilities involved in a position of public trust have been violated, as well as the more general responsibility not to steal or cheat. Terry could compensate Lia for losing a paper or for returning one late. If the compensation is accepted, the teacher-student relationship is maintained, and the broader relationship between academy and society is not injured. But Arthur can offer no such compensation. Is he to pay Lia for the labor involved in writing the paper of hers he plagiarized? Is he to offer her some of his work for her to plagiarize? And how can he compensate the students to whom he sold papers for undermining their education and the colleagues he deceived? Refund the students money? Re-grade new papers? Manifestly, while any of these actions might redress some of the costs of his betrayal, the massive and pervasive damage done to the relationships of trust that under gird academia and his place in it are irreversible. His career is (should be) over, and the damage he has done is irreparable. Trust and its goods are not fungible in the way that reliability and its goods are. If trust is constituted and maintained by a commitment, when it is breached, the deal is off, whether distrust or betrayal has broken down the dependence. Personal and public relationships can survive failures of reliability but not failures of trust. This is because such failures undermine the relationship itself; they go to its heart. There is a further, and perhaps more catastrophic, consequence of the regular or significant betrayal of trust in a domain central to public or private life, and that is the diminution of the degree to which an individual is trusting (We are indebted to Karen Jones for this point). Let us first ask why it is good to be trusting (keeping in mind, that like any virtue, being trusting comes with a pair of correlative vices—one of deficiency, to which we refer in this discussion, and one of excess, gullibility, which we leave aside for present purposes). Our lives as moral and epistemic agents demand that we in general possess the virtue of being trusting. Knowledge is a collective good. Its production, transmission, storage, and retrieval require that we can trust our sources, our collaborators, and the social institutions that mediate these activities. In order to collaborate effectively, this trust must be implicit. Without it, we would be solo operators, and would hardly be able to know anything; nor could we contribute effectively to the extension of knowledge. Similarly, much of our morally and politically significant action is joint or collective action that requires our pervasive and implicit trust in our fellows. Even such mundane acts as contributing to charities or voting demands considerable trust. The betrayal of trust by another in significant domain can damage this virtue, rendering us untrusting. This can thus make us less effective epistemic

100

CYNTHIA TOWNLEY AND JAY L. GARFIELD

and moral agents, and less effective citizens. Being less trusting makes us more individually fragile, and less capable of collective action. Failures of trust can have significantly deleterious effects on the quality of our collective social and political life. Conversely, the trustworthiness of our fellows and of our social institutions can engender the virtue of trust and can enhance our collective social and political life. Integrity and mutual commitment to one another’s good, individually and socially, make no small difference. 2. Institutions and Trust Above we showed that trust and reliance are distinct forms of dependence that can coexist between the same parties, here, we will show that patterns of reliance and trust between an institution and the public it serves are analogous to those between parties understood in terms of their social, professional or institutional roles. We have made a start by considering trust relations that are constituted in part by public roles, those of teacher and student. But we can go much further. The community depends on universities to educate some of its members and to contribute to a culture of intellectual excellence by teaching and research. Here too reliance and trust are distinct forms of dependence, not just between members of the university community such as a teacher and her student, but also between the institution and its direct stakeholders such as employees and students, and between the institution and the community at large. A university could fail in terms of reliability or in terms of trust. Take reliability first. Suppose Sandstone University fails to maintain the infrastructure necessary for its computing services, and as a result a server crashes, deleting students’ grades from the system. This is a serious failure of the university, and the public, the students, and the staff should expect the reasons for the failure to be identified and remedied without delay. Perhaps the reason is financial, the institution is struggling to meet its costs, and cut just a few too many corners in the technical and maintenance sections. Perhaps there was a simple mistake in operating procedures. This would be a significant mistake, but not necessarily one that indicates a bad or improper attitude to the enterprise of education, instead a lack of competence in pursuing that enterprise. Sandstone might be perceived as unreliable, but not as having betrayed the public trust, or that of the affected students. Disappointment is reasonable; feelings of betrayal are not. Vanstone University is coping with a similar financial squeeze, but instead of reducing costs, seeks to extend its sources of income. Identifying the marketable skills of its Philosophy faculty as a useful resource, Vanstone University contracts with the Australian Defense Force to have its philosophers produce the technical and operational manuals for different pieces of equipment. (An alternative scheme would contract with a college admissions consultancy service to have its philosophers coach prospective

Public Trust

101

students and prepare applications.) They are therefore relieved from all teaching responsibilities and research obligations. Philosophy classes and research will be suspended at Vanstone until the work is completed. Such lucrative consultancy work will fill the university’s budgetary gap for the term of the contract. This arrangement is manifestly inconsistent with the goals proper to a University whose mission is teaching and research, contribution to intellectual culture. The purpose even of the philosophy faculty of a public university is not the direct material support of the military, nor to provide college admissions services. When it acts in this way, Vanstone University betrays the public trust. This is because it has acted contrary to its entrusted purpose by abandoning its commitment to the cultural good of the nation in favor of commodifying the resources at its disposal. Even if the institutions so supported might be worthy of public support and needy, and while the University might well be in need of additional revenue, selling the university’s expertise is not the way to meet either of these needs. This case is like that of Arthur’s misappropriation of his student’s work and like that of his sale of student papers. He is in a position of trust and of power, which he exploited in order to meet his financial needs and/or the needs of his students to pass classes. Vanstone’s governing council likewise is entrusted with making decisions in the public interest. While their trust includes the financial position of the University, it is impermissible to pursue this end by violating the larger responsibility with which the public at large has entrusted it. The university cannot compensate the Philosophy Department by ensuring that it benefits financially from the contract deal. They have betrayed the academic integrity of these academics by treating them as guns for hire and by disregarding the intrinsic value of their philosophical projects, treating them as employees with “billable hours” to be deployed to the university’s best financial advantage. Once this betrayal has occurred, while compensation for damages might be considered, the relation of trust necessary to the flourishing of an academic community has been destroyed. Repair would amount to reconstruction from scratch of a new university community. Nor can Vanstone compensate the public by employing more philosophers to contribute to intellectual excellence—these moves do not fix the problem, which is that the nation once trusted the university faculty to serve educational goals and can now no longer trust that this institution is committed to this purpose. Simon Clarke argues that several attributes of government action enhance trust in government. He identifies sincerity, significance, the absence of alternatives, considered judgment, publicity, competence, locality, transference of trust, reciprocity of trust and encapsulated interest. Clarke argues that to the degree that these conditions are present in government action, government is likely to be trusted by its citizens in that action domain. Using this framework, we can say that Vanstone University’s action fails the

102

CYNTHIA TOWNLEY AND JAY L. GARFIELD

conditions of sincerity, absence of alternatives, reciprocity of trust and encapsulated interest. (Clarke, this volume.) 3. Why Public Trust is Important Public trust is important because in any culture as articulated and complex as ours we must entrust some things to institutions that transcend the individuals who might at any time be occupied in advancing their purposes. Educational institutions are indispensable for our way of life, as are city councils, the stock market, banks, benevolent societies, unions, and other such bodies that serve the public. Our dependence makes us vulnerable, to unreliability and untrustworthiness, but we are better off with that vulnerability and those institutions than we would be without them. We do not trade material or cultural benefits for vulnerability: our relations to institutions often mediate our relations to our societies and compatriots and themselves knit communities that are too large to be bound directly by interpersonal ties. Unless institutions (and the people who make them up) are committed to values that preserve education and intellectual excellence, creativity and artistic excellence, the cultivation of excellent sports people, the preservation of historical artifacts, and so on, such things will be lost, derailed or deformed and our society and culture loses much that gives collective life its point. Public institutions that are the repositories and beneficiaries of public trust make tacit or explicit commitments to uphold the requisite values. Most of these institutions—museums, art galleries, and the like—extend over generations, hence must enculturate members of their direct communities and the public in general to continue to preserve the values. Many such values cannot thrive without such interdependent institutions and without these institutions enjoying public trust. This entails that these institutions are proper objects of moral reactive attitudes as well as parties to contracts, with attendant obligations for compensation when they fail to perform reliably in their purely instrumental roles. These institutions are responsible not only for reproducing and maintaining skills and artifacts, but also for collective moral education—for showing us what matters and why, and for ensuring that what matters to us thrives. Trust is unavoidable not only because to verify that public institutions and their officeholders are meeting their commitments would be too time consuming, but also because to engage in such monitoring would be to invoke relations of suspicion that are poisonous to the social order, just as such constant monitoring of children, friends, parents or life partners would poison those relationships. Moreover, even when we find it necessary and desirable to establish accountability mechanisms to scaffold the reliability of public institutions, trust in those institutions must already be presupposed in order for them to perform this first-order function. Trust must hence provide the context and the point for such social enterprises, making it possible even for those on

Public Trust

103

which we merely rely to perform their functions in a healthy society. Overinstrumentalizing these public institutions can deflect attention from their central roles in forging community, and can in turn lead to the failure to sustain the relationships of trust and trustworthiness on which these institutions and the public life they enable depend. It is important that both members of the general public and of the institution take seriously the preservation of trust and what it protects. 4. The Limits of Trust Trust may be a good, but it is possible to trust too much. In particular, it is possible both to be too trusting and to expect trust to be a basis of too many social relations. Some institutions, like some people, are not trustworthy, even when they may be perfectly reliable. For others, trust may be too much to ask. Our relations with such institutions are too businesslike, or too unimportant to our daily life for trust to matter. Some institutions we may expect to be trusting; others we may wish to eschew trust in favor of measured suspicion, in order to maintain their reliability. We now turn to an exploration of these limits of trust. There are public institutions of different kinds to which the public as a whole or individual members of a society bear quite different kinds of relations. Let us remain with our academic examples for a moment and contrast two quite different kinds of tertiary institutions, a small private college and a large state university. Sophist College is highly selective, enjoys a small student-faculty ratio, and a loyal alumni body. Its students, alumni and teachers regard themselves as bound by institutional loyalties and membership relations to a common, temporally extended and valued institution. Generic University is relatively unselective, maintains a high student-faculty ratio, and is regarded by its students as an education service-provider and by its teachers as a workplace. Its alumni regard it with as much loyalty as they might a bus they once rode. As different as these institutions are from one another in many respects, they are nonetheless similar in many other respects. In each case, studentteacher relations depend on a particular framework of trust. In each case, the larger social community in which the institution is constituted trusts the institution with certain social and cultural goods. In each case, the institution itself and its employees, students, and others involved in its operation must be reliable in several respects. Let us first return to the two examples of domains in which questions about trust might be raised that we considered earlier, namely integrity in scholarly/student work and in research reporting. We can then turn to some more general considerations of the boundaries between public trust and reliance. When a student enters Sophist College, she participates in an orientation

104

CYNTHIA TOWNLEY AND JAY L. GARFIELD

program in which the principles of academic integrity are explained, including the definition of plagiarism and the reasons that it constitutes such a serious betrayal of academic trust. While it is true that many students might, given the ambiguities of ethical principles regarding the use of publicly available information in the current internet environment, arrive at Sophist with an incomplete understanding of these issues, within a few hours of discussion they achieve a pretty clear understanding of the bounds of legitimate use of source material and of the reasons for these boundaries. Once the academic year begins, students are expected to act with integrity. To be sure, a few do not. Some, but not all, of these, are caught and punished. But there is no regular investigative procedure in place to which all student work is subjected. In short, there is a default trust in students’ integrity. When a student enters Generic University, on the other hand, he is handed a code of student conduct, part of which informs him of the bounds of legitimate use of source material, and part of which informs him, correctly, that all academic work must be turned in through software that checks electronically for plagiarism. At Generic, students can be relied upon not to cheat, but there is no default trust in play. At the end of each year at Sophist College, each member of the faculty reports on his research and teaching activity. These reports constitute part of the basis on which decisions regarding merit pay, promotion, and academic leave applications, etc are made. No documentation is expected regarding statements made in these reports. The institution trusts in the integrity of the individual faculty member. It is possible that some faculty members misstate the enrolments in their classes, or even that some claim credit for nonexistent publications. But nobody checks. At the end of each year at Generic University, each faculty member reports on her research and teaching activity. Each publication claimed must be accompanied by a photocopy of the original work, and evidence of peer review; each enrolment report must be accompanied by documentation from the university registrar. There are hence quite few, if any, misreports. At Generic, faculty members can be relied upon not to cheat, but there is no default trust in play. This might sound like the contrast between good and evil in academic life, but it is not. It is not even the contrast between institutions grounded in trust and those grounded in reliance. Let us take the second point first. Sophist College does not trust its students and faculty in every domain: library books are checked out, and security devices ensure that people can be relied upon not to remove materials from the library without checking them out. Faculty members must submit receipts for reimbursement for travel expenses. Generic University trusts its faculty members to teach conscientiously and its teachers typically trust their students’ honesty when discussing academic matters. More broadly, the accounts of Sophist College are audited, and donors and government agencies that are involved in funding many of its activities

Public Trust

105

rely upon the College and its officers not to misappropriate their funds. On the other hand, nobody audits the academic records: a student’s transcript is trusted as evidence of her attainment. Generic University is trusted by its community to educate the youth, support industrial development and generally to contribute to the progress of culture. Life with respect to public institutions, like interpersonal human life, is characterized by a patchwork of trust and reliance, in combination with countless other relations. Moreover, even the contexts in which we might find an absence of trust in the case of Generic University we find a background of trust. A faculty member from Sophist, for instance, who leaves for a position at Generic, might at first find it unpleasant and even insulting to be asked for documentation of her scholarly activity. Her colleagues are surprised. They point out that equity with other institutions in the state, who cannot be trusted to report accurately, demands such reporting, and they point out that they trust the government audit body who distributes funds in accordance with these research reports. The new hire from Sophist might express surprise that anyone can trust the government. Her new colleagues might be equally surprised that anyone would trust students not to plagiarize or faculty members not to inflate their research performance. The lecturer who leaves Generic for Sophist might be equally surprised: How do we ensure that students do not cheat? What keeps me from claiming work I never published? Her new colleagues express shock. What would the point be of an educational community if we did not trust each other? The new lecturer adverts to other goods—the production of skilled students; equitable distribution of research money, etc—that depend on confidence in equity that would erode absent accountability guaranteed by verification mechanisms in which all have confidence. All of this merely shows that, although we require a network of trusting relationships in order to flourish, and that these relationships extend well into the public domain, we do not require that all of our relationships, private or public, must be relationships of trust. It also suggests that there are no general rules governing where each is appropriate. A great deal might vary with culture and context. There are goods available to members of the Sophist community that are not available to those at Generic, but a relationship to the broader community at Generic that is absent at Sophist. Sophist faculty members risk betrayal by colleagues or by students; at Generic, that is hardly a possibility. Trust is necessary in order to sustain moral accountability, moral development, and a sense of community, and these are important human goods. When public trust is violated, community suffers. But community also suffers when trust is not violated, but not presupposed. This is because building and maintaining community involves trust, not only sets of relationships in which the parties are not interchangeable like interpersonal trust, but also institutional relationships within which the goods pursued and

106

CYNTHIA TOWNLEY AND JAY L. GARFIELD

protected are not fungible. Intellectual integrity is not preserved or ensured by checks and balances unless educators and researchers care about it. If educators judge that the likely risks and costs of selling assignments to students and appropriating their work mean that complying with the rules is the best strategy we no longer inhabit a virtuous educational community. There is a further lesson here. We demand accountability from public institutions and from public officeholders. Sometimes this accountability is obtained from conventions of trust; but where trust is impossible, mechanisms to assure reliability must be in place. The mechanisms in turn must be grounded in further networks of trust if our society is not to degenerate into a complex of coercive practices. There is always room for tradeoff, but never room for wholesale abandonment of trust if we are to have a society in which there is anything worth making reliable. 5. Modernity, Community, and Trust Modernity is often regarded in Heideggerian (or Simonic) terms as constituting the inexorable march of technology, or, in Habermassian language, as the progressive colonization of the lifeworld by instrumental reason. The human lifeworld comprises not only individuals, but also institutions. The public sphere in which we educate, are educated, debate, govern, and transact our common life is integral to our lifeworld. Characteristically human relations between human beings, between social institutions, and between human beings and their public institutions comprise trusting relations. It is possible in principle to construct a public sphere in which trust is absent, a public sphere in which the only ground for any confidence in institutions of government, of education, of finance, and so forth is enforcement that guarantees reliability. Totalitarian and bureaucratic regimes exemplify this possibility. Such a society, however, undermines the cultivation of the virtues and attitudes we value in each other and in our public institutions as well. Undermining these virtues makes life less tolerable and less efficient even by the standards of a totalitarian society. Our collective moral and aesthetic goods are diminished, as is our effectiveness in political and epistemic activity. Nobody voluntarily inhabits such a society. In such a society not only does the Tao of humanity degenerate, but so does benevolence, virtue, and even duty. In short, community degenerates, and it becomes hard to see what is worth preserving, or why we should care even about the reliability of any of our institutions. Thanks to Simon Clarke, Karen Jones, Neil Levy and other members of the Macquarie University workshop on trust, 2005 for helpful critical comments on an earlier draft and for many ideas incorporated into the current version of this paper. Without such trustworthy colleagues intellectual life would be

Public Trust

107

impossible. We also gratefully acknowledge a visiting scholar’s grant from Macquarie University that made this collaboration and the workshop on trust possible.

WORKS CITED Baier, Annette. (1986). “Trust and Antitrust,” Ethics, 96:2 (January), pp. 231–260. Jones, Karen. (1996). “Trust as an Affective Attitude,” Ethics, 107:1 (October), pp. 4– 25. Becker, Lawrence C. (1996). “Trust as Noncognitive Security about Motives,” Ethics, 107:1 (October), pp. 45–46. Pettit, Philip. (1995). “The Cunning of Trust,” Philosophy and Public Affairs, 24:3 (Summer), pp. 202–225. Strawson, Peter F. (1974). Freedom and Resentment. London: Methuen.

Seven CAN YOU BUY TRUSTWORTHINESS? Vittorio Pelligra Trust[worthiness] is an important lubricant of a social system (. . .) Unfortunately this is not a commodity which can be bought very easily. If you have to buy it, you already have some doubts about what you’ve bought (Arrow, 1974, pp. 23–24). Introduction A socially responsible organization ought to be responsible first to its “internal society” by promoting, for instance, the conditions for fulfilling human relationships among its members. If, as Matthew Rabin put it few years ago: “Economics should be concerned not only with the efficient allocation of material goods, but also with designing institutions such that people are happy about the way they interact with others” (Rabin, 1993, p. 1283), this appears to be also a concern for economic theorizing. There is nowadays a wide consensus that one of the principal determinants of both individual well-being (Helliwel, 2002) and organizational success (Kramer & Tyler, 1996) is the existence of relationships based on trust and trustworthiness. Standard economic models, however, depict agents as individualist, outcome-oriented and materially self-interested, and as a consequence, show serious limitations when confronted with the phenomenon of trust. This descriptive inadequacy is not neutral. As noted by Robert Gibbons, there is a risk that “management practices based on [incomplete] economic models may dampen (or even destroy) non-economic realities such as intrinsic motivations and social relations” (Gibbons, 1998, p. 130). Economists generally maintain that reducing the cost of a certain behavior leads to an increase in the frequency of that behavior. The relationship between extrinsic incentive and effort is supposed to be always monotone: effort increases with positive incentives and decreases with negative incentives (as long as income effects are smaller than substitution effects.) Psychologists however suggest that material rewards and sanctions may cause conflict between intrinsic and extrinsic motivations that may even yield counterproductive outcomes especially in activities that people generally undertake for their own sake, instead than as a means to some other end. Trustworthy behaviors imply always a material (opportunity) cost for the

110

VITTORIO PELLIGRA

trustee. How might the reduction of such a cost or the introduction of a material reward affect people’s choices in a cooperative interaction? In this paper I present an experiment designed to investigate the effects of material rewards on agents’ trustworthiness. The results suggest that paying for trustworthiness can have a counterproductive effect, as it increases opportunism. This result has profound implications for human resource policies. A socially oriented organization, or a “good company,” should take this evidence into account by, for instance, designing work environments, relational schemas, compensation policies and other incentive provision systems, in a way that respect agents’ intrinsic or vocational motivations as well as their psychological, relational and moral ones. 1. The Birth of a Market Suppose you are a manager in a day-care center where some parents are systematically late at picking up their children. You need to reduce the number of latecomers. Deterrence and economic theory suggest that imposing a fine for each delay would be an appropriate remedy. In an influential study Uri Gneezy and Aldo Rustichini (Gneezy and Rustichini, 2000) analyzed this situation in a natural experiment considering real day-care centers in the Israeli city of Haifa. They considered ten centers for a period of twenty weeks, implemented the fine in six of them, and compared the results with those of the remaining four centers that formed the control group. In the first four weeks the number of parents who arrived late each week was simply recorded. At the beginning of the fifth week, a fine was introduced for a delay of ten minutes or more. At the beginning of the seventeenth week, the fine was removed with no explanation and the behavior of parents was recorded for three more weeks. Figure 1 presents the results. The data show that in the daycare centers where the fine is adopted in week five, the number of latecomers significantly increases. In the centers with no fine, it remains constant. Furthermore, after week seventeen, when the fine is removed, the number of delays does not decrease. The negative effect of the fine appears to persist, at least for the remaining three weeks of the study. What happened? A new market was born where parents exchange minutes of delay for money. They can now buy overtime.

Can You Buy Trustworthiness?

111

Figure 1. Parents’ behavior across time (source: Gneezy and Rustichini, 2000)

2. Trust, Trustworthiness, and Incentive The choice problem considered by Gneezy and Rustichini has the form of a parametric choice where the only relevant aspects seem to be the implicit cost of punctuality and the explicit fine for lateness. However the situation can be better understood, as the authors seem to suggest, if we consider it as a strategic situation where two players, the management and the parent, interact in a sequential game (figure 2). The management has to decide whether to open the day-care or not. If it opens, it may impose the fine or not. The parent has to decide whether or not to arrive late. The payoff is identical in both situations, apart from the fine that increases management’s payoff and decreases the parent’s payoff by two, in case of lateness. If we frame the situation in this way, it appears that the relationship between management and parent has the form of a fiduciary interaction when there is no fine, and takes the form of a market exchange when the fine is introduced. With no fine, parents decide to be late, while with the fine parents should be forced to be on time. The management, by backward induction, decides to introduce the fine. The equilibrium of the game is given by the pair of strategies (fine) for the Management and (on time) for the parent. However, as Gneezy and Rustichini’s study shows, the real effect is much more ambiguous than that. It becomes, therefore, important to analyze the way market exchange and fiduciary relationships interact. And in particular to examine whether trust and trustworthiness are reinforced or crowded-out by material incentives. This is precisely the question that the experiment we present in this chapter will

VITTORIO PELLIGRA

112 address.

Figure 2. The Management-Parent Game

A fiduciary interaction, like that in the subgame with no fine, can be formally described by a trust game depicted in figure 3.

Figure 3. The Generic Trust Game A

B (c,f)

(a,d)

(b,e)

c>a>b; e>f.

Can You Buy Trustworthiness?

113

This game incorporates all the relevant features of a fiduciary interaction: conditional gain; conditional loss; and temptation (or risk of opportunism). Conditional gain is described by the condition c>a. If B behaves trustworthily (choosing “across”) the trustful course of action leads A to a better result compared to the status quo (a,d). Conditional loss is captured by assuming bf. Finally, since e>f, B is subject to an opportunistic temptation. The Nash equilibrium of this game is given by both player A and B playing “down.” Such an outcome is worse than a situation when player A trusts B playing “across” and player B respond trustworthily choosing “across” as well. A large amount of data have been gathered in many controlled experiments that show that people choose this “irrational” combination of strategies much more often than the theory would suggest. The explanations for such findings can be many: altruism, inequity aversion, reciprocity, trustresponsiveness. Elsewhere I have discussed the relative merits of these alternative theories (Pelligra, 2006). From the analysis of the available evidence, it emerges that the principles of reciprocity of trust-responsiveness seem to account for trust and trustworthiness in their pristine form (Bacharach, Guerra & Zizzo, 2004; Guerra & Zizzo, 2004; Pelligra, 2005; 2006a). Reciprocity, however, can explain trustworthiness only in a limited subset of all the trust interactions (Pelligra, 2006b). In this paper I investigate a more limited question: if and how the willingness to behave trustworthily is affected by the introduction of material incentive. We can think of the difference between what the trustee gets in a cooperative outcome (f, see figure 3) and what she or he gets in the equilibrium (d), as the reward for trustworthiness. Different variants of the trust game can provide different levels of reward for trustworthiness. 3. The Experiment The experiment compares behavioral responses in four variants of the trust game, characterized by increasing levels of reward associated with B’s trustworthiness: two treatments present positive rewards (basic trust game, BTG); in another treatment the reward is kept equal to zero (gratuitous trust game, GTG) and in a fourth treatment B’s trustworthiness is associated to a cost (coercive or costly trust game, CTG) (see table 1).

VITTORIO PELLIGRA

114

Figure 4. The experimental Trust Game A

B (10,10)

(5,d)

(0,15)

Table 1. Variants of trust games d’s value

variant of TG

B’s gain from Trustworthiness

5

BTG

5

9

BTG

1

10

GTG

0

11

CTG

-1

I consider the observational implications of the standard theory, reciprocity theory and the trust-responsiveness hypothesis. Standard theory represents a benchmark. Reciprocity theory and the trust-responsiveness hypothesis are theoretically and empirically better at explaining trusting behaviors. The standard theory based on self-interest predicts that B chooses “down” and (if it is also part of the theory that A believes that B is self-interested) that A chooses “down.” According to reciprocity theory (Rabin, 1993) one should expect to observe B players to play trustworthily as long as the difference between what they get from being trustworthy (f) and what they get from playing Nash (d) is positive. In the experimental trust game depicted in figure 4 that is true whenever d