I argue (in chapter 3) that certain alternative, strongly asserted individual-level accounts are implausible as general accounts of trust. They might fit some of the apparent trusting relations we see, but they do not fit many trusting relations.

XREF: Connects to broader philosophical debates about whether micro-level explanations can account for macro-level phenomena like social trust.

Trust is a three-part relation that is grounded in the truster’s assessment of the intentions of the trusted with respect to some action. Typically, the intentions of the trusted will be based in self-interest (as in the account of encapsulated interest), moral commitment, or idiosyncratic character. Other views, that trust is an attitude of ungrounded faith or belief or that it is inherently moral, are not convincing as general accounts of trust.

DEFINE: Defines trust as a three-part relational assessment grounded in the truster's evaluation of the trusted's intentions, opposing views that see trust as ungrounded faith or inherently moral.

One reader challenged me with the Yiddish quip that "'for example' is not proof." In fact, of course, it can be proof—of existence. If you have seen one example of the black swans in Taiwan, you cannot any longer truthfully say that all swans are white. At most, my examples from fictional accounts establish conceptual possibilities, which is a bit short of proving existence. It is in this spirit only that I invoke them. I do so because I think it is often important to reify any conceptual claim by giving an example of it.

Dmitry Karamazov says that the lieutenant colonel implicitly trusted Trifonov. After his sad day of reckoning, the lieutenant colonel would presumably have said that Trifonov was not trustworthy. Unfortunately, Trifonov was trustworthy just so long as there was some longer-run incentive for him to be reliable in their mutually beneficial relationship. The moment there ceased to be any expectation of further gains from his relationship with the lieutenant colonel, Trifonov had no incentive to be trustworthy in this highly irregular commercial dealing between de facto crooks. Not surprisingly, he ceased to be trustworthy.

XREF: Connects to game theory's shadow of the future — trust in repeated interactions collapses when the relationship ends.

The encapsulated-interest conception fits three distinct categories of interaction: interactions that fit the model of the iterated oneway trust game, iterated exchange interactions as modeled by the prisoner's dilemma, and interactions in thick relationships. These categories of trust relations represent increasing complexity of the interactions. In all of these interactions, trust is relational. That is to say, my trust of you depends on our relationship, either directly through our own ongoing interaction or indirectly through intermediaries and reputational effects. If we have no or only a passing relationship, we are not in a trusting relationship.

That is, I trust you because your interest encapsulates mine, which is to say that you have an interest in fulfilling my trust. It is this fact that makes my trust more than merely expectations about your behavior. Any expectations I have are grounded in an understanding (perhaps mistaken) of your interests specifically with respect to me. … More generally, it is principally those with whom we have ongoing relationships that we trust. In addition, the richer an ongoing relationship and the more valuable it is to us, the more trusting and trustworthy we are likely to be in that relationship.

There might even be relationships that I value in themselves and not primarily because they are causally connected to certain other benefits that I get from them. For example, I love certain people and have rich friendships with others. Such relationships are at one extreme of the range of encapsulated interest, and the relationship between Trifonov and the lieutenant colonel is at the other, minimal, extreme. Many of our relationships with others develop from relatively minor exchange interactions and become much richer relationships of fairly broad reciprocity.

Note that our merely having the same interests with respect to some matter does not meet the condition of trust as encapsulated interest, although it can often give me reason to expect you to do what I would want you to do or what would serve my interests (because it simultaneously serves yours). The encapsulated-interest account does entail that the truster and the trusted have compatible interests over at least some matters, but such incentive compatibility, while necessary, is not sufficient for that account, which further requires that the trusted values the continuation of the relationship with the truster and has compatible interests at least in part for this reason. Other drivers on a highway and I enjoy incentive compatibility, and therefore each of us can be expected to try to drive on the appropriate side of the road to avoid accidents with one another. Generally, however, there is no sense in which the other drivers want me to be in the relationship of driving on the same road with them, and therefore I am not in a relationship of trust as encapsulated interest with them.

Continuity of the relationship is not enough, of course, because the commitments matter. In a favorite philosopher's example, Immanuel Kant's neighbors may have relied on his punctuality in his morning walk to set their own schedules. To trust him, however, would require more: that they rely on his having their interests at heart in deciding when to take his walk. If they could not think he did, they could not be said to trust him in the strong sense of the encapsulated-interest account (Baier 1986, 234).

DEFINE: Clarifies the 'encapsulated-interest' account of trust, distinguishing it from mere reliance on regularity.

A rational analysis of trust of another intentional being, as opposed to "trust" of a force of nature (our "trust" that it will not rain on a July day in Palo Alto), must take account of the rationality of both intentional parties. Indeed, in a trust relationship, I must think strategically, because my purposes are served by the interaction between what I do and what another does (or others do). My outcome is the joint outcome of both our actions. In this sense, mere expectation accounts are only half strategic, and they therefore fail to address the central nature of trust relationships. They have a liability not unlike that of the similarly half-strategic Cournot theory of market behavior. In the Cournot theory, actors assume regularity of behavior on the part of others in the market in order better to decide how to act themselves; but, although they are strategic in responding to others' actions, they suppose that others are not strategic in responding to them. Hence they fail to take account of secondorder effects of others' responses to their actions. Cournot actors are somewhat smart but they think others are dumb.

First, trust is generally a three-part relation that restricts any claim of trust to particular parties and to particular matters. Second, trust is a cognitive notion, in the family of such notions as knowledge, belief, and the kind of judgment that might be called assessment. All of these are cognitive in that they are grounded in some sense of what is true. These cognitive notions—and trust, in particular—are not a matter of choosing: we do not choose what is to count as true, rather we discover it or are somehow convinced of it. Hence we do not trust in order to accomplish anything, although our trust might encourage us to enter beneficial interactions. (We may well choose to be trustworthy in order to encourage others to cooperate with us.) Generally, we wish to explain cooperation or its failure by reference to trust. To make the cooperation itself a matter of trust would make the thing we want to explain the explanation of it. Thus we wish to keep trusting and acting from trust cleanly separated. Finally, acting on trust typically involves risk.

XREF: Contrasts with game-theoretic and rational-choice accounts of trust that model it as a strategic choice, the common framework in economics and political science.

A characteristic of trusting relationships, one that is not uniquely relevant to the encapsulated-interest view, is that trust is generally a three-part relation: A trusts B to do X (Baier 1986; Luhmann 1980, 27).

DEFINE: The three-part structure of trust (A trusts B to do X) is a definitional clarification central to the author's conceptual framing.

But here note that this category has two odd features. First, it sounds more nearly like a simple expectations account than a richer trust account. In this account, I supposedly think everyone is reliable up to some degree independently of who they are or what relationship I have with them. I think this of them the way I might think the typical person would behave in certain ways in various contexts.

DEFINE: Distinguishes a thin expectations account from a richer relational trust account.

Second, when survey respondents say they trust most people most of the time, this is almost surely an elliptical claim. They do not mean that, if a random stranger on the street were to ask for a loan of, say, a hundred dollars, they would trust that person to repay and would therefore make the loan. This ellipsis might be covered by the phrase “most of the time.” Hence even this open-ended answer to a badly framed, vague question is almost certainly just a loose way of saying they would trust most people within somewhat narrow limits. Moreover, it is also elliptical in its reference to “most people.” Few of the respondents would genuinely trust just anyone much at all.

Trust is in the cognitive category with knowledge and belief. To say I trust you in some way is to say nothing more than that I know or believe certain things about you—generally things about your incentives or other reasons to live up to my trust, to be trustworthy to me. My assessment of your trustworthiness in a particular context is simply my trust of you. The declarations “I believe you are trustworthy” and “I trust you” are equivalent. If it is cognitive, it follows that trust is not purposive (Baier 1986, 235). I do not trust you in order to gain from interacting with you. Rather, because I do trust you, I can expect to gain from interacting with you if a relevant opportunity arises. Moreover, if trust is cognitive, it is not behavioral. I may act from my trust, and my action may give evidence of my trust, but my action is not itself the trust, although it may be compelling evidence of my trust. If I trust you, I trust you right now and not only in some moment in which I act on my trust by taking a risk on you.

XREF: Contrasts with decision-theoretic accounts of trust as calculated risk-taking, and with behavioral economics models where trust is inferred from costly actions.

I can, however, also choose to take the risk of cooperating with you on some matter even if I do not trust you. While he was the prime minister of Israel, Ehud Barak, when asked if he trusted Yassir Arafat, said, “I don’t know what it means to trust. He is the Palestinian leader, not the Israeli leader, and he is determined to do whatever he can to achieve Palestinian objectives. The real question is not whether we trust him. The question is whether there is a potential agreement that could be better overall for both sides, a win-win, not a zero-sum game” (quoted in Goldberg 2001, 66). Cooperation or coordination is the general goal, but there are many ways to achieve it, some of which do not depend on trust. Hence my actions are not simply determined by the degree of my trust, although they are often likely to be influenced by my trust or distrust.

XREF: Distinguishes cooperation from trust as distinct concepts; relevant to negotiation, game theory, and international relations.

In the encapsulated-interest account of trust, the knowledge that makes my beliefs about you a matter of trust rather than of mere expectations is my beliefs about your incentives toward me in particular. These are not merely bald, unarticulated expectations about your behavior. I have bald expectations that the sun will rise tomorrow, and I might not be able to give any account of why I think that, other than induction from the past. (As a physicist, you might be able to give a very good account, so that your expectation of the sun's rising is theoretically grounded.) What matters for trust is not merely my expectation that you will act in certain ways but also my belief that you have the relevant motivations to act in those ways, that you deliberately take my interests into account because they are mine.

DEFINE: Distinguishes trust from mere expectation by foregrounding perceived motivations and goodwill, not just predicted behavior.

That trust is essentially rational is a common view. For example, James Coleman (1990, chapter 5; also see several contributions to Gambetta 1988) bases his account of trust on complex rational expectations. There are two central elements in applying a rational-choice account of trust: incentives of the trusted to fulfill the trust and knowledge to allow the truster to trust. The knowledge at issue, of course, is that of the potential truster, not that of the theorist or social scientist who observes or analyzes trust. Hence we require an account of the epistemology of individual knowledge or belief, of street-level epistemology, to complete the rational theory of trust (see chapter 5).

DEFINE: Distinguishes two central elements of rational-choice trust: incentives of the trusted and knowledge of the truster — framing trust as a rational calculation. XREF: Connects to philosophical debates about trust vs. rational choice and to 'street-level epistemology' as a distinct research program.

A full statement of the rational theory, including the incentive and knowledge effects, is as stated earlier: Your trust turns not directly on your own interests but rather on whether these are encapsulated in the interests of the trusted. You trust someone if you believe it will be in her interest to be trustworthy in the relevant way at the relevant time, and it will be in her interest because she wishes to maintain her relationship with you. Some accounts of trust do not specifically include reference to the trusted's interest in being trustworthy toward the truster but merely require an expectation that the trusted will fulfill the trust (Barber 1983; Gambetta 1988, 217-18; Dasgupta 1988). Adequate reason for such an expectation, however, will typically turn on an assessment of likely future incentives.

DEFINE: Clarifies the rational theory of trust: trust rests on the trusted party's encapsulated interests in maintaining the relationship, not just on an observed expectation of fulfillment.

The encapsulated-interest account backs up a step from a simpler expectations account to inquire into the reasons for the relevant expectations—in particular, the interests of the trusted in fulfilling the trust. The typical reason for the expectations is that the relations are ongoing in some important sense. There are two especially important contexts for trust: ongoing dyadic relationships and ongoing—or thick—group or societal relationships. The two classes are closely related, and both are subsumed in the encapsulated-interest account of trust. The first class is divisible into one-way and mutual trust relations, both of which are grounded in ongoing dyadic interactions. Such interactions pose incentives to the trusted that are of increasing severity. The sanction that compels the trusted party in a one-way trust game and both of the trusted parties in the mutual trust exchange interaction is withdrawal by the other party and therefore the loss of future benefits from the interaction. The sanction in thick relationships can go beyond such withdrawal to include shunning from the whole community of those who share in the thick relationships.

DEFINE: Encapsulated-interest account frames trust as grounded in the trusted party's self-interest within ongoing relationships, where withdrawal from future interaction is the sanction.

If, however, the players were like the lieutenant colonel and Trifonov and were able to play the game repeatedly over the years, their interests might change dramatically, because both could do very well over many plays of the game.

Hence so long as he can foresee two more plays of the game beyond the current play, it always serves his interest to repay the money. Once the relationship is clearly over, however, and there is no longer any chance of reprisal, there is only a short-run gain from a single final loan, and that is substantially trumped by the gain from cheating and keeping the 4,500 rubles (plus any profit he has made from them).

The game as represented might not include all the relevant payoffs. For example, so long as he has the resources of his power over the military base and his standing in the local community, the lieutenant colonel has power to take vengeance on Trifonov if Trifonov cheats. When he was embarrassed by an unannounced audit and discovered that he would be replaced in office, however, he lost that power.

XREF: Connects to game theory's assumption that payoff matrices capture all relevant incentives, and how real-world power dynamics (like a commander's leverage over a military base) fall outside the formal model.

Hence the main incentive that one faces in a particular exchange in which one is trusted by the other is the potential benefit from continuing the series of interactions. The sanction each of us has against the other is to withdraw from further interaction.

If the backward induction argument is compelling, it is hard to see how rational individuals could ever enter into normal relationships of trust and exchange. All such relationships would have to be grounded in something extrarational, perhaps in normative commitments to be altruistic or more decent than is rational. On this view, the fact that there is apparently a great deal of trust in our lives suggests that we are not rational. I think, on the contrary, that trust is eminently rational and that the backward induction argument is flawed. In brief, the flaw is this: Suppose I know that you are eminently rational and that you believe the backward induction argument. I also know that we could gain substantially from entering a series of exchanges that must terminate, perhaps unhappily, at some distant future point. I can now wreck your backward induction by simply cooperating at our first encounter. You may now suppose I am irrational, or you may reconsider your induction. Either way, you may now decide it is in your interest to reciprocate my cooperation, so that we both gain far more than we would have from continuous mutual defection. Indeed, I think you must reconsider your induction because if I, by acting cooperatively, can get you to cooperate, you should realize that you could do as well with others in such an interaction. That is to say, you must agree that it would be sensible for you to cooperate initially rather than to defect.

XREF: Connects to game theory literature on the finitely repeated prisoner's dilemma and the paradox of backward induction, which many sources treat as a genuine puzzle undermining cooperation.

The correct way to see the role of thick relationships, however, is as one possible source of knowledge for the truster about the trustworthiness of another and one possible source of incentives to the trusted to be trustworthy. The first of these is essentially an epistemological role. Obviously, however, thick relationships yield only a part of the knowledge we have of others. Our understanding should not stop with only the thick-relationship class of epistemological considerations. In practice, this class may often have priority among our sources in our face-to-face interactions, but this descriptive fact does not give it conceptual or theoretical priority.

The thickrelationship theory is therefore merely a special case of the encapsulated-interest theory of trust. It is a partial theory that does not generalize to many contexts. In any theory of trust, the restriction to small-scale thick relationships must follow from other principles. Going back to those principles is a first step in generalizing the theory.

It is common to say that people are rational in some contexts and not in others. One might be seemingly rational in choosing between two jobs but not in choosing a spouse. It is even supposed that some whole cultures are less rational than others. James Scott (1976) has argued that the peasants of Southeast Asia, for example, are driven by a “moral economy.” What he means is that they do not maximize their production of rice. Rather than adopting seed grains that would have large average annual yields, they stick with grains that will almost always produce enough to keep them from starving but much less on average than the most productive seeds.

XREF: Connects to risk-aversion theory and behavioral economics' loss aversion; the 'moral economy' framing parallels prospect theory's asymmetric valuation of losses versus gains.

But preferences make sense only over whole states of affairs, in which income is only part of what matters. The peasants are like anyone else; they want income only for what it will buy for them. If in a bad year with the higher-yielding grain they starve, their income will have done little for them. As is presumably true of Scott and virtually everyone else as well, I also do not have an unrestricted preference for higher income. If higher income entails giving up my academic life or more of my leisure time, I may not prefer it to my present income with my present lifestyle and consumption pattern. There are no preferences that “make sense in terms of income alone.” Income is just a proxy for what we really want.

XREF: Connects to behavioral economics on revealed preferences and utility theory beyond simple income maximization.

The encapsulated-interest account of trust holds that the trusted encapsulates the interest of the truster and therefore has incentive to be trustworthy in fulfilling the truster's trust. The encapsulation happens through causal interactions in the iterated one-way trust game, iterated exchange (or prisoner's dilemma), and thick relationships. None of these, however, is itself definitive of the trust relation. They are all merely ways to give the trusted incentive to take the interests of the truster into account.

If I love you, or am your close friend, or am altruistic toward you, I might directly count your interest to some extent as my own. In economists' jargon, I might partially include your utility in mine. Hence you can trust me to some extent just because the effect of our interaction on your welfare will matter to me. We commonly trust our parents, siblings, close friends, spouses, and others who are close to us in this way within varying limits. One might wish to call these normative instances of trust. But the actual trusting is not different from the purely interested cases under the trust game or iterated exchange. If there is a normative quality to these instances from love and so forth, it is in the fact of the love or friendship and the caring for another that follows from these.

If we consider all the trust relations we experience, we find that a large fraction of them fall into three categories: relationships or interactions that are iterated, those that are backed up by institutions, and those that are mediated by other (noninstitutional) third parties. This chapter has focused on the first of these categories. … The first category—iterated interactions—seems to be far and away the largest in ordinary interpersonal life. That is because much of our lives is spent in ongoing relationships, such relationships constitute much of what is most valuable to us, and we make substantial commitments to one another in such relationships.

I once had an acquaintance of whom many people said, with genuine force, that he was a person you could trust. Alas, that depended on who “you” were. Many people did not trust him at all because they thought him deceitful and manipulative. The latter group included people whose interests often conflicted with his and whose future value to him he had seemingly written off. He could be richly and deeply trusted by those who shared enough of his interests, not at all by those who did not. He was almost mythical in his capacity to put people into two distinct classes. On a cynical reading, he was not trustworthy on the encapsulated-interest account. He was merely reliable to those whose interests he happened to share.

DEFINE: Illustrates the encapsulated-interest account of trust: trustworthiness is not a fixed trait but conditional on shared interests.

When people say, in ordinary language, that they trust the government, they do not mean anything closely analogous to what they typically mean when they say they trust another person. That can become clear, however, only if we first unpack what ordinary individual-level trust is about in common instances.

Second, we can subject ourselves to external inducements. We can try to arrange matters so that our interests will be aligned with our commitment when the time for acting on that commitment comes. We can do this by setting up ad hoc personal devices or by relying on societal and institutional devices. An individual cannot do much about the availability of societal and institutional arrangements—they either exist or they do not. But if they do, I can hope and often even contrive that they will help me rely on you. Legal and other institutional constraints can give us strong incentives to be trustworthy.

The encapsulated-interest account of trust is grounded in the partial alignment of the interests of the trusted with those of the truster, and it comes primarily under the second of these categories. Because many norms and social conventions are backed by sanctions, it is commonly in our interest to follow them. Hence it will often be in our interest to be trustworthy even though it would sometimes be wrong to say that this is because our interests encapsulate the interests of someone who might trust us. If you are driven by institutional constraints or societal conventions, then you have a strong incentive that does not depend on me or any ongoing relationship with me. In such a case, we can have merely congruent or compatible interests without having the least concern for each other.

DEFINE: Clarifies the encapsulated-interest account of trust and distinguishes it from merely congruent interests driven by sanctions rather than shared concern.

Surprisingly, much of the literature on trust hardly mentions trustworthiness, even though implicitly much of it is primarily about trustworthiness, not about trust. Consider four striking cases. First, under the guise of discussing trust, the philosopher Bernard Williams (1988) gives an account of the possibilities of general trustworthiness, from which trust is merely inferred. Second, the economist Roland McKean (1975) ostensibly addresses the economics of trust, but his actual problem is that of trustworthiness. It is not trust itself that is the collective good in his account but trustworthiness. Creating institutions that help secure trustworthiness thus helps to support or induce trust.

DEFINE: Highlights the conceptual slippage between trust and trustworthiness in the scholarly literature—two distinct concepts often conflated.

The fourth case is more complicated. The sociologist Niklas Luhmann (1980, 8) says trust constitutes a more effective form of complexity reduction. This is a very elliptical claim. Clearly, where there is trust that is justified there are increased possibilities for beneficial experience and action. Trust by itself, however, constitutes nothing. Presumably Luhmann is saying that we cannot handle enormous complexity without having others act de facto on our behalf. If we cannot count on their acting in our interests, however, we may be reluctant to empower them or to follow their advice. Hence whatever can secure their trustworthiness enough for us to trust them will help us manage complexity. That is to say, again, the focal problem is trustworthiness, not trust. Elsewhere, Luhmann (1988, 95) says he is concerned with social mechanisms that generate trust. Once again, he has substituted concern with trust for concern with trustworthiness. Commonly, the best device for creating trust is to establish and support trustworthiness. As before, acting on trust without the latter can bring harm.

Linger with this issue for a moment. In The Remains of the Day, Kazuo Ishiguro portrays Mr. Stevens, an aging butler rethinking his life with his late master. In an imagined debate with another servant, Stevens says, "the likes of you and I will never be in a position to comprehend the great affairs of today's world, and our best course will always be to put our trust in an employer we judge to be wise and honourable, and to devote our energies to the task of serving him to the best of our ability" (Ishiguro 1990, 201). He slowly revalues his master from the point of view of others, who detested the master's reprehensible and foolish politics. "At least [his lordship] had the privilege of being able to say at the end of his life that he made his own mistakes," Stevens reflects. "I cannot even claim that. You see, I trusted. I trusted in his lordship's wisdom. All those years I served him, I trusted I was doing something worthwhile. I can't even say I made my own mistakes. Really—one has to ask oneself—what dignity is there in that?" (Ishiguro 1990, 243). Stevens recognizes that trust can finally be stupid and, when it seemingly justifies action or inaction, even culpable.

XREF: Connects to political philosophy on delegation of judgment to authority figures — Milgram's obedience experiments and debates on the Banality of Evil (Arendt).

Hence merely trusting in itself obviously need not help in managing complexity well—it could lead to dismal results, including quick destruction. Again, the core of Luhmann's account of the role of trust must be an account of the importance of trustworthiness. If his account really commends trust in its own right, it elevates Stevens's culpable stupidity. Trust has led Stevens not to manage complexity so much as to fall victim to it.

DEFINE: Clarifies that trust and trustworthiness are distinct: the former is mere reliance, the latter is what actually enables effective complexity management.

In general, the complexity of the problem of trust derives primarily from the complexity of the problem of trustworthiness. As this chapter demonstrates, the motivations for being trustworthy are manifold. In a sense, trusting someone in some context is simply to be explained as merely the expectation that the person will most likely be trustworthy. Trust is little more than knowledge; trustworthiness is a motivation or a set of motivations for acting. Motivationally, there is much to be explained about why someone acts as though with trust when there is little ground to believe that the other is trustworthy, so that, in fact, there is little or no trust. The explanation of your actually trusting in some context will be simply an epistemological, evidentiary matter—not a motivational problem. The explanation of your trustworthiness can be either simple or complicated, and it will depend on motivation.

DEFINE: Distinguishes trust (epistemic expectation) from trustworthiness (motivational), a conceptual split that frames the chapter.

Dylan is wrong, of course, about how surely one can trust oneself. Trustworthiness not only requires that one share the interests of the other as directly or indirectly one's own but also requires the competence to serve those interests.

DEFINE: Breaks down the concept of trustworthiness into two necessary components: aligned interests and actual competence.

The demands made by this commitment might sound commonplace—after all, monks of various religions have chosen celibacy. They do so, however, with substantial social and institutional constraints to keep them celibate, and even then they often fail. Alberich commits himself without any social structure to give him the incentive to stay unloving.

One cannot simply start trusting people as of tomorrow unless the people one deals with and one's relationships with them are suddenly different in relevant ways and one is privileged to know this.

QUESTION: What does the author suggest as the actual path toward building trust if it can't be simply started?

The issue for this chapter is what a new person can do to convince me of his or her commitment to fulfilling my trust. For example, it might have been in Alberich's interest to say he would forswear love forever even if he knew he would later cheat. What could have made his curse credible?

DEFINE: Explores the concept of costly signaling — a commitment is only credible when retracting it is expensive.

What we lose when receipt of the benefit and fulfillment of the commitment are separated in time is the coupling of the motivations into, in essence, a single motivation. It would be strange in the future moment to say that we violate our earlier motivation when we no longer wish to fulfill our commitment. This is the central problem of bald commitments into the future: we cannot tie our future and present motivations together to yield a single “net” motivation for action.

In either case, if I believe you have strong moral compunctions to be reliable in some context, then I have strong reason to rely on you in that context.

Note that this disposition seems genuinely to be a disposition to be trustworthy, not merely to act in a trustworthy manner, although the latter may often be entailed. A comparable disposition to be trusting seems much less likely, because its range is radically larger than that of any disposition to be trustworthy. If I am always trusting, I will be the gullible target of increasingly many people. After losing repeatedly from my excessive trust, I will have overwhelming incentive to alter my disposition. I might still have a risk-taking disposition, but this will be more limited in that it will not lead me to take egregious risks. Hence trust and trustworthiness are not analogous or symmetrical in this respect, because one can be disposed to trustworthiness without significant risk.

Also note the important potential fact that contrary incentives might not be very effective in altering motivations grounded in moral dispositions. We do not have a good account of the psychological or other force of such dispositions, but they seem to be extremely strong in some cases. We should qualify this claim, however, because psychological research suggests that we tend to overestimate dispositions as a cause of behavior (Jones 1979). On the contrary, we may tend to underestimate the role of interests, as is suggested by the account of supervenience in chapter 3.

XREF: Relates to fundamental attribution error research (Ross / Jones 1979) and the disposition- versus-situation debate in social psychology.

One thing we can say fairly conclusively is that, in a moral rule system that is articulated enough to cover much of the moral life, the rules must, as in the case of the norms discussed earlier, occasionally be in conflict with one another. For example, in one of the most contested remarks by any major moral theorist, Immanuel Kant holds that one is bound not to lie even when faced with an assassin who asks whether his intended victim is in one's apartment. Kant supposes that effectively turning over the intended victim to the assassin is a less grievous wrong than lying to the assassin. He writes that "although by a certain lie I in fact do no wrong to any person [such as the intending murderer], yet I infringe the principle of justice in regard to all indispensably necessary statements generally (I do wrong formally, though not materially); and this is much worse than to commit an injustice to any individual" (Kant 1909 , 365, final emphasis added; see further Hardin in press b, chapter 6). A rule to be trustworthy could run aground on Kant's account of his rule against lying. (Beware of such "moral" rules.)

XREF: Connects to Kant's absolute prohibition on lying and its tension with justice — relevant to work on deontological ethics and rule conflict. QUESTION: Worth interrogating how any articulated moral rule system could reconcile inherent rule conflicts; the assassin example provides a concrete test case.

In a utilitarian assessment of our various dealings with another we can commonly suppose that the interactions are more or less decoupled from all else, and we can simply choose according to what's best for the pair of us. In standard mutual trust interactions, I typically cannot justify cheating you on utilitarian grounds because I cannot judge that the pair of us is made better off on the whole unless we complete our exchange. We will know that the move from the status quo ante of no exchange to the condition of a completed exchange benefits both of us, so that it is better overall than not exchanging. If we cannot perform accurate interpersonal comparisons of utility, we cannot conclude that either of the outcomes in which I cheat you or you cheat me is better overall than the status quo ante of no exchange. Hence consummation of a mutual trust relationship is typically utilitarian, so long as it does not have substantial negative effects on others. A strong utilitarian commitment therefore generally implies a strong commitment to trustworthiness.

Nevertheless, one can sometimes "will" oneself into future action indirectly. One can predict well enough how one would act in certain circumstances and can therefore sometimes set matters up to stimulate a particular response. For example, I may know that, once I get out of bed in the morning, I am really up and I will not return to sleep. Therefore I set my alarm clock on the dresser a few feet from my bed, so that I will have to stand to stop its noise. I have then used an external device to get myself on my feet, but it is something internal about me that keeps me from returning to bed and to sleep. If the alarm were too near, I might often turn it off and go back to sleep.

When such quasi-internal devices are unavailable or weak, a more forceful move is to invoke external constraints. For example, we can make our commitment in public in full self-understanding that our embarrassment over publicly failing the commitment would most likely trump our costs of keeping to it. Then we do not face motivation from the commitment itself but from social constraints that make it in our interest to fulfill the commitment. That is to say, we no longer depend on contriving commitments to be motivators in their own right. Indeed, by making the costs of reneging on our commitment high, we can virtually bring our future action of fulfillment into the present so that we tie our present and future motivations into a single net motivation now for action in the future. In a suggestive term of older contract law, we presentiate that future.

In her society, however, Maria may need to construct her own reinforcements to keep her faith. Suppose that many American Catholics are like Maria—active because they want to be strongly tied to the church rather than because they share all of its politics. Commitment to the antiabortion movement might therefore be much stronger in a less Catholic society, such as the United States, than in a more solidly Catholic society. The difference is, however, not evidence of the greater intrinsic importance of the morality of abortion in the less Catholic society. Rather, abortion is taken on as an issue on the way to something else—namely, maintenance of the personal faith of future selves.

XREF: Connects to theories of minority group solidarity and religious boundary-maintenance; also echoes 'sacred canopy' decline arguments where committed minorities become more active than complacent majorities.

Under the social conventions of her time, Ukifune's was a virtually irrevocable commitment. In our society of radically looser conventions, you and I cannot so readily constrain ourselves as Ukifune did—our shaved heads might be nothing more than a frivolous style of the moment—and we cannot be fully believed if we assert our undying commitment to a particular religious creed or to any other purpose. We can be believed if we say we are committed in this very moment—but we cannot be fully trusted to stay committed into the distant future. The most assertive believers among us are all too often the most fickle. We are like Alberich without magic to control us.

XREF: Connects to the reader's thinking about commitment, identity, and how social structures shape belief — relevant to moral philosophy or cultural commentary.

Ukifune wanted to commit to leaving her world, the world of the court in Heian Japan. She did so by taking vows that would preclude her return to the secular world. First, the rules of her commitment are easily stated: she must live in a convent and not range into the larger world. Second, her actions were extremely easily monitored. If she left, her absence would be instantly noticed by the small group of her fellow nuns, and her presence elsewhere would be instantly seen as wrong even by those who knew nothing of her, because they would be able to see that her hair had been shorn. Third, the sanction against any attempt on her part to return to the court would be essentially of no cost to most of the courtesans, courtiers, and other hangers-on at the court—they would almost all have coordinated on shunning her. The sanction would have been devastating to Ukifune if she had really been intent on returning to the court. Finally, the rules on her behavior after taking vows could firmly be expected to last long beyond her life. They were stable.

XREF: Connects to game theory and credible commitment mechanisms — the logic resembles reputation-based commitment devices described in economics and political science.

If I trust you to act on my behalf, I set myself up for the possibility of disappointment, even severe loss. To avoid that possibility, I might try to find institutional backing to get you to do what I trust you to do. This might be easy to do. For example, if I wish to get you to pay in several installments for the work I do for you or for your purchase of my car, in many societies I can propose that we sign a legally enforceable contract. Now you are faced with the likelihood of real costs if you renege on doing what I have relied on you to do. If the only reason you comply with our agreement is the threat of contract enforcement, then ours is finally not a trust relationship.

DEFINE: Distinguishes genuine trust from compliance driven by institutional enforcement threats.

Had his dealings with Trifonov been reputable, the lieutenant colonel could have been protected by a contract that gave both parties the right to be sued. “Who wants to be sued?” Thomas Schelling (1960, 43) asks. Well, he notes in answer to his own question, “the right to be sued is the power to make a promise: to borrow money, to enter a contract, to do business with someone who might be damaged. If suit does arise, the ‘right’ seems a liability in retrospect; beforehand it was a prerequisite to doing business.” The odd right to be sued is the “power to accept a commitment.” It enables one to establish that one has a strong commitment to fulfill one’s half of a bargain. Trifonov had no right to be sued by the lieutenant colonel, who, indeed, could not even publicly accuse him.

XREF: Connects directly to Schelling's commitment and strategy game theory, and to commitment devices in decision-making literature.

The contract or audit may protect a relationship against the worst of all risks it might entail, thereby enabling the parties to cooperate on less risky matters. This is a milder and more specific variant of the central argument in Hobbes’ theory of political order. The threat of sanctions to protect each makes all better off. Hobbes’ sovereign enforces order in the protection of life and property. Once these are secured, we have less reason to be defensive, and we engage in productive investments and beneficial exchanges.

XREF: Connects to Hobbes' Leviathan and broader social contract theory — the enforcement machinery that underwrites trust and cooperation.

Consider a highly articulated differentiation of people into different classes of apparent trust. In Leo Tolstoy’s Anna Karenina, Count Vronsky’s code of social rules is probably well understood by all concerned, who therefore have differential grounds for expecting good or bad behavior from him. As Tolstoy (1949 [1874–76], 1.347) puts it, “The code categorically determined that though the card-sharper must be paid, the tailor need not be; that one may not lie to a man, but might to a woman; that one must not deceive anyone, except a husband; that one must not forgive an insult but may insult others, and so on.” Karenin therefore can be confident that Vronsky will be honorable in repaying a gambling debt but he cannot be confident that Vronsky will be honorable with his wife, Anna—and therein lies a long novel.

What was the nature of Vronsky's concern to repay his gambling debt? It was not merely a reciprocal concern with his gambling partner of the moment. It was, rather, a concern to live up to the standards of behavior of the nineteenth-century Russian aristocratic community of which he was a member. If he failed to repay the debt, he would be embarrassed in that community and would lose standing in it, so that his opportunities within that community would be constrained. He faced a norm of exclusion, a norm that required his submission to the group in various ways at the cost, if he did not submit, of being shunned within the group and perhaps even excluded from the group.

DEFINE: Introduces the 'norm of exclusion' — a mechanism through which group standards enforce individual conformity via threat of shunning.

If we wish to break the hold of such norms as those that reinforce the exclusionary status of some group such as the aristocrats, a forceful way to do so would be to offer contrary incentives. That is what the modern industrial economy has inadvertently done to many aristocratic norms, such as the dueling norm (Hardin 1995, 91–100). Some of Vronsky's class norms might, however, seem generally good, so that we might wish to keep them. Unfortunately, if they are merely universalistic norms, so that following them is not motivated by the exclusionary sanctions of the status group, they will most likely have far less force. No one other than the victims of violations will have incentive to sanction violations of them. Universalistic norms that are not enforced dyadically, as the norm of promise keeping or that of truth telling typically is, are generally weak norms (Hardin 1995, chapter 5).

XREF: Connects to Russell Hardin's work on norms and to the Anna Karenina example of dueling norms undermined by industrial economy.

Suppose the lieutenant colonel thought that Trifonov was under a norm of trustworthiness of some internalized kind. Such a belief, while perhaps common, would actually be quite odd in this case. The lieutenant colonel was, after all, engaging in criminal activity with Trifonov, and he was violating the trust of his command and of his nation. Perhaps the lieutenant colonel would himself have felt bound by a norm of trustworthiness in strictly personal relations while blithely violating any such norm with respect to the Russian army and the Russian nation. More generally, any broad system of norms is likely to include within it quite different norms that would commend contradictory behaviors in common contexts. It is perhaps therefore not surprising that universalistic norms tend to be vague and limited, and they might tend to be weakened by their potential conflicts.

Finally, consider how internalization might work. Perhaps internalizing norms is akin to developing habits. In a sense, although it need not be a deliberate move, we economize on thinking and calculating. We adopt habits and norms as devices for deciding in repeat contexts how we should act. Then, following a commonplace but seemingly compelling fallacy, we moralize these shortcuts from being merely devices that we should pragmatically use to being devices that are morally incumbent on us to use. This is the standard fallacy of supposing that what is also ought to be—the is-ought fallacy of reasoning.

DEFINE: Explains how the is-ought fallacy operates in practice: pragmatic habits get moralized into obligations.

It would be useful to have systematic analyses of trust that is grounded in each of the bases of trustworthiness discussed here: internal, external, and mixed motivations. In this book, I focus primarily on encapsulated interest as the ground for trustworthiness, although I also discuss categories that are closely related to encapsulated interest. I think it will be harder to give systematic accounts of trust in all its workings for trustworthiness that is grounded in moral commitments, norms, and bald dispositions. The knowledge problems of assessing such commitments in general seem likely to be harder than those for assessing interests.

DEFINE: Distinguishes encapsulated-interest trustworthiness from grounds of moral commitments, norms, and bald dispositions, with the latter posing harder knowledge problems. QUESTION: Why are interests inherently easier to assess than moral commitments? Is this an empirical claim or a conceptual one?

Interest is one of the best and most useful of internal motivations. It leads us to and through much of the best that life has to offer. To suppress interest on occasions in which it leads us in directions we might consider wrong is to put ourselves at war with ourselves. In waging that war we are supported by social controls based in our own longer-term interest. A strong network of laws and conventions is needed to make any kind of behavior reliable if it is likely to conflict with powerful considerations of interest. All too often, we have neither law nor convention nor love to make us trustworthy, and it is often only the interest we have in maintaining particular relationships that makes it in our interest to be trustworthy.

XREF: Connects to intrinsic motivation research (Deci & Ryan self-determination theory) and the wellbeing cost of suppressing natural drives.

Still, trust can be grounded in the belief that another person is guided by norms or moral dispositions. Whether we know that these, and not encapsulated interest, are the grounds for seemingly trustworthy behavior will commonly be important for explaining behavior and understanding social practices and institutions. We can build institutional devices that mimic the incentives of the encapsulated-interest account, so that we can relatively easily see how to overcome problems of the lack of trust and trustworthiness in many contexts in which, for example, ongoing relationships cannot motivate cooperation.

DEFINE: Distinguishes two grounds of trust: encapsulated interest vs. shared norms or moral dispositions.

Other motivations to be trustworthy might not be so readily mimicked by institutionally devised incentives, especially if those motivations are not matters of simple incentives. When these motivations are at work, trustworthiness might be overdetermined in the sense that one of these motivations might be coupled with encapsulated interest. Hence one might say that one or the other of these is the essential reason for the trustworthiness. In truth, however, parsing the motivations would be quite difficult unless we could compare the same person's behavior over a wide range of interactions.

DEFINE: Introduces the idea that trustworthiness can be 'overdetermined' — multiple motivations coupled with self-interest, making the essential reason hard to isolate without observing a person across many interactions.

Trustworthiness may be inherently moral in part for at least some people, or it may be compelled by the force of norms. That many accounts of trust are really accounts of trustworthiness therefore suggests that the moralizing of trust might be more reasonably seen as a moralizing of trustworthiness.

Trust is a three-part relation: A trusts B to do X, and even this is restricted to certain contexts and ranges over degrees of trust. The fourth and fifth slippages involve the reduction of trust to a two-part relation. In our daily invocations of trust we commonly do not include the third part. This might merely be an ellipsis of what we mean, but it might also begin to inform what we think, so that I begin to assert that I trust you, purely and simply, as though with respect to anything and everything.

DEFINE: Defines trust as a three-part relation with an object of trust, arguing that collapsing it to a two-part relation is a slippage that can reshape our thinking.

As a reductive term, trust is similar to other major theoretical terms such as power, equality, and justice that must be explained and are not primitive terms but are reducible to other terms. For many trust relationships, we can determine what these other terms are from the structures of those relationships, as in the account of mutual trust in ongoing exchange relations discussed in chapter 1.

DEFINE: Clarifies that trust, like power and justice, is a reductive theoretical term requiring explanation via other constituent terms rather than being primitive.

It is hard even to imagine a conception of trust that is nonreductive and still plausible. One might argue that joy is conceptually not reductive, although it is probably causally explicable in reductive terms. Trust is not merely an unvarnished emotion, however, in the sense that joy is. Emotions might come into particular trust relationships, but trust is not itself simply an emotion.

DEFINE: Distinguishes trust from a mere emotion by arguing trust cannot be reductively explained, even though joy potentially can.

Finally, note a peculiar implication of an ordinary-language analysis of trust. Anyone who wishes to make trust a simple primitive or who wishes to take it as a deontological moral concept should have trouble with real-world experience, which often lacks any such notion. (A deontological concept is universally applicable and derivable from pure reason.) Even today, many languages have no direct, perspicuous equivalent of the term.

DEFINE: Clarifies what a deontological concept is: universally applicable and derivable from pure reason.

That trusting someone and acting on that trust are different is trivially evident. For example, I may trust you in various ways but never have reason to act on that trust. Against this distinction, Jane Mansbridge (1999) argues for “altruistic trust,” which implicitly runs trust into action. To speak of altruistic trust is implicitly to say that it is not trust, that one acts cooperatively beyond what one would do if one acted only from the degree of trust one had in another. One acts altruistically despite lack of sufficient trust to justify the action. Hence to speak of altruistic trust is to make trust a term of action rather than of knowledge.

DEFINE: Distinguishes trust as epistemic (a form of knowledge) from altruistic action that exceeds one's actual degree of trust.

We can choose to put ourselves in a position to come to know something, but we cannot look at the evidence and then decide to know. The evidence might compel us or it might not, but we do not choose the degree to which it does compel us. The recognition of this point and its denial are in contest in such claims as that it “is not possible to demand trust of others; trust can only be offered and accepted” (Luhmann 1980, 43). No. My trust of you can be neither offered nor withheld. It just is. I cannot withhold it from you—and it would make no sense to do so even if I could because it would be contrary to my interest as I understand it. Similarly, you can neither accept nor refuse my trust. Both of us, however, can choose to act or not to act on it in various contexts.

XREF: Contradicts Luhmann's framing that trust is offered and accepted; also challenges common exchange-model thinking about trust in relationships and institutions.

Commonplace claims that one chooses to trust entail mistaken implications that trusting is a matter of acting. Kenneth Arrow (1974, 26) speaks of the “agreement to trust each other.” John Dunn (1988, 73, see also 80) says that trust is “a more or less consciously chosen policy for handling the freedom of other human agents or agencies.” Annette Baier (1986, 244) speaks of “conscious trust the truster has chosen to endorse and cultivate.” Luhmann (1980, 24) speaks, as do many others, including Baier (1986) and Virginia Held (1984, 65), of trust as a gamble, a risky investment. Held (1968, 158) also says that one may be obligated to trust. All of this is wrong. I just do or do not trust to some degree, depending on the evidence I have. I do not, in an immediate instance, choose to trust, I do not take any risk in trusting. Only actions are chosen—for example, to act as I would if I did in fact trust or to take a chance on your being trustworthy beyond any evidence I have that you will be trustworthy.

XREF: Contrasts with game-theoretic and rational-choice treatments of trust popular in economics and political science.

Therefore, it is not credible that the standard survey results really imply merely one-part or two-part relations. It is virtually inconceivable to suppose it is a one-part relation if we pay even slight attention to what it could mean: I trust, period, everyone and with respect to everything. The standard survey question refers to “most” people; but then, it also says “most of the time,” which could be taken to mean that the respondents must restrict the range of matters on which they would trust “most people.” This is not genuinely generalized trust. The respondents are forced by the vagueness of the questions to give vague answers, and it is a misdescription to label their responses as generalized trust.

QUESTION: What would a properly formulated survey question for generalized trust look like, and would it change measured levels?

Many, maybe even most, claims for generalized trust can readily be restated as claims that, in contexts in which trust generally pays off, it makes sense to risk entering into exchanges even with those whom one cannot claim to trust in the encapsulated-interest sense—because one does not yet have an ongoing relationship with them nor does one have reasons of reputation to trust them. This is not a claim that one trusts those others but only that one has relatively optimistic expectations of being able to build successful relationships with certain, perhaps numerous, others (although surely not with just anyone). If the context is even slightly altered, this conclusion may be wrong, as it is in dealings with con artists who propose quick-profit schemes or, often, with sellers in tourist traps.

DEFINE: Distinguishes encapsulated-interest trust from a weaker, context-dependent optimism about exchange with strangers.

The moral burden, if there is one, is typically on promise keeping and being trustworthy, not on either promising or trusting. That is because the immediate, short-run incentives in relationships of promising and trusting are on the side of failing to fulfill the promise or the trust. The potential promise keeper, however, is first the promiser. I make a promise to you, and I am expected to keep it. On standard understandings of obligation, I impose my own obligation on myself when I promise. The trust relationship is very different. The one who faces the burden of trustworthiness is not the one who trusts. I trust you, and you are the one who is expected to fulfill the trust. But I cannot impose an obligation on you to fulfill that trust. Moreover, it would be odd generally to moralize promising, as opposed to promise keeping; similarly, it should be seen as odd generally to moralize trusting, as opposed to trustworthiness.

The strategic structure of promise keeping is quite varied, with three distinct kinds (Hardin 1988b, 59–65). First, there are promises made without any reciprocal benefit to the promiser, whose fulfillment is essentially altruistic. Such promises are commonly called gratuitous, and in some societies there is law to govern these while in others there is little or none (Dawson 1980). Second, there are coordination promises, which do little more than coordinate joint actions at future times, as in our mutual promise to meet for lunch tomorrow. Finally, there are exchange promises, which are explicitly a reciprocal part of an exchange relation. In a case of the latter, for example, you may do something for me now in return for my promise to do something for you later.

DEFINE: Distinguishes gratuitous, coordination, and exchange promises as a taxonomy.

Laws of promise keeping can have substantial force for exchange promises but little or none for coordination promises. They commonly have little force for gratuitous promises—unless there is subsequent reliance that gives the promisee a loss relative to the status quo ante if the promise is not fulfilled. Even for exchange promises, the law usually enters only if the stakes are relatively high and there has been some reliance that would prove costly if the promise were not fulfilled.

DEFINE: Distinguishes exchange, coordination, and gratuitous promises by how much legal force each receives.

The law, as opposed to the mere practice, of promise keeping is an instance of the substitution of institutions for trust relationships. That law primarily covers some classes of promises, especially exchange promises involving significant values in exchange, while neglecting others, especially small-scale exchange promises for which recourse to the law would be disproportionately expensive and coordination promises, in which there are no values in exchange. … The law takes over those areas in which there is significant value at stake and in which trustworthiness would be inherently less reliable, because it would face endgame incentives either in a once-only interaction or in an ongoing relationship faced with unusually high stakes in a single interaction.

XREF: Connects to game theory concepts of one-shot versus repeated games — endgame incentives and defection risk. DEFINE: Clarifies the role of contract law as institutional substitution for informal trust, covering selective promise classes.

With trust, however, we cannot even formulate an analog of either the conventional or the logical account of an obligation. It is I who promise and who thereby assume an obligation. But, if it is I who trust, it is the trusted who would have to be burdened with an obligation.

DEFINE: Distinguishes trust from promising by showing the direction of obligation is inverted — I assume the obligation when I promise, but the trusted would bear it in trust.

In actions that come under any of these terms, our purpose is commonly to bring the future into the present in some sense—a move that contract lawyers call presentiation. There may be some joint project or exchange that we cannot complete in this moment, but we might wish to secure our expectations about its future before we expend effort or resources now or make commitments in the future whose value will turn on completion. … As Luhmann (1980, 13) says, "Managing complex futures means corresponding performances by people in the continuous present." He then adds, "Trust is one of the ways of bringing this about." The addition is not quite right. We hope to bring it about by entrusting, contracting, or accepting a promise, by turning over some future part of our project to another or others. It is the turning over—that is, an action—that fits that hope.

DEFINE: Explains the contract-law concept of 'presentiation' — bringing the future into the present through commitments.

Several other visions of trust that are fairly clearly articulated, sometimes only implicitly in part, differ from trust as encapsulated interest. … These assumptions are that trust is not a matter of knowledge (in other words, it is noncognitive), that it is normative in some moral sense, that it is dependent on the characteristics of the trusted other than the trusted's commitments to the truster, and that it is a good in much the way that various consumption goods are goods or that capital of various kinds is a good. … The extant accounts of trust as a good are misleadingly said to be about trust when in fact they are about trustworthiness. When they are cast as being about trust, they are incoherent; when recast as being about trustworthiness, they can be compelling.

DEFINE: Distinguishes trust (encapsulated interest) from trustworthiness, arguing accounts cast as about trust are really about trustworthiness.

One may ask whether this is a credible psychological view. Robert Frank (1988, chapter 3) thinks we are likely to trust even against reason if we develop or inherit moral characters of trustworthiness. He might further think we ought to develop these moral characters (even if we did not inherit them) if doing so is—as he seems to think—the cost of developing a character that will benefit us. Here, however, rationality, or self-interest, stands in the background to justify the character development.

XREF: Connects to rational-choice theory debates and the puzzle of why people cooperate against narrow self-interest.

If our differentiation of whom to distrust and how much is so finegrained as to allow us to trust our close associates very much and to distrust our ruler almost entirely, the role of our noncognitive trust in motivating us seems quite limited

Because the infant has no choice but to depend on its parents, there is no point in saying the infant trusts them. Indeed, it is not sensible to say that Abraham trusts his god, given his beliefs. He might have failed to follow that god’s orders because of weakness of will, or he might have revised his view of the goodness of his god and the rightness of following his orders. If Abraham could not revise his beliefs, however, then there would be no question of choice for him of whether to sacrifice Isaac, any more than if his son had been taken from him by disease.

DEFINE: Clarifies that genuine trust implies choice — if beliefs can't be revised, there is no question of trusting, only of compulsion or dependence.

On Baier’s account, trust is an extension of the infant’s relation to its parent, especially its mother. Yet, she notes, a “constraint on an account of trust which postulates infant trust as its essential seed is that it not make essential to trusting the use of concepts or abilities which a child cannot be reasonably believed to possess” (Baier 1986, 244). If this is so, then the encapsulated-interest view of trust is inherently wrong because assessments of trustworthiness could only be based on an infant’s instinctive, behavioral learning.

XREF: Connects to Baier's philosophy of trust and to developmental psychology debates about whether infant behavior reveals genuine trust or mere conditioned learning.

An account of the life of the infant and its necessary dependency might seem cogent as an account of how we come psychologically to be able to trust or to know (as argued in chapter 5). That is not an account of what trust is or of how it works, however. It is not only preadult, it is prehuman. It is plausibly the way an infant bird works, turning a wide-open mouth to the sky with an instinct behaviorally equivalent to acting on trust that good things will fall into it. In the survival lottery, this works for at least some number of the instinctively openmouthed. Even at poor odds of survival with a dumbly open mouth, however, opening up is more beneficial than keeping the mouth closed, which guarantees starvation.

XREF: Connects to the account of infant psychological development in chapter 5, which the author argues is not an account of what trust itself is.

Such instinctual considerations are arguably a compelling part of an account of knowledge, including knowledge that backs up trust, knowledge of the reliability of any particular other and of others in general. That kind of knowledge is inherently inductive, and one might suppose it wise to be skeptical of inductive knowledge until the run is fairly long. Our normal, biologically inherited proclivities may be to make an optimistic assessment of a short but so-far positive inductive run. If we live in a culture in which that optimism typically is justified by longer-run experience, we develop rich relationships from trusting others. Still, although it befalls us rather than being chosen by us, trusting depends on objective data and is subject to correction if experience recommends.

Recall Buber's (1951, 7) supposition that I trust someone "without being able to offer sufficient reasons for my trust in him." Note how peculiar this would be in a real-world relationship. When I first meet you, perhaps I take risks merely optimistically or out of hope. For Buber this could be genuine trust, because it has no grounds, at least no grounds in you. Ten years later, when I have great experience of your trustworthiness, I genuinely do trust you. This would not be trust for Buber, however, because it is based on sufficient reasons. Buber insists that trust is, in the end, ungrounded.

QUESTION: This inverted definition begs the question: does trust that earns sufficient reasons stop being trust for Buber, and is that tenable?

Almost everything written on trust recognizes the possibility of accounting for much of what we commonsensically call trust with a rational-choice, expectations theory. Much of the literature, however, including contributions by the economists Oliver Williamson and Robert Frank, the philosophers Baier, Becker, and Hertzberg, and the theologian Buber, suggests or even insists that there is something left over. Some of these people seem to want to say that the rationalchoice elements or cases of apparent trust are not trust at all. Williamson (1993, 479) says that what he calls personal trust is "nearly noncalculative."

XREF: Connects to debates between rational-choice theory and social norms/relational accounts across sociology and economics.

Part of the claim in each case seems to be that one just knows there is something else at stake—something moral or psychological or whatever. This putative fact recommends that we try to understand how or why someone comes to think of his or her own actions as other than rational. … Oddly, however, there is not “a” case for an extrarational notion of trust. Indeed, there is little analytical agreement at all on any notion of trust, other than perhaps a residual interest notion. What case there is for something beyond this residual is essentially a cluster of descriptive claims about trusting behavior. The case is that it seems that more than the self-interest of the trusted is involved in the commitments of the truster.

DEFINE: Clarifies the 'residual interest notion' of trust versus the extrarational notion built on descriptive claims about trusting behavior.

If we wish generally to moralize the notion of trust, we will have to take out cases of acting on trust to accomplish bad ends. Such a move, however, is perverse in either an explanatory theory or a conceptual account. It is better just to let trust work its way and then to judge the morality of acting on it. Members of a community may trust one another in ways that are commonly all to the good, and yet their trust may enable them to subjugate and brutalize a neighboring community. For example, Ratko Mladic and Radovan Karadzic might well have been able to cooperate to wreak gruesome harm on Bosnian Muslims only because they trusted each other.

XREF: Connects to discussions of how virtuous traits or institutions can have dark applications — parallels the problem of loyalty in organized crime or trust within authoritarian regimes.

Consider whether it could be required in common contexts. Is it moral or immoral to know or not to know, say, that Afghanistan is ruled by the Taliban? … Unless you can say, as one might of the State Department official, that I am in the wrong not to know those things about you that attest to your trustworthiness with respect to what is potentially at issue between us, then you cannot say it is morally incumbent on me to trust you. Similarly, to say that it is a virtue to trust is to say that it is a virtue to know that others are trustworthy. This is largely incoherent. It is surely a virtue in many contexts to act as one would if one did trust. It is the behavior or the inclination to such behavior, however, not knowledge, that is a virtue.

DEFINE: Reframes trust as an inclination to behave as if one trusted, separating it from knowledge of another's trustworthiness.

Moreover, although one should not conclude too much from etymology, note that a strictly moralized notion of trust runs up against an odd history. The Anglo-Saxon word “trust” is cognate with “tryst.” Tryst today has a somewhat salacious ring, as in an appointed meeting with a paramour. In Middle English, it merely meant, especially in hunting, holding one’s place in a team effort. I hold my place while you drive the game toward me, and I catch or kill it. In Middle English, I stand tryst. It is my trust to hold my place, so naturally you might come to say you trust me to do so. Note, of course, that it is likely then also to be in my interest to fulfill your trust by holding my place.

DEFINE: The etymology of 'trust'/'tryst' reveals its original cooperative, non-moralized sense of holding one's place in a team hunting effort.

These considerations might commend acting as though one trusted others in some matters. However, these possibilities do not entail having the knowledge actually to trust—indeed, they are defined by the lack of such knowledge. To call these actions trust is to slip into making trust a behavioral term, so that the phrase “acting on a trust” would be redundant.

DEFINE: Distinguishes genuine trust (an epistemic state) from mere trust-like behavior, warning against collapsing trust into a purely behavioral term.

Suppose I rely on you more than can be enforced by withdrawing from future interactions with you. I act as though I trust you when in fact I do not. My action is virtually a gift—or, better, it is a gamble and a hope, not a manifestation of trust.

DEFINE: Distinguishes genuine trust from merely acting as if one trusts when enforcement mechanisms are absent — the gamble is simulation, not trust itself.

In contrast with the case of altruism, however, it does not make sense to cut trust free of mooring in expectations and hence, at least potentially, in interests. We cannot cut it free because our expectations will be grounded in factual assessments of the motivations of anyone we might trust. Among the most important of these must commonly be interests.

Even when we think someone is morally obligated to be trustworthy, we might still doubt that he or she will be. For example, we might suppose a police officer is obligated to be fair and helpful in various contexts, but we might have reason to think she will not act from that obligation. Despite her presumed moral obligation, then, we might not trust her.

For the rug dealer in the bazaar, the difficulty of selling to customers unable to pay on the spot might be commonplace. If experience shows that those customers are reasonably reliable, then the rug dealer might profit substantially from increasing total sales by taking the risk of letting people send the money later. … Barth's rug merchant might have had some sense from experience or guessing, rightly or wrongly, that some types of people (those who struggle to speak the local language, women, men, the well-dressed, the friendly, those who seem to be knowledgeable about rugs or who praise the workmanship and artistry) will often enough send the money to make it worth the risk of parting with rugs on expectations—perhaps with less willingness to bargain down to the lowest price to cover for the sometime person who does not send the payment.

Just because their interaction is in this forced form, they actually have reason to expect all others to want to live up to the requirements of their own parts in the enterprise. Each therefore at least partially encapsulates the interests of the others. We may call this swift trust (Meyerson, Weick, and Kramer 1996). … In such contexts, too much personal knowledge and individuating data might even be counterproductive.

In general, it seems that, in the end, all that one can invest in is reputation, which is not necessarily correlated with trustworthiness. The only way to actually affect trustworthiness is by changing one's incentives—for example, by entering into long-term ongoing relationships with those whose trust one would like to have. Hence trustworthiness is not a commodity, even though perceived trustworthiness (that is, reputation) is.

As Gambetta (1988, 225) says, trust is "a result rather than a precondition of cooperation" (though Bacharach and Gambetta hold a contrary view). It is actually both, as is not surprising in an iterated interaction in which there can be feedback from each to the other. Nevertheless, Gambetta's (1988) point seems to be basically correct. I risk cooperating and, if it pays off, I begin to trust you.

XREF: Relates to iterated game theory work like Axelrod's The Evolution of Cooperation, where repeated interaction enables trust to build incrementally.

Coleman, Putnam, and others say trust is social capital, or an element of social capital. It appears, however, that the core of the meaning of social capital in the work of these scholars is not trust but rather the social relationships, or the networks of such relationships (as emphasized by Brehm and Rahn), that enable us to undertake cooperative endeavors. These relationships, of course, ground trust among the participants in them. They do so because we have incentive to be trustworthy to those in our networks, therefore making it beneficial to us to trust one another in various undertakings. Hence it seems wrong to think of the trust itself as an element of social capital.

There might be some feedback between trust and further development of trust. I cooperate with you, discover your trustworthiness, and therefore cooperate even more or on even more important matters with you. If I trust most of the people with whom I interact, I might also begin to take the risk of cooperating with almost anyone I meet, at least if they are likely to remain in my ambit. Hence my general optimism about others is a benefit to those others when they might wish to cooperate with me (or even to abuse my optimistic expectations). Again, however, it is the high level of trustworthiness of people in my network that generates this benefit.

XREF: Connects to network theory and the logic of social capital/reputation, echoing recent book club readings on how trust scales within bounded communities.

When I act in a way that seems to fit my interests, I might nevertheless assert that I am acting out of moral commitment, character, or something else. Moreover, I may be extremely good at acting in my interests and still be genuinely committed to the claim that my motivation is not that of self-interest. Given that more than one of these motivations—interests, morality, norms, identity—might govern my behavior, trustworthiness, and therefore trust, can be overdetermined.

DEFINE: Introduces 'overdetermination' as a concept applying to trust: more than one motivation (interests, morality, norms, identity) can jointly ground trust.

In a somewhat analogous sense, the ordinary supposition that we act from norms may supervene psychologically on motivation by interests. For example, I develop a collection of friendships, each of which is rewarding to me in its own ways. I value the friendships for what they do for me, and I spend more time on and develop those that do most for me. I also begin to think of them as normatively valuable. I would strongly assert that they are good—and, of course, they are good, in the sense that they are good for me, which is to say that they benefit me in various ways. Once they are in place, however, I think of them as morally grounded.

Consider an odd example of a claim that some behavior is normative rather than merely straightforwardly self-interested. In North America virtually everyone drives on the right side of the road, and in any other nation, on the side that is the norm in that nation. I have asked many people how they explain that action. Almost everyone answers that it is immoral or illegal to drive on the wrong side and moral to drive on the correct side. Of course, that answer from morality is analytically true because it would be murderous to drive on the wrong side. It seems motivationally unlikely, though, that we drive on the legally prescribed side because we would feel immoral if we did not or merely because we are law abiding. Indeed, if someone is stopped by a highway patrol officer for going the wrong way on a divided highway, the driver's first reaction must commonly be gratitude for being saved from a potentially devastating accident. The problem is merely coordination of interests, not moral truths.

DEFINE: Illustrates the distinction between normative behavior and straightforward self-interest through the driving-side example.

Finally, note that there may be a disanalogy between supervenience in the theory of mind and the psychological supervenience of normative commitments on interests. There may be no necessary connection in the latter. For example, we might suppose in many cases that people have learned to have normative reactions to particular matters and that this is their first motivation in many actual instances.

QUESTION: Potential disanalogy between mental supervenience and normative supervenience — whether normative reactions are learned rather than necessitated by interests is worth pursuing.

Surprisingly many people have a quasi-Platonic vision of trust. They know what it is. Unfortunately, because they know it to be so many different things the concept is so plastic that we cannot even know what is at issue in many accounts that explain trust or use trust to explain some behavior. If our discussions of trust are to be understood, we must specify more narrowly how we mean to use the term. In this book, I canvass numerous claims about trust, what it is and how it works. I put forward one fairly general model of trust—trust as encapsulated interest—that seems to be what is commonly at stake in many of the interpersonal relationships we have. In particular, it fits the fact that most of our trust relationships are ongoing, not oneshot, relationships and with people we know relatively well, not with strangers.

DEFINE: The author distinguishes the plastic, quasi-Platonic lay sense of trust from a narrowly specified model: trust as encapsulated interest, applied to ongoing relationships with known persons rather than one-shot dealings with strangers.

In the encapsulated-interest account, it is the reason for the expectations that defines trust. If I expect you to fulfill my trust because I think you encapsulate my interests to some extent in your own, then I can be said to trust you. If I expect you to fulfill my trust because you would act in the relevant way even if I were not in the picture, then I have expectations but not trust as encapsulated interest. Often, as in mutual trust in an ongoing exchange relationship, the encapsulated interest is simply the interest the trusted has to continue the relationship, because that relationship is valuable to the trusted. Hence the trusted's trustworthiness is not a general characteristic of the trusted but is specifically related to the truster. It is relational, often perhaps even wholly relational.

XREF: Connects to game theory and repeated-game cooperation theory, where relationship value sustains cooperation rather than intrinsic virtue.

It is a peculiar implication of the thesis that trust is inherently moral, and of much of the current literature on the need for trust in society, that distrust must evidently therefore be bad. But, distrust is sometimes the only credible implication of the evidence. Indeed, distrust is sometimes not merely a rational assessment but it is also benign, in that it protects against harms rather that causing them. For example, parents who do not entrust the safety of their children to unworthy caretakers, or international institutions that do not entrust the welfare of the Cambodian people to Pol Pot, are acting morally according to almost any conception of morality. Although one might argue that acting as though one trusted some person would be a moral, praiseworthy act in its own right (see Mansbridge 1999), it would surely not be in such cases. Because distrust is benign whenever it is justified, it seems implausible that trust is inherently a moral notion

DEFINE: Clarifies that distrust is not inherently opposed to morality — it is morally neutral or positive when justified.

One might think of distrust as the complement of trust, so that if I do not trust you with respect to something, I distrust you. As is implied in the opening paragraph of this chapter, I treat it here rather as the negative of trust. If I trust you, I have specific grounds for the trust. In parallel, if I distrust you, I have specific grounds for the distrust. I could be in a state of such ignorance about you, however, that I neither trust nor distrust you. I may therefore be wary of you until I have better information on you. … Sometimes, I can create the grounds for trust by giving you incentive to be trustworthy.

DEFINE: Distinguishes the negative (opposition) sense of distrust from the complement (simply lacking trust) sense, allowing for a third state of wary ignorance.

As Coleman (1990, 101) observes, misplaced trust typically entails a large loss, while forgone trust entails only a small loss. For example, the lieutenant colonel lost 4,500 rubles—an enormous sum of money at that time—on one interaction in a long series, in which the total benefits from many of the previous interactions might even have been less than this one loss. If his benefit from each successful deal was 300 rubles, it would have taken fifteen successful deals to make up for his one unsuccessful deal. Distrust may therefore come more quickly to us than trust. Moreover, the slightest distrust or even a slight degree of trust can recommend against risking cooperation with someone if the potential loss is typically greater than the potential gain.

XREF: Connects to loss aversion and negativity bias findings in behavioral economics and psychology.

Unfortunately, forgone trust can entail enormous losses if it blocks establishing a longer-term relationship … Distrust produces an aggregate of lost opportunities, each one regular and predictable. Trust leads to an aggregate of some real losses plus some real gains. Overall, the gains from acting on trust may far outweigh the losses from doing so, because the gains from any given relationship are repeated over and over again while the loss in any relationship tends to end because we correct our misjudgment that the other was trustworthy and we take no further risks on an untrustworthy person. Hence the gains from trust can far outweigh the savings from distrust, as they typically do in many groups or societies, especially including prosperous societies. In deciding whether to take a risk on another's possible trustworthiness, therefore, we should look to the potential gain from a longer run of repeated interaction and weigh that gain against the loss from a single risk of cooperation. Because of this asymmetry between potential losses and gains, we can further suppose that it is relatively easier to flip from trust to distrust than to create trust where there has been distrust for very good reasons of the risks at stake.

Turn now to epistemological asymmetries. Distrust comes easily because it can be built on a limited bit of behavior by the distrusted. Trust, however, requires too rich an understanding of the other's incentives for it to come easily to many people. In a world in which there are many potential partners in cooperation for various purposes, the standard, relatively friendly tit-for-tat strategy might be too generous. In that strategy, I cooperate with you if you cooperated on our last round, and I defect if you defected on our last round. If we both start out by cooperating and we both follow the tit-for-tat strategy, we may readily establish a long-term cooperative relationship. Just because we have many potential partners available, however, we might often use the more draconian trigger strategy of dropping efforts with anyone who fails an initial attempt at cooperation. One failure at the outset of our potential relationship and you are out of

It would be impossible for individuals in their daily interactions to escape the suspicion of distrust, because that suspicion is inherently well grounded. The slogan, “Every betrayal begins with trust,” suggests not only that it is principally when we trust or rely on someone that they can betray us but also the sense that we cannot be sure of another. There are at least four reasons for this. First, the other might change between the time we place our trust and when the other must act on it. Second, people can and often do strategically misrepresent themselves in order to gain advantage.

DEFINE: Explains the claim 'Every betrayal begins with trust' and unpacks why distrust is well-grounded, offering a conceptual framework for trust's fragility.

Fourth, and even more fundamentally, as Georg Simmel (1950, 311– 12) observes, “All we communicate to another individual by means of words or perhaps in another fashion—even the most subjective, impulsive, intimate matters—is a selection from that psychological-real whole whose absolutely exact report (absolutely exact in terms of content and sequence) would drive everybody into the insane asylum.”

DEFINE: Simmel's 'selection' concept — that communication is inherently reductive, framing the boundary between what can and cannot be conveyed.

Finally, in many contexts, it seems only reasonable that the underlying messiness of an institution’s deliberations should be concealed lest it lead to conflict and complications. For example, a jury should arguably deliberate outside the public eye. The National Academy of Sciences in the United States was recently burdened by the general requirement of public openness under the 1972 Federal Advisory Committee Act. Such openness would make it, in the view of many, virtually impossible to reach reasonable judgments about scientific matters of public importance that affect specific interests.

XREF: Connects to debates about transparency vs. effectiveness in governance, jury secrecy, and deliberative democracy theory.

The shambling, slightly dishonest nature of the niceties of private discourse and the concealment of official debates behind closed doors may actually contribute on balance to the possibilities of trusting relevant individuals and groups. In fact, your willingness to say polite things to me, whom you may dislike or otherwise harshly judge, may be a clue to how committed you are likely to be to fulfilling some agreement we make. By covering your feelings, you focus on the benefits from dealing with me, and you give yourself reason for trustworthiness in our dealings.

Distrust in a world in which others are untrustworthy does, of course, protect one against losses that would follow from taking the risk of cooperating with others. But it can wreck one’s own opportunities in a society or context in which others are generally trustworthy. The meaningful result of trust, when it is justified, is to enable cooperation; the result of distrust is to block even the attempt at cooperation. Trust is functional in a world in which trust pays off; distrust is functional in a world in which trust does not pay off.

XREF: Connects to the reader's understanding of social capital, game theory (prisoner's dilemma), and how individual risk-aversion interacts with collective trust environments.

Indeed, it seems unlikely that Barth's accounts tell the whole stories of trust in these two societies. For example, because there was presumably no good legal system covering the vast regions in which they traded, the Omani must have been able to trust one another in their commercial dealings; otherwise they could not have been such successful traders. Their society might have been organized to secure trust in some areas while lacking it in many other areas. In this, their pattern was not so different from much of commercial relations in many other societies or of various other relations. We trust one another with respect to some matters but not with respect to all.

Data from the General Social Survey in the United States for the years 1972 to 1994 show that blacks have much lower levels of socalled generalized trust than do others in the society (Patterson 1999, 175, 190–91). This could follow for various reasons. Life in poor ghettos and in broken families most likely tends to offer less trustworthy relationships than does life in more prosperous conditions. If these conditions produce little experience of trustworthiness in early life, they might lead to a pervasive distrust that is hard to overcome through later experience (see further chapter 5). In addition, blacks individually have proportionately far more interactions with whites than whites have with blacks. This follows from the simple mathematics of the differences in the sizes of the two groups. (Suppose there are eight whites for every black and that on average each black has eighty interactions with whites every month. Then on average each white has only ten interactions with blacks every month.)

XREF: Group-size asymmetry logic parallels other contexts where minority groups have more intergroup contact than majority groups—relevant to networks and trust formation.

Perhaps the greatest deliberate experiment in pervasive distrust was that of Joseph Stalin's Soviet Union during the purge years of the 1930s. Although the apparatus of those years was not fully dismantled soon thereafter, the war with Germany required enough cohesion to make such deep distrust harmful even to Stalin's personal rule. Pol Pot's rule of Cambodia similarly depended on radical use of distrust. In general, pervasive distrust is more quietly used than it was in these two dreadful cases.

XREF: Connects to general knowledge about authoritarian regimes using surveillance and mistrust to maintain control.

There may also be systematic distrust that sustains a destructive equilibrium. For example, familial trust can turn exclusionary, so that outsiders may become actively distrusted. Various studies of impoverished peasant subsistence communities in southern Italy, Mexico, and Peru suggest that people in such communities sometimes develop what the political scientist Edward Banfield (1958) calls amoral familism (also see Aguilar 1984; Westacott and Williams 1976; and, for a brief survey, Govier 1997). Family members stick together and both lie to and distrust others.

People in these societies therefore fail to cooperate, in part because it would do no good in many contexts and in part because they naturally focus on the competitive, zero-sum aspects of their lives. If there is no point in cooperation, there may even be advantage in deceit, secrecy, and cheating, as there typically is in a zero-sum conflict. Of course, if everyone outside my family is likely to cheat me, I have good reason not to trust them.

Such perverse equilibrium states might be extremely stable, unless there is finally intervention from outside. They are stable because anyone who tries to be cooperative is at risk of being exploited by others. To move away from pervasive untrustworthiness, and hence distrust, requires moves by several or many at once and cannot easily be started by a single member of the community.

XREF: Connects to collective action problems and game theory (prisoner's dilemma, tipping points) and to literature on social capital and trust collapse. SEED: Could be an article angle on why trust breaks down in organizations or online communities, and what triggers the coordinated shift back to trust.

That we might not trust those who have power over us, especially when they have little reason to care for us individually, is no surprise. I depend heavily on your favor, while you depend not at all on mine. You can therefore do me substantial harm, while I can do you little or none. The mutual trust that depends on reciprocal relations cannot easily develop in such unequal, nonreciprocal contexts.

XREF: Connects to research on power asymmetries in organizations, institutions, and principal-agent problems where dependence is one-directional.

There is a further problem in power relations. The subordinates may come actively to distrust one another. For example, courtiers are commonly in contest with other courtiers over who has the greatest favor of the powerful king. The French court in Versailles elevated vacuity to its high status in self-defense against the harms that could befall anyone who ventured to say anything with real content that might be reported to the king.

XREF: Connects to the broader discussion of power relations and how subordinates under authoritarian control develop coping mechanisms like self-censorship.

In general, therefore, the weaker party cannot trust the more powerful much at all. Inequalities of power therefore commonly block the possibility of trust. One might mistakenly suppose that a powerful Henry VIII is trustworthy, but this supposition cannot be grounded in a belief that a mere courtier's interests are encapsulated in Henry's interests, at least not over the long run.

Moreover, in the context of great differences in personal power, distrust is malignant. I cannot trust the egoistic Henry VIII, and therefore I cannot trust anyone else at court much either, because they might use anything I say or do to undercut me with Henry, to my great peril. Indeed, I may merely not trust Henry while actively distrusting my fellow courtiers. Henry need have no competitive interest with me, but my fellow courtiers do.

Annette Baier (1986, 247) argues that “voluntary agreement, and trust in others to keep their agreements, must be moved from the center to the moral periphery, once servants, ex-slaves, and women are taken seriously as moral subjects and agents.” Great power differences undercut the very possibility of agreement that is voluntary and uncoerced. For example, Immanuel Kant's kingdom of ends—that is, rational beings—makes little sense where there are gross inequalities.

XREF: Connects to critiques of Rawlsian and Kantian contractarian ethics from feminist and critical-theory perspectives, and to discussions of coercion in unequal power relations.

Her most important criticism of the idea is the conspicuous lack of participation by women in the social contract of any past time and, arguably, even of today, but her criticisms are more general than this in that, with Baier, she thinks much of moral life is grounded in the assumption of agreement. Agreement between those of unequal power is inherently suspect.

For example, the structure of prisons and the behavior of prison guards often provoke distrust between prisoners. Among the most obvious devices, setting up the cells so that any one prisoner cannot see any other but every one of them can hear the noise of every other creates tensions and hostility between individual prisoners. They can become virtual enemies rather than allies who might pose a common front against the guards of the system. Some of what critics think of as unnecessarily punitive and insulting treatment is functional in provoking interprisoner distrust and hence in blocking group mobilization.

QUESTION: This raises the question of whether such architectural tactics reflect deliberate design or emergent dynamics within penal systems.

Not having confidence in the state leads to not trusting other individuals in many contexts, because the state cannot be relied upon to prevent the worst possible outcomes from various joint endeavors and contractual relations. Rather than risk such outcomes, individuals may refrain from cooperating at all. Hence by wrecking the conditions for trust between pairs of individuals or families, the Spanish could substantially reduce the prospects of cooperative endeavors to oppose or hinder their rule (Pagden 1988). In this, Spanish Italy and contemporary American prisons have much in common.

XREF: Echoes the mechanism of prison social dynamics and fractured trust under unpredictable authority; parallels Hobbesian and game-theoretic accounts of coordination collapse.

In such contexts, absence of conditions for trust may breed distrust through paranoid cognition (Kramer 1994). Economic backwardness in these polities exacerbated conflict because it gave little opportunity for advancement through productive endeavor. Economic stasis is sufficient to produce such positional conflicts, which can become lethal, as has occurred in Rwanda and elsewhere in recent times.

Under these conditions, one might expect two general responses. First, people turn inward to concentrate on relations with those with whom they have rich enough ongoing relations to establish trust and trustworthiness. Hence social and economic organization tends toward familism or even Banfield’s (1958) amoral familism. Second, alternative organizations for social control might arise in the relatively anarchic vacuum left by the Spanish neglect of normal governmental functions. Thus we might expect to see the rise of such organizations as the Mafia (Blok 1974, 89–102).

Unfortunately for Sicilians, members of the Italian Mafia themselves engaged in the promotion and selective exploitation of distrust. In this, they mirrored the devices of the Spanish overlords of southern Italy in the seventeenth century (Gambetta 1988, 159). Whereas the Spanish merely suppressed Sicily, the Mafia were parasites who exploited Sicilians. The Mafia system survived changing governments into the twentieth century for at least two reasons. First, out-migration of those hostile to the Mafia left less opposition to it. Second, in the early, weak days of democracy, participants in the political system dragged the Mafia into politics as an ally.

XREF: Connects to discussions of how criminal organizations entrench themselves as parasitic alternatives to weak state institutions.

Breton and Wintrobe (1982, 140–46) present an analysis of the organization of the Japanese large firm that suggests the sophisticated use of barriers to the “wrong” kinds of trust, wrong in the sense that they are dysfunctional for the firm. … In such a system, individual opportunity does not depend on group success so much as on individual differentiation from the group. Hence ties within the group of workers on the line are overlaid by ties with those above that must often be more compelling to many workers than are ties with other workers in general. Indeed, many workers must develop relationships of trust with their superiors, with whom they have reciprocal relationships, while their relationships with nominal peers are competitive and perhaps even distrusting. Weakening ties among peers works against their succeeding in group action against their employers.

QUESTION: This challenges the familiar picture of Japanese firms as harmony-driven collectives; worth exploring whether this vertical-trust model generalizes or is an outlier.

The prison, Mafia, and Habsburg overlords and to a lesser degree perhaps the managers and owners of Japanese firms all benefit from the same structural failure of their subjects. Their subjects do not cooperate when it would be mutually beneficial to do so because they lack the seeds of trust and the institutional backing that would allow them to risk reliance on one another. As Hobbes (1968 , 1.13: 184–85) argues of his state of nature, they may even become preemptively hostile and untrustworthy—indeed, they rationally ought to do so for the sake of their own survival. Furthermore, they compete in harmful ways, taking advantage of one another because, unable to cooperate for mutual benefit and unable to protect what they have, they struggle over shares of what exists rather than producing more to use in trade.

Gambetta (1988) calls these behaviors the conditions for the Mafia’s existence. They are conditions that Mafiosi, prison designers and guards, political overlords, factory owners, and others can manipulate and contrive to gain or maintain control. The Habsburg rulers and prison overlords have power that they can use to create these conditions so that their power will not be challenged by their wards. Because Habsburg rule produced these conditions, the Mafia was able to rise to power.

XREF: Connects to Gambetta's work on the Sicilian Mafia and the economic/social conditions enabling organized crime.

A central value of contract and even criminal law is to reduce the likelihood of severe losses from cheating, such as the lieutenant colonel's loss of 4,500 rubles. Had he been in a contractual relation with Trifonov, he would not have been completely protected, but the asymmetry between his potential gains and losses would have been much less acute.

The most important institutional alternatives to trust in our lives are government and the market, but there are many others, including various institutions and customary devices. Indeed, one could arguably go far toward explaining various institutional structures by showing how they obviate the need for acting as though we trust, at least when doing so involves risk of great loss.

XREF: Connects to discussions of trust economics (e.g., Fukuyama) and institutional design replacing personal reliance.

Much of our ability to trust others on ordinary matters of modest scope depends on having institutions in place that block especially destructive implications of untrustworthiness. Indeed, this is, in different vocabulary, Hobbes' argument for government and submission to it (Hobbes 1968 , 2.30: 376–94). If the potential losses of interacting with others are always high, our interactions are like prisoner's dilemmas in which the loss payoff for one who cooperates while the other defects is very large, so that the prospective loss trumps the gains even of several successful cooperations in repeated play. In such games, we may not ever be able to initiate joint cooperation and may therefore never enter into repeated play.

XREF: Extends the Hobbesian social contract logic into a game-theoretic framing; connects to trust literature and institutional economics. QUESTION: Whether this framework actually explains how real trust gets built versus merely constrains betrayal.

Government generally protects us against the worst that might happen so that we may take risks on modest cooperative ventures. Even while we are often wary of government and its agents, we rely on them to reduce the need for trustworthiness in many realms that government regulates or otherwise oversees. We rely on contract law and court enforcement to achieve successful cooperation in contexts in which, without such protective institutions, we would not risk cooperating with others.

XREF: Connects to trust literature and institutional economics — how formal institutions lower the need for interpersonal trust (e.g., Fukuyama, North). QUESTION: What are the downsides when protective institutions erode interpersonal trust rather than merely substituting for it?

Trust is not, as altruism may be, a motivation that we must use sparingly because we have limited amounts of it. It is a cognitive judgment that depends on perceived facts. It may or may not be in short supply, but using it—that is to say, acting on it—does not reduce the supply. In some contexts people can trust one another across broad ranges of issues. Indeed, establishing trust in one area might lead to expanding it to other areas by generating diversified or higher levels of trustworthiness. Hence trust can build on trust.

XREF: Contrasts with altruism, which is treated as a depletable motivation — connects with any resource-scarcity framing of social capital.

In many discussions of trust the authors assume a benign situation or society. Under this assumption, they can argue that trust is a good thing in the sense that it makes society work better. Clearly, this background assumption may be violated, and it often is.

One might, in such a context, be able to develop a small number of intensive relationships in which trust is justified, but even then, as in Stalin's cruelest years, the people one trusts may finally see it as in their interest to abuse the trust or perhaps to violate it, for example, in order to protect loved ones. … This raises the question of how to create a stable order in which trustworthiness prevails and those who are trusting therefore prosper.

XREF: Connects to work on institutional design, credibility, and how social orders incentivize trustworthiness versus betrayal.

If everyone seems to be relying on the police or the banks or the government, it is relatively safe for me to rely on them too. Still, this may be an unstable relationship that can suddenly tip.

XREF: Connects to trust and social proof dynamics, and to notions of institutional fragility and systemic risk collapses.

My prior experiences with trust may have been so charmed that I optimistically take the risk of cooperating with this new person. On the other hand, they may have been so disastrous that I pessimistically avoid that risk. The new person is no different in the two cases; my prior experiences, unrelated to him or her, are the source of difference. Hence my experience molds my expectations of trustworthiness. If my past experience too heavily represented good or poor grounds for trust, it may now take a long run of contrary experience to correct my initial expectations. Indeed, even my capacities for assessing trustworthiness will reflect a commonsense learning process.

XREF: Connects to schema theory and confirmation bias in cognitive psychology — how prior expectations shape interpretation of new evidence.

In a Bayesian account of knowledge, for example, I make a rough estimate of the truth of some claim—such as that you will be trustworthy under certain conditions—and then I correct my estimate, or “update,” as I obtain new evidence on you. If I take the risk of cooperating with you, I soon have some evidence on whether you are trustworthy in that single context. I might test further and further, updating until I have a good sense of your degree of trustworthiness in various contexts. I might do this—indeed, typically would do it—not necessarily to test you but rather to benefit from cooperating in new ways. Hence trust—the belief in another's trustworthiness—has to be learned, just as any other kind of knowledge must be learned.

DEFINE: Explains Bayesian updating as applied to trust — a conceptual framework for how trust is acquired incrementally rather than given wholesale. XREF: Connects to game theory and iterated cooperation models, where reputation is built through repeated interactions in prisoner's dilemma-type scenarios.

Because general optimism about the trustworthiness of others enables us to enter mutually beneficial relations, we might readily conclude that a utilitarian should encourage such optimism. It does not follow that the utilitarian should be more trusting, however, because a person's degree of trust is determined by the knowledge that the person has. I speak in this chapter of the street-level epistemology of trust, which is the knowledge—good or bad—that a person develops through ordinary life experiences. That knowledge includes the capacity to assess the trustworthiness of others in light of relevant evidence. This epistemology is subjective in the sense that it is a theory of the knowledge of a particular person rather than of the correctness of claims of knowledge in the abstract, which is the usual focus of philosophical epistemology. All that the utilitarian should do is encourage pessimistic distrusters to take risks on cooperating with others up to the level that the utilitarian thinks is justified in the relevant population. In the model discussed in this chapter, one might go somewhat further and say that the utilitarian should encourage people to take somewhat greater risks than what the utilitarian, on present expectations, thinks correct, because the more open person will have greater opportunity to learn from experience than the less open person. Hence erring on the side of optimism is more readily corrected than erring on the side of pessimism.

DEFINE: Introduces 'street-level epistemology of trust' as the subjective knowledge of trustworthiness a person builds through ordinary experience.

Note the nature of the beliefs one must have about another to trust that other. In the encapsulated-interest account, one must know something about the incentives the other has to fulfill the trust (and something about their understanding of those interests, as well). Although valuations of various things are idiosyncratic, we are apt to be fairly adept at assessing someone's interests in many things; thus we might learn to assess trustworthiness that is grounded in the potential trusted's encapsulation of our interests. Trust that is grounded in moral commitments, norms, or bald commitments may be much less generalizable and therefore less easily assessed.

Many market economists have long asserted that eventually the people of the Eastern nations would give up on communist command economics. Interests must eventually trump an ideology that runs counter to interests. Market economists and rabid anti-Communists agreed that the ideology was counter to interests. But the economists had better (more generally applicable) grounds for understanding citizens of communist states than did the American far right because their conclusions were grounded in expectations of rationality, not expectations of irrationality.

XREF: Connects to rational choice theory and interest-based explanations of political change versus ideological/pathological explanations.

In addition, we require a theory of how to act on relevant streetlevel knowledge. I presume here that this theory of decision is a simple commonsense learning theory. You may start with such limited information about me that you can only estimate the likelihood that the typical person in my position would be trustworthy with respect to what you might entrust to me. You might even have such limited information about me that you can only assess from your past experience whether cooperation has paid off in similar circumstances. Suppose it did, and now you risk cooperating with me. (This is not to say you choose to trust me. Rather, gambling on me seems to you the rational thing to do.) You either gain or lose from your cooperation, and this experience is added to your evidence on trustworthiness for future occasions. If I am somehow a new kind of person in your experience, your initial estimate may be unstable, and my behavior might tilt your assessment heavily for or against my kind in future encounters.

The best condition for humans is an environment in which they are fortunate enough to have well-founded confidence (Dunn 1988, 84). This is not an individual-level problem but is, rather, a collective problem. For me to rely on not locking up my home or shop would require that I have confidence or trust in almost everyone. The individual-level problem here is to judge rightly what the collective behavior on trustworthiness is.

If we must develop such a capacity at an early age, two potentially large groups are at a cruel disadvantage: those whose early years are spent in fractured conditions of caprice and neglect, as in the case of many children of American inner-city communities wrecked by poverty, drugs, and broken families, and those who have suffered substantial abuse in their early years from the very persons who might have provided the first experience of trustworthiness.

QUESTION: What specific developmental windows determine whether trust capacity is lost permanently?

On this account, in some societies trust must be beneficial. It would be in the interests of one's children to teach them to be optimistic about others. Trust cannot be produced at will, although it can be willfully instilled, as in children. Moreover, acting as if one trusts can be willed repeatedly so that one may slowly develop optimistic expectations of trustworthiness. In many potentially iterated prisoner's dilemma interactions, one should open with a cooperative move in the hope of engaging the other also to be cooperative. This is not merely a moral injunction. It is a rational claim of self-interest (see discussion of the iterated prisoner's dilemma in chapter 1). One should open with cooperation when the expected long-run benefits of iterated cooperation outweigh the short-term risk of loss.

XREF: This builds directly on iterated prisoner's dilemma strategy from chapter 1, likely echoing Axelrod's tit-for-tat findings on cooperation emerging through self-interest. QUESTION: How is 'rational' trust distinguished from moral trust here? Does the author consider the two reconcilable?

Optimistic assessments can be beneficial only if the general social conditions the optimistic assessor faces are relatively favorable, so that statistically acting on this optimism will be rewarded by trustworthiness. The huge genre of post-apocalypse films and novels of our time portray conditions in which trust is generally not justified. Life is impoverished beyond measure, with the demands of survival and struggle preempting almost all else. To have optimistic assessments in these conditions might be to risk suicide.

XREF: Connects to evolutionary psychology debates about whether optimistic biases are universally adaptive or context-dependent.

The failure of early investment by my parents and others need not correlate with the untrustworthiness of my associates in later life, but it might. The very fact that I have a hard time trusting even those who would turn out to be trustworthy may mean I fail to establish ongoing cooperative relations with such people and therefore disproportionately face short-term relations with people who, on average, are less trustworthy, thus reinforcing my attitude of distrust or wariness. Trusters and the trustworthy may interact chiefly with each

Because I have such low expectations, however, I am willing to test very few interactions. If you had a generally good experience of the benefits from trusting, you would readily enter into far more of these interactions. It therefore would take me longer to gain enough information to recommend changing my pessimistic assessments. All the while, I also enter fewer interactions and therefore benefit less than you do as you enter many, which, on average, pay off well. If we start with similar levels of welfare, you soon outdistance me.

XREF: This is a classic trust dynamic akin to optimism/trust research and network effects — the pessimistic person gains less information to update beliefs, locking in lower trust.

Suppose, on the other hand, that I started life with such a charmed existence that I now am too optimistic about trusting others, so that I often overdo it and get burned. Because I am optimistic, I enter into many interactions, and I coincidentally, therefore, collect information for revising my estimates of the trustworthiness of others very quickly. My aggregate experience soon approaches the aggregate average, and I reach an optimal level of trust that pays off well in most of my interactions, more than enough to make up for the occasions when I mistakenly overrate the trustworthiness of another. Oddly, therefore, if parents are to err on one side or the other in instilling a belief that others are trustworthy, a simple learning account suggests that they should err on the side of instilling greater optimism in their children than they think is objectively warranted, at least with respect to modest issues of cooperation for which the losses at risk are not major.

QUESTION: Does this learning argument hold only for low-stakes cooperation, or does the risk calculus change for major losses? The author restricts the claim but the boundary case is worth probing.

One reason for stereotyping people is to set a baseline estimate of their trustworthiness in order to get the analysis and its revision of information under way.

The low truster nevertheless loses ground and suffers severe relative welfare losses. An equal opportunity program cannot stop that morose trend. The losses are not merely of opportunities but of the capacity to capitalize on opportunities.

XREF: Connects to debates on equal opportunity vs. capability approaches in welfare economics and theories of social capital.

Consider the condition of an outsider or new immigrant to a community. The outsider may initially seem untrustworthy to others in the community. This could follow merely from a guess, a judgment of that person as less likely to be trustworthy than those who are long well known. Until more knowledge of the newcomer's trustworthiness is generated, he or she is given fewer opportunities to demonstrate trustworthiness. Hence it may take depressingly long for the outsider to become trusted. Since, again, the combination of trusting and being trusted conveys benefits in various kinds of exchange and mutual aid, the outsider faces greater difficulties getting ahead and may begin to seem less able and worthy. Superficially, we might suppose a group that held off from trusting an outsider for a long time was prejudiced in a racist or related sense, although the rational expectations of the wary assessor of trustworthiness might be sufficient to explain the community's attitude.

These loans involved seeming trust by the Latin bankers who made them. Portes and Sensenbrenner call that trust enforceable. Why? Because the Cuban exiles were virtually trapped. They could prosper in this transplanted Cuban community only if they proved reliable in repaying their loans. They could not prosper as well anywhere else. They could not return to Cuba, and there was no welcoming community elsewhere. No other community of people had natural access to their reputations and could have trusted them, in the sense of having good reason to believe that their incentive to be trustworthy was compelling, even overpowering. They were an unusual case: outsiders with almost insider status in one exclusive community on which they were fully dependent. The availability of that status was brief. By 1973, character loans ended, because the newly arriving Cubans were no longer known to the local banking community, and they might also have had little or no recent entrepreneurial success in Cuba on which to ground a reputation.

XREF: Connects to the sociology of social capital and enforceable trust in closed networks.

Sociologists writing on trust are generally concerned with social mechanisms that generate trust (for example, see Luhmann 1980, 95). Luhmann supposes that the structure of trust relations requires that calculation of risk remain latent. Yet contractual relations may require that such calculation be overt and present and may therefore introduce an atmosphere unfavorable to trust. When such interdependence already exists, so that the risks are openly known to all without need of specific present discussion, mutual understanding and trust may be enhanced (Luhmann 1980, 36). This merely means that the transition from informal to formal regulation of relations may be uneasy, as the street-level epistemologist should expect. New conditions that have ill-defined prior expected probabilities generally introduce initial instability, but the result of successfully completing the transition to formal regulation can be to enhance trust.

Coleman (1990, 114) says development of norms with sanctions enhances cooperation. This is also Hobbes' theory: Creation of strong sanctions to protect each from the others makes all better off. In Hobbes' account of the need for a powerful state, what is mainly needed for most people is merely enough security to be able to enter into exchange relations with others without fear of being killed for what one has or merely preemptively to prevent one from stealing from or killing another (Hobbes 1968 , 1.13: 184; Hardin 1991b). Trustworthiness is underwritten by a government that enforces con-

XREF: Connects the Hobbesian state-of-nature as a security enabler to game-theoretic/Coasean framings of trust and cooperation that appear elsewhere in this kind of literature.

The incentive to cooperate (or to fulfill a trust) and the sanction not to do so are one and the same. They are the benefits of future interaction that ride on present trustworthiness.

DEFINE: Clarifies that cooperation incentives and sanctions are the same thing — the value of future interactions contingent on being trustworthy now.

Most parties to Hambros loans probably prove to be fully trustworthy to repay. The background possibility of sanction greatly enhances this trustworthiness, however, and justifies Hambros's general risk taking. We might say that the use of courts is the centrally important consideration to Hambros in making any particular loan, but such a statement is likely to be misleading. What is centrally important is that the existence of the court sanction changes relations drastically across the board, making Hambros's relations with virtually all potential borrowers more cooperative than they would otherwise be.

In many accounts of trust, including Coleman's (1990, chapter 5) strictly rational account, one might suppose those who are to trust are all interchangeable in the following sense: given the same incentives (potential objective payoffs), we would all trust to the same degree. There is an important prior element, however, that some might think of as psychological: we may have different capacities for trust. This issue should not be treated as psychological in the sense of irrational or not rationally justifiable but, rather, as essentially epistemological and hence as pragmatically rational.

DEFINE: Reframes differing trust capacities as epistemological and pragmatically rational rather than irrational psychological variation, challenging the interchangeable-agent assumption in rational accounts like Coleman's.

A general problem with inductive knowledge, if it is completely atheoretical, is that it cannot justify a claim that what has always happened before must happen again. Most of us are willing to live with inferences that various things will continue to happen in the ways they have always happened so far.

XREF: Connects to Hume's problem of induction and the is-ought/induction problem in philosophy of science.

To break the hold of a bad and misleading past, the inductivist requires a lot of new experience or a bit of theory that runs counter to prior experience. … Understanding that others will be trustworthy when their incentives are right, as in the encapsulated-interest account of trust, may hold greater, quicker promise for grasping those opportunities. This requires seeing the choices of others from their perspective to comprehend their incentives. Then trust becomes fully strategic.

It follows that I will have to make choices with less information to go on than other, more optimistic people have. Although there might be specific instances in which information is deleterious to my well-being, it is generally beneficial to have more information about one's circumstances. It is not an intention, but merely an effect, of distrust that it can put one at a relative disadvantage on this score. Hence in a society with pockets of distrust, such as in hostile ghettos or in the cases of individuals with particularly bad experience of misplaced or undeveloped trust, some people will have to act from crippled knowledge.

In a relatively benign society, we can afford to take chances with people in general until we discover any of them not to be worthy of cooperation. If, in contrast, we had come to be distrusting in general, the result in a benign society would surely be far worse than what would follow from taking risks on people.

XREF: Connects to social capital research (Putnam, Fukuyama) on how trust functions as a societal resource, and to Trivers' work on reciprocal altruism.

For many people, such a stance would be too tedious and would cost far more in lost opportunities than it would save in avoided harms. To act according to the backward induction argument against cooperation in the iterated prisoner’s dilemma would not generally be in their interest (see discussion of the iterated prisoner’s dilemma in chapter 1).

XREF: Connects to the iterated prisoner's dilemma discussion in chapter 1 of the same work, and to broader game theory literature on backward induction vs. practical rationality.

Once you have a cluster of associates on whom you can rely, however, it might not seem worth the risk of developing new relationships. At its extreme, this might mean a society of close-knit families closed off to most of the rest of the world, as in the impoverished subsistence economies discussed in chapter 4.

If the account is to fit the encapsulated-interest model, then part of the learning that goes into revising one’s assessment of another’s trustworthiness is learning of the other’s interest in continuing the interaction. It is easy to misread the learning account as grounded in a purely dispositional view of trustworthiness, as discussed in chapter 2. If it is taken as an account of my assessing your capacity to judge your interests in cooperating with me, however, then it is fully consistent with the encapsulated-interest view.

XREF: Relates to the encapsulated-interest account of trust discussed throughout the book, distinguishing it from a purely dispositional view.

The simplest psychological path for such development would merely be a learning model, as discussed in chapter 5. The more I encounter people who reciprocate my cooperative gestures, the more I come to understand the nature of our potentially beneficial interaction, so that I become trustworthy in the sense that I begin to take others' interests into account in deciding what I do in joint undertakings. When furthering their interests furthers mine and I recognize this fact, they will have potential reason to judge me to be trustworthy.

If I have optimistic expectations I will more readily take risks on the trustworthiness of others, and if others in the context are relatively trustworthy, I will benefit greatly from the cooperation that we can achieve. If, however, others are not trustworthy in the context, I may quickly learn that and will protect myself against betrayal by them. If on the contrary I have pessimistic expectations, I will not readily take risks with others. Hence I may not even gain the experience to revise my expectations when I come into a context in which others would commonly be trustworthy.

Trust and trustworthiness often involve a severe commitment problem that may be hard to resolve. A particularly absurd resolution of it is that of Wagner's (1977, 16) The Rhinegold, in which Alberich adopts celibacy and forswears love forever to gain the power of the Rhinegold (see chapter 2). The sniveling Alberich has impossibly great willpower (as perhaps befits the ego of Wagner).

XREF: Connects to commitment device theory in economics and game theory, where costly signals make threats or promises credible. QUESTION: Does the author explore other real-world commitment devices beyond this mythological illustration?

One nonabsurd way to resolve the commitment problem is genetically. One who blushes when lying may be able to establish credibility easily without any willful commitment or manufactured disposition. One who does not blush when lying may have to create a reputation for honesty that is worth more than any gain he or she might make from cheating on that reputation.

DEFINE: Explains how a physiological trait (blushing when lying) functions as a genuine commitment mechanism, contrasted with manufactured reputation-building.

Sentiments of vengeance, guilt, and so forth often incur substantial avoidable costs—hence, Robert Frank (1988, 54, 57) supposes, they must also confer some sort of compensating gain to have persisted socially or genetically. These sentiments will be most effective if they can be properly communicated to deter or encourage relevant actions from others. If devices for communicating, such as blushing, are genetic, we may then explain the genetic selection as the result of interests—it may even develop through intensive cultural conditioning.

XREF: Connects to Robert Frank's work on emotions as commitment devices — the idea that seemingly costly sentiments actually serve strategic social functions.

Frank's model yields four conclusions. First, if cooperators and defectors look alike, then cooperators will be extinguished. Second, if cooperators and defectors are easily identified, defectors will be extinguished. Third, if mimicry has no cost or delay, cooperators will be extinguished. Fourth, if mimicry entails fixed costs of detection, there will be a stable mix of cooperators and defectors.

If sentiments for vengeance, generosity, trustworthiness, and so forth have developed genetically, one may act from them even when it is not in one's interest to do so.

Of course, intermediaries in trust may be more available to some kinds of people than to others, giving the former great advantages over the latter, as was evidently true for me by error, so that, unlike many people stranded without money abroad, I did not have to go to a consulate. Among the many groups that have long had difficulty owing to a lack of relevant intermediaries to help establish their trustworthiness have been wives seeking credit, the self-employed, and new arrivals to a community.

Most people we meet and come to trust may be people we meet through others, who are de facto intermediaries in trust. This fact might make optimism about others plausible in many cases. Experience must still trump if the new person fails our optimistic cooperative gestures.

Perhaps this account of love as leading to risks that, in the presence of trust, would not be risks suggests the wisdom of Jane Austen's eventual marriages between cousins and other longtime but not too close associates, as in the case of Emma Woodhouse and her neighbor, Mr. Knightley.

XREF: Connects to the philosophical account of love, trust, and risk being discussed, and draws a literary parallel to Austen's novels. FICTION: The Austen example and the trust/risk dynamic could inform a story about a marriage or partnership forged through long familiarity rather than passion.

Annette Baier (1985, 58) says that “it is fairly obvious that love . . . involves trust,” but this is not obvious at all. Many people have had the doleful experience of deeply loving someone they did not trust. The pattern of “misplaced” love in some cases recalls a spoof of the standard theme of country and western songs: “I can't love you, baby, if you won't leave me.” Even reciprocal love need not require extensive trust. It seems likely that falling in love leads to experience that justifies trust, so that love commonly begets trust rather than the other way around.

XREF: Challenges a widely assumed conceptual link between love and trust, echoing philosophical discussion of the nature of both emotions.

When I trust you in the sense of believing that your interests encapsulate mine in at least the matter with respect to which I trust you, we can, naturally, be said to share interests to some extent. We might share interests in two quite different ways. First, we may merely share them causally, because what serves your interest serves mine as well. Second, one of us may genuinely adopt the other's interests as our own (see further Hardin 1999c). If I love you, I may literally encapsulate your interests in mine to some extent in the sense that I value your happiness so much that I act as though for your benefit without full regard for my own benefit.

DEFINE: The passage distinguishes two distinct ways interests can be shared: causal overlap versus deliberate adoption of another's interests, as when love leads someone to value another's happiness.

It is this kind of relationship that allows the child to develop at all in the account in chapter 5. In the end the child's independence or autonomy as a person may depend on the parents' capacity to love and give without commensurate reciprocation.

The striking thing about a love relationship, whether it is mutual as between two lovers or more nearly one-way between a parent and child, is that the trusted genuinely encapsulates the interests of the truster in the trusted's own interests. Moreover, this is commonly true ab initio for the mother of a newborn child, and it is very nearly true ab initio for lovers who fall suddenly in love. The lovers need not have known each other for very long, and they need not have a rich array of associates in common to protect them in their risk taking with each other. Their love allows them to escalate the risks they take very fast and relieves them of the need for support from a larger community. Falling in love moves everything ahead much faster, and the thrill of such momentum may even enhance the love just because it is associated with the beloved.

XREF: Connects to the broader trust literature in the book on encapsulation of interests and how communities scaffold risk-taking.

For the ordinary trust relationship, such as in an iterated prisoner's dilemma or exchange or in a trust game interaction over a relatively narrowly defined range of issues, you and I might share limited interests and might not genuinely care much about each other's general welfare, but we can still trust each other because we need each other's contributions to some common effort or production. In this case our interests are weakly shared through the causal effects of our efforts on our joint good. For the lover and the parent, the interests of the other are much more strongly shared: they are directly included in the parent's or lover's own interests to some substantial extent because the lover and the parent care directly about the general welfare of the other. Therefore, the other has strong reason for trusting. The range of matters of concern to the lover or parent is typically very broad; for the ordinary trust relation, it is typically narrowly defined.

Once they have been initiated, trust relationships are maintained in at least two large classes of ways: through direct recognition of the value of the relationship and through indirect feedback, which stimulates continuation or iteration of the reciprocal dealings that constitute the relationship. The mode of direct recognition may be the whole story for many dyadic trust relationships. You and I each know that continued interaction with each other will benefit us. One reason mutual trust seems, on a subjective assessment, to be so commonplace in the range of all trust relationships is that it is so easily maintained and understood. Additionally, as noted in chapter 1, mutual trust gives each party an extra reason for thinking the other trustworthy: that the other benefits from being in the relationship.

Our chief problem is not maintenance in this sense but only the prevention of the collapse of our ongoing interaction in the face of anything that might disrupt it, such as sudden escalation of the stakes that lets one or the other of us treat the next round of our interaction as though it were part of a new, perhaps only one-shot, interaction. In such an interaction, the modest level of trustworthiness that we might have maintained through many iterations might no longer be enough to prevent opportunistic actions

XREF: Connects to game-theoretic work on iterated prisoner's dilemma and Axelrod's findings that repeated interactions sustain cooperation.

Hence, given the condition that enough people are trustworthy, Luhmann's theory is a functional explanation in the demanding sense of this paradigm. What does this tell us? It does not say that because trust is functional it will happen. Rather, it says that if the causal chain producing trust ever gets under way, it will tend to be sustained by the feedback mechanism of the explanation. That fact raises an additional question. How does the causal chain get under way? If others are generally untrustworthy, we can generally expect everyone to learn not to be readily open to cooperative gestures. We could imagine that trusting would be a result of social evolution from a prior, less complex world in which ongoing interactions are dense enough to ground trust, as in the thick-relationships variant of trust.

XREF: Connects to Luhmann's broader systems theory and evolutionary accounts of social trust, as well as Putnam's social capital literature on dense networks fostering trust.

Acting on the incentive for trustworthiness is to be trustworthy on the encapsulated-interest account of trust. In his functional explanation of the maintenance of trust in a complex society, Luhmann does not make the direct mistake of confusing trust and trustworthiness. Nevertheless, it is much easier to account directly for trustworthiness, which then begets trust. It begets trust because trust is essentially in the category of knowledge, and evidence of trustworthiness ultimately defines trust.

XREF: Connects to the encapsulated-interest account of trust from game-theoretic and social capital literature; the distinction between trust and trustworthiness recurs across the trust literature.

On this account, trustworthiness is, in a sense, prior. Hence there is reason to suppose that it is the rise of trustworthiness that allows for the development of complexity, which actually results from successful trust. Trustworthiness not only enables us to handle complexity when we have it, but it also therefore enables us to develop complexity.

XREF: Connects to theories of social capital and institutional trust as prerequisites for economic and organizational complexity.

In general, functional explanation fits especially well with rationalchoice understandings because the feedback can work through the creation of incentives for acting according to the pattern of behavior that is to be explained (see further Hardin 1980). The relevant functional pattern of behavior commonly just is a response to incentives. No one need know the general implications of everyone’s acting from those incentives.

Other issues in the management of trust could be addressed. For example, we might suppose that trust relationships develop in one arena in response to distrust in another or others. If the larger society is a place of distrust or the government is a source of threat, we might develop stronger local relationships to protect ourselves against intrusions from the larger, malevolent society or government. The threatening world of the Omani (briefly discussed in chapter 4) leads to strongly maintained familial homes that are virtual fortresses, with high walls offering protection against the outside. The awful world created by Stalin led later to the rich underground culture of the samizdat. Pernicious race or ethnic relations might lead a minority group to define its own ghetto.

XREF: This echoes the general principle that external threat consolidates internal cohesion — relates to social identity and conflict theory.

We might also analyze the range of things that can disrupt trust relationships, as the endgame effect in the disreputable dealings of the lieutenant colonel and Trifonov disrupted their relationship. Some relationships seemingly have to escalate to some relatively high level or else founder, and that gives opportunity for breakdowns that mimic endgame effects.

I argue that, in any strong sense of trust, trust in government is not a major consideration in the working of a modern society. A claim to trust government is typically implausible if it is supposed to be analogous to a claim to trust another person.

XREF: Connects to broader literature on institutional vs. interpersonal trust, and to the author's overall argument about how modern societies actually function. QUESTION: What does the author propose as the real basis of modern societal functioning if not trust in government? Worth pursuing in the surrounding argument.

Citizens must often be compliant if government is to work, and they may more readily be compliant if they are confident that government actions will serve their own interests or some broader public good that they support. For example, H. L. A. Hart (1961, 201) argues that the Hobbesian vision of using coercion to motivate obedience to law or to the state depends on the background fact that most people comply willingly, perhaps for normative reasons. It is their compliance that makes it possible for the state to focus its limited resources for coercion on the potentially disobedient. This sounds like a fundamentally important claim. In fact, of course, for most of us most of the time in a benign society, the police generally work in our interest, and compliance is easily motivated. This fact leaves the police free, as Hart notes, to coerce selected others, including minority groups as well as suspected criminals.

What is called trust in government may in fact be something short of trust as we experience it in interpersonal relations. This is confidence, or what we might call quasi trust. It is based on reasons for believing government agents to be trustworthy, reasons that do not involve direct relationships with the quasi trusting citizen.

DEFINE: Distinguishes quasi-trust (confidence based on reasons and indirect evidence) from interpersonal trust built on direct relationships.

A claim that is worthy of investigation is whether a government that depends on extensive reciprocal participation by citizens requires trust to work well. Largely for empirical reasons, I argue that even such a claim cannot be sustained—or, rather, it cannot be sustained if what we mean by trust when we speak of trust in government is conceptually the same as what we mean when we speak of trust in another individual. Often, all that is needed for government to work is for citizens not actively to distrust it.

QUESTION: What empirical evidence supports the claim that passive non-distrust suffices for government function? The distinction between active trust and passive tolerance is worth pursuing.

To bring trust of government into political theory requires a microlevel account of how government works at the macro level. This must largely be an account of rational expectations of what government and its agents are likely to do. In the encapsulated-interest account, I must know that the agents or the institution act on my behalf because they wish to maintain their relationships with me. That is generally not possible for government and its officials.

XREF: Contrasts the personal, relationship-based encapsulated-interest trust model with impersonal institutional trust in large-scale governance.

Expectations about human behavior are much less reliable than many of the most common inductive expectations about nature. Indeed, their unreliability is the central driving force of most great literature. In a cute moment, one might say that one of the strongest expectations we must have of people in the long run is that they will defy our expectations.

XREF: Connects to the psychology of prediction and narrative theory — the idea that literature thrives on breached expectations resonates with work on reader anticipation and surprise. SEED: A possible essay angle: how unreliable human behavior underpins narrative craft, contrasting scientific induction with literary unpredictability.

Declining faith in religion has historically followed increased understanding of the nature of the world. Declining faith in government, its agents, and various others may similarly be the result of increased understanding or, more likely, increased knowledge of the nature of government. For a trivial example, the kinds of information we had about President Bill Clinton went far beyond what was popularly known about any previous president, including some whose actions as president were arguably more scurrilous than Clinton's. The knowledge we have about many governmental actions is also astonishing in comparison with past times. The civil rights movement may well have been so successful largely because of instant television coverage of the brutality and stupidity of many southern officials (Garrow 1978). The White House tapes drove Richard Nixon out of office when mere testimony probably could not have done so.

XREF: Connects to broader theme of knowledge undermining trust, echoing shifts in religious belief paralleled by political disillusionment. SEED: Could develop into an essay on how transparency and information abundance reshape public trust in governance.

However, other changes—especially the period effect of changing generations—are clearly also important in the sense that they correlate with reduced civic engagement over recent decades. Putnam's data suggest that the generational effect is about twice the effect of television viewing, although these two are not independent. The television generation accounts for between 10 and 15 percent of the total decline in group participation rates over the period from 1965 to 2000 (Putnam 2000, 284).

XREF: Connects to broader Putnam thesis on declining social capital and its competing causal explanations.

The third form of spillover argument is essentially about the creation of a disposition on which people then act even without direct reasoning in particular instances. This is roughly Oliver Williamson's (1993) view, as discussed in chapter 3, that trust is not generally calculative. There is calculation or at least solid reasoning somewhere in the past but not in this moment when I am dealing with you. This raises the question of whether there is calculation when I first meet someone or whether my disposition is applied generally to virtually any and everyone, including those newly met and those never to be met again. To assess whether the dispositional (or spillover) thesis is correct would require psychological data that go beyond the correlations between measures over time of trust in government and participation in group activities. This is a fundamentally important issue on which we have inadequate knowledge.

QUESTION: The author notes we lack adequate psychological data on whether dispositional trust extends to newly met or never-to-be-met individuals — a genuinely open empirical question.

If you have ungrounded faith in me, that might benefit me (but not you) by getting you to do things for me out of misplaced trust. If you have ungrounded faith in our government, that would benefit me only if that government happens to serve my interests fairly well and if, as we may assume, your faith in the government helps to license its actions and reduces the chance of its being successfully challenged by you and similarly placed others.

XREF: Connects to critiques of institutional legitimacy and social contract theory, where consent licenses state power.

The cause, however, might not be Putnam's indirect one of displacement of group activities in favor of time before the television screen, which leads to fewer trusting relationships in group activities, which in turn spills over into less trust in government. The intervening step need not be this claim of spillover but might instead be the claim that citizens, in part because of the visual power of television, now know too much to have confidence in many officials. Television does not tell us as much about politics as our former newspapers did, but it gives us information far more viscerally. Television helped to destroy the credibility of many local politicians in the South during the civil rights era (Garrow 1978), and it helped to destroy the credibility of the war in Vietnam.

XREF: Challenges or offers an alternative to Robert Putnam's 'Bowling Alone' thesis on declining civic trust via television.

To see the decline in faith in government as a result of, in a sense, the decline of citizens is to treat it as a problem of trusting when it should more cogently be seen as a problem of trustworthiness. The differences in what we must explain in these very different visions of the problem are categorical. We would need data on psychological dispositions toward trust and an account of how these work for the Putnam thesis. On the other hand, we would need data on evidence people have to trust or not trust government officials—or to have or not have faith in them—for the thesis that trust is primarily dependent on trustworthiness.

XREF: Contrasts with Robert Putnam's thesis on declining social trust/civic engagement, framing trust as a matter of citizen dispositions rather than institutional performance.

Recall the view cited in the previous section: that having only 5 percent of the citizenry alert and committed to politics is all it takes to make things work. If this is even roughly true, then the survey evidence on declining trust, confidence, or faith in government across the general population is of little interest. What we need to know if we are to assess the prospects for government in this era is how the 5 percent who are alert and active in politics rate government.

QUESTION: The author concedes this 5% figure may only be 'roughly true' — worth probing what empirical support exists for this threshold.

Ordinarily, I am likely to distrust you if I believe your interests strongly conflict with mine. (My distrust might make no difference for any action of mine, however, because I might have no interaction with you.) Often, this is the position we are in with respect to government officials. We can imagine that their interests are not clearly ours. Trust and distrust of government and its agents may therefore be asymmetric. We may have knowledge and theory to distrust when it would be hard to have knowledge or theory to trust. From a Humean or Madisonian view of the corrupting influence of having power with the discretion to use it, one can theoretically distrust government officials in principle as sometimes likely to use their offices for personal benefit in ways that conflict with the public interest or any citizen's individual interest. It is not so sensible to argue for a converse general principle that such officials are likely to share my interests and therefore to serve them.

Yet it would be wrong to say that Gorbachev actually succeeded in generating trust in government in the Soviet Union or its successor states. All he did was greatly diminish grounds for distrust in certain contexts, by both Soviet citizens and the West. He did so by disabling the Soviet government.

Consider a more extreme grounding for distrust in government. For an easy case, consider an individual who has been abused by government in the past, as were the subjects of the notorious Tuskegee experiments designed to watch the progress of syphilis in untreated black southern prisoners, or the subjects of supposedly harmless exposure to radiation in various experiments by U.S. government agencies to find out what effects such exposure might have (for the latter, see D’Antonio 1997, 41–42). Such an individual has good personal reason to be wary of distant government officials, indeed to distrust them in a relatively unspecific and vague way that might have no counterpart in trust.

XREF: Connects to the Tuskegee syphilis study and Cold War radiation experiments as historically documented cases of state harm against citizens.

This dispiriting conclusion might be exacerbated by an unfortunate characteristic of many government policies, including many of the best understood and most important policies on economic benefits. If the populace is distributed along some rough continuum of preferences with respect to some policy arena, then any specific policy that is adopted will be very near the positions of only a fraction of the populace. Among those who are at all aware of policies, then, most people might think that they fare relatively poorly from any given policy. Hence they might not even reach the inductive generalization that an actual government does serve their interests at all well.

In sum, it makes far readier sense to distrust government than to trust it. The kinds of understanding necessary for trusting government are almost logically ruled out for typical citizens, while the kinds necessary for distrusting it are commonplace and resonant with ordinary life experience.

QUESTION: This flips conventional civic framing that sees trust in institutions as default virtue. Worth exploring the empirical basis for why trust-making knowledge is so constrained.

When distrust in government is endemic, as in the Eastern European and Soviet worlds at the end of the 1980s, there may be no better move than to weaken government substantially (Hardin 1999d, chapters 5 and 6). Elimination of agencies and powerful bureaus and bureaucrats will eliminate the objects of distrust. “Strong” leadership is precisely what is not wanted when strong leaders have been the problem. Weak leaders unable to intervene capriciously are what is wanted to make the society develop successful trusting relationships.

As is commonly true, they seem to have come to view their interests as tied to the interests of their organization rather than to the interests of their clientele. This can happen simply because it is within the organization that they are rewarded, so that it is their organization and not their clientele who give them their incentives for action or inaction. This problem is exacerbated when there is the credible threat that some of these institutions will be abolished. Hence I cannot trust the agents—because I believe their interests do not encapsulate mine.

At the most extreme change in government, as after a major social revolution, the problem of establishing stable expectations may make reducing distrust nearly impossible for a while. Tocqueville (1955 , 176) notes the seeming paradox that typically a revolution intended to improve things initially makes things worse. Many might have faith in the new regime because they think it represents their interests, and they might therefore say they trust the regime. They might continue to have such faith even after arduous years of failure, as in the Stalinist years in the Soviet Union, when the virtual deification of Stalin was evidently effective in many parts of the population. This would not be grounded trust, although it might be an unusual case of what some philosophers call trust as blind faith.

XREF: Connects to Tocqueville's observation about revolutions making things worse before better, a classic paradox in political theory.

There are, of course, many issues for which politicians do not face such a problem because the distribution of preferences will not be normal. For example, preferences on abortion policy and gun control in the contemporary United States are partly bimodal, with fairly sharply defined pro and con positions. In the face of the absolutist stance of some opponents of any abortion for any reason, the arguments of such writers as the lawyers Ronald Dworkin (1993) and Laurence Tribe (1990) that the two sides are not so far apart are prima facie unconvincing

XREF: Connects to public opinion and political polarization research; the normality assumption underlies median voter theorem and Downsian spatial models.

For issues that have a more or less normal distribution of preferences, on the other hand, staking oneself to a precise position is tantamount to putting distance between oneself and most voters. The task of gaining the trust of constituents, then, is complicated by the virtual certainty that effectiveness in office will correlate strongly with disappointing or even offending large numbers of constituents on particular issues. Dwight Eisenhower had the nearly unique advantage of being elected by a populace ignorant of any of his views (if he had any), and he did a fairly good job of maintaining such vagueness through eight years in office. Ronald Reagan, the best president the American right has ever had, was reviled by much of that right soon after he entered office. Franklin Roosevelt campaigned as a fiscal conservative and then soon abandoned the policy as president. Thomas Jefferson, among the most committed of democrats, bought Louisiana from the French without constitutional authority or congressional approval.

XREF: Connects to the median voter theorem and spatial models of electoral competition.

Low voter turnouts in many nations, including, notoriously, the United States, are commonly taken as evidence that government has failed to elicit support. Prima facie, an equally or even more plausible conclusion may be that such turnouts are evidence that government has not engendered grievous distrust and opposition. Silence cannot unambiguously prove the case for or against government. If the crippled epistemology of mimicry underpins our expectations of government, then the limited commitment of most people to try to change or affect government makes epistemologically good sense.

XREF: Passing familiarity with political science debates about participation and legitimacy — this flips the standard 'low turnout = low legitimacy' equation.

In speaking of government, John Dunn (1988, 90) supposes we should choose Bernard Barber's (1983) trust in capacity, rather than trust in good intentions, in large-scale politics. (Throughout this book, I assume that both elements must be met for trust, although, as noted at the outset, I have not constantly addressed the potential problems of lack of capacity.) This “trust in capacity” is a substantially different meaning from the usual sense of “trust” in interpersonal relations, where it includes concern with both competence and intentions.

It seems true that, when people say they trust government or the president or some other officeholder, they typically mean something different from what they mean when they say they trust a friend. Maybe what they mean is merely that the government or officeholder has the capacity to do the job well. Perhaps it includes an element of the intentions of the trusted, but only an inductive sense of these and not a sense that the government or its officials specifically care about any citizen's interests. If trust reduces merely to this inductive expectation, however, it loses its usual positive valence. In this relatively hollow expectations sense, I can trust an official to act against my interests or to follow policies I oppose.

DEFINE: Distinguishes institutional trust (an inductive expectation of competence) from interpersonal trust (which carries positive valence and care for one's interests).

If capacity is the measure of trust, we might have to say we trusted Richard Nixon more than most presidents, but that seems to be a quite unlikely claim in the vernacular. Perhaps Europeans, who were not subject to Nixon's tax audits, FBI investigations, and election shenanigans, could give most weight to his capacities in attempting to reduce the scale of international conflict in at least his China policies. Many Americans would not so easily have slighted the political abuses, which seemed to be driven not by concern with his better policies but by personal paranoia.

QUESTION: Does trust hinge on capacity or on moral/political conduct? The passage suggests these diverge sharply in Nixon's case.

In a poll reported in Business Week (17 May 1999, 8), people responded that, when looking for advice in assessing some product to be purchased, they most trust Consumer Reports, a friend's recommendation, a news article (not an ad), and a magazine article, in that order. This is presumably what trust in government often means in the survey literature. “I trust” for some of these responses means no more than that I can rely on government or I can expect it to act in decent ways. “I trust” Consumer Reports does not mean what “I trust” my close friend means. It reduces to the French, “j'ai confiance,” I have confidence.

DEFINE: Distinguishes institutional confidence (as in Consumer Reports) from the deeper interpersonal trust reserved for close friends.

Hence we probably should not adopt Dunn’s redefinition of trust for politics as merely about competence. Rather, we should generally speak not of trust in government but only of confidence in it, as Luhmann (1988, 102) argues.

XREF: Distinguishes trust from confidence, echoing Luhmann's framework — connects to broader trust literature in political theory and sociology.

In sum, government and its agents might be genuinely trustworthy in many cases, but most citizens cannot be in a position to know that they are. Hence most citizens cannot be said to trust government in any of the standard senses in which individuals can trust one another. This follows for simple epistemological reasons. Most of us most of the time cannot know enough to trust government (or other large institutional) agents or agencies to judge their fit with any of the standard conceptions of trust.

QUESTION: If citizens can't be positioned to know whether government deserves trust, what legitimate basis can political trust rest on?

Trust is merely one reason for confidence in taking cooperative risks, and trustworthiness is merely one reason such risks can pay off. In the large contemporary literature on trust, the clear point of wanting more trust is that trust eases the way to cooperative social relations (Luhmann 1980; Putnam 1993; Fukuyama 1995; and many others). Similarly, much of recent experimental work on trust is an outgrowth of earlier identical or nearly identical experiments that focused on cooperation. Cooperative relationships constitute a broader and more inclusive category than trusting relationships—generally a much larger category. At the end of the day, therefore, when trust has run out, we still need to explain most cooperation in other ways. A large part of the explanation of such cooperation in the face of obstacles to trust rests in the social structures of norms and institutions that can be explained as devices to get us around such obstacles.

XREF: Connects to institutional economics and social capital literature — Putnam, Fukuyama — where trust is often treated as the primary mechanism of cooperation.

On every account of trust based on assessments of trustworthiness, we obviously face epistemological and time constraints that prevent us from having strong trusting relationships with more than a limited number of people. The trustworthiness can be morally motivated, the result of encapsulated interest, or a matter of character or disposition. The constraint of time is clear enough if we must have ongoing relations with others in order to build trust in them. The epistemological constraints even cut against the possibility that we can trust large numbers of people through their reputations.

Does social order grow out of trust? It might prosper better with widespread trust and trustworthiness, but it does not follow that it must initially be grounded in such trust. Consider the velvet revolutions in Eastern Europe in 1989. Masses coordinated behind the expression of hostility to the prior regimes (see further Sztompka 1996). Distrust must have been endemic in, for example, East Germany at that time, with a large fraction of the population implicated in the STASI (secret police) oversight of citizens at all levels. It was partly distrust that stimulated the quest for a new order. On Piotr Sztompka's (1999, 160–90) account, the new order soon produced higher levels of trust in Poland after 1990. It would be odd to suppose that the Polish quest was grounded in trust. Many moves signaled the commitment to joint action against the dying Communist regime, and these enabled people to think that the risks of their opposition would not be disastrous. The last days of the regimes in Poland, Czechoslovakia, East Germany, and Romania saw mass actions in major public squares in which, by coordinating with so many, all were substantially protected against violent reprisals from the regimes.

XREF: Challenges conventional social capital theory linking social order to pre-existing trust; connects to discussions of collective action problems and revolutions without prior trust.

Suppose there were no cases of low trust and high social order. Unfortunately, this fact would not settle the issue because social order provides the background conditions that facilitate trust by creating the conditions for stable ongoing relationships and backing them with law to block the risk of massive losses from wrongly trusting someone. That is, social order produces the conditions for trust and therefore must commonly produce trust.

Much of social structure can be explained as a response to obstacles to resolving problems of cooperation and coordination through simple individual-to-individual trust. Indeed, Hobbes' political theory addresses the need for government to back cooperative life when trust relationships without such coercive backing would be inadequate to the task. In contexts in which individual-to-individual trust commonly fails, we find institutional and cultural arrangements that make life work more or less well in lieu of trust. These devices often work by securing trustworthiness on the part of those with whom we must deal, even though epistemological and time constraints may keep us from knowing enough to assess their trustworthiness or to trust them.

DEFINE: Introduced the concept that social structures and institutions function as substitutes for trust-based cooperation.

The biggest and most pervasive problem for us in trusting others is not the malign problem of dealing with cheaters but the relatively neutral problem of often having to deal with people with whom we cannot expect to have ongoing relationships in which to ground incentives for trustworthiness.

XREF: Connects to game theory literature on repeated interactions and tit-for-tat cooperation, where ongoing relationships ground trustworthiness.

Consider the limits on how many individuals one can trust. Even before getting to trust, someone who already has several friends may not think the possible benefits of investing in developing a friendship with yet another person is worth the risk. This problem may explain at least part of the phenomenon of clique and friendship-group formation. It might also be part of the explanation of familism in contexts in which families are relatively large, that is, large enough to exhaust a substantial part of the resources any member might have for investing in rich relationships. It may also be part of the explanation of ethnic exclusion.

XREF: Connects to social capital and Dunbar's number literature on cognitive limits to relationships. Also echoes network closure arguments in sociology. SEED: Angles: why trust is a finite resource in polarized groups; how relationship-investment limits explain in-group/out-group boundaries and familism across cultures. QUESTION: What is the actual empirical ceiling on deep relationships, and how does it scale with group size?

In the substantial literature on the subject, generalized trust is loosely seen as unspecific trust in generalized others, including strangers (Rotter 1980). From various experimental studies and surveys, it is supposed by many social scientists that Americans have higher levels of generalized trust than do people in many other societies (see, for example, Fukuyama 1995; Yamagishi and Yamagishi 1994).

DEFINE: Clarifies the technical meaning of 'generalized trust' as opposed to trust in specific known individuals.

What is in fact needed if we want successful cooperative relations is trustworthiness, which is likely to beget trust from those who learn of our trustworthiness and recognize its utility in reputational and trial-and-error experience.

Do I trust the vastly larger number of people with whom I have no dealings in those matters? No—but this is not a harsh answer. Most of these are people I do not even know and have no reason either to trust or to distrust. Unfortunately, if we begin to articulate our questions precisely enough to get at such discriminating differences, we virtually have to explain what we are seeking to those we survey (or those whom we put through experiments). We thereby give them theoretical understandings they did not have, and we elicit answers or experimental responses to those understandings rather than to their normal experiences.

XREF: Relates to the observer effect in social science research and the problem of reactivity in survey methodology.

If we compare across nations, we find lower levels of reported generalized trust in some societies than in the United States. What questions are people in the United States answering when they say they trust most people? What questions are those in certain other societies answering when they say they do not trust most people? Evidently, we are answering different questions, perhaps because we are differentially alert to the problem of dealing with those with whom we do not have ongoing relationships or perhaps because our background institutional structures differ in the scope of the interactions they protect.

XREF: Connects to comparative political culture and social capital literature (e.g., Putnam, Inglehart) on how institutional context shapes survey responses. SEED: Angle for an essay critiquing raw cross-national trust comparisons as misleading — survey responses are filtered by institutional backdrop and relational norms.

If there is widespread trust, it is therefore of the form that lots of people trust other particular people. That is to say, there are pockets of trusting relationships, and possibly almost everyone is included in such pockets. Alternatively, it may be that a typical individual is involved in various relatively small networks in which each trusts the others with respect to some range of issues.

One can sensibly question the claim for each of these terms here. For example, one might note that society without trust is very nearly impossible, but in a sense very different from that apparently intended by Eisenstadt and Roniger. That is to say, if we do build stable, continuing relations with others, we will commonly have the conditions, including the relevant incentives, for trustworthiness and trust. It would therefore be virtually, although not logically, impossible to escape the development of some trust.

Generalized trust is, at best, like trust in government in the following sense: We are merely confident that most of those with whom we might interact will be at least moderately trustworthy. It is not truly generalized but has scope constraints. Those who rely on generalized trust for explanations of various things must grant that their claims depend on context. Hence they are saying, for example, that middleclass Americans are likely to be trustworthy with respect to certain kinds of things, so that other middle-class Americans can relatively safely rely on them.

DEFINE: Clarifies that 'generalized trust' is not truly universal but bounded by context and social group.

Much of our ordinary cooperative activity is embedded in groups of which we are members. Groups are themselves the focus of cooperation for collective purposes, and they can help to induce cooperative behavior by their members even in essentially dyadic relationships. In the former case, when we need collective action, we might wonder whether our groups can be trusted to cooperate in accomplishing our purpose. In the latter case, we often find that our groups oversee our individual actions and induce us to be cooperative even though we could not say that the group and its members trust us. The way small communities induce their members to behave in certain cooperative or coordinative ways is through sanctions for the violation of communal norms, among the most important of which is a norm of cooperativeness within the group.

Can a collectivity be trustworthy on the encapsulated-interest account? Consider two very different strategically defined classes of groups: groups mobilized by coordination and groups mobilized for collective action (see further Hardin 1991a). Suppose a group is coordinated behind a leader, as happens with charismatic leaders, in response to what the leader wants it to do, as in the case of the seventeenth-century Sabbatai Sevi, "the mystical Messiah," who led a messianic movement in central Europe. Coordinators can commonly count on such a leader because they will withdraw support if the leader violates their mission, making it generally in the leader's interest to attempt to fulfill their expectations. Of course, the leader's interest in fulfilling that expectation can be trumped by changed incentives or preferences. So long as the leader has an interest in pursuing the goals behind which the followers are coordinated, however, he or she can be trusted and derives power from the coordination of the followers. It is limited power in the sense that it cannot be used for just any purpose. Rather, it can be used only for the group's purposes or mission (Hardin 1995, chapter 3). Coordination power therefore fits very well with the three-part relational account of trust.

DEFINE: Distinguishes groups mobilized by coordination from groups mobilized for collective action, and defines coordination power as limited to the group's mission.

In standard contexts of collective action that takes the form of a large-number prisoner's dilemma exchange, however, the group cannot be trustworthy. The members of the group might well share some set of interests as, for example, the people of Los Angeles and Houston virtually all share an interest in reduced pollution in their cities. If all of them would stop barbecuing over open flames in their backyards, pollution would be significantly reduced. The interest of any given individual in these cities, however, is typically to renege on acting for the collective benefit, to free ride on the efforts of others, while barbecuing as usual. According to the logic of collective action, your own interest, as in this example, is to free ride on the efforts of others even with respect to your own personal interest in the product of group effort (Olson 1965; Hardin 1982a). Eo ipso, your interest is not likely to encapsulate mine with respect to that joint product if it does not even include your own.

DEFINE: Explains the free-rider problem in large-number prisoner's dilemma contexts: individual incentives favor reneging even when interests are shared.

Note a peculiar asymmetry in the possibilities for trust and distrust in such collectivities. The worst implication of the grand transition from a small-scale to a very large-scale, impersonal society might eventually, because of power differences and general cynicism, be to produce fairly generalized distrust, which can make sense in a way that generalized trust does not. Generalized distrust would not, however, block the possibilities for trust as encapsulated interest in ongoing relationships.

XREF: Connects to broader sociological work on social capital and trust (e.g., Putnam) that treats generalized trust as an unqualified civic good.

In the logic of collective action, small groups are often expected to succeed even though very large groups cannot. The principal reason for their success is that the dyadic relationships between each pair of group members play a large role in motivating cooperative action. Hence trust relationships can enable us to cooperate beyond the dyadic level to some extent even though such relationships and their effects must run out for interactions within very larger groups.

XREF: Connects to game theory and the free-rider problem in collective action literature (Olson). The dyadic trust mechanism is a novel micro-foundation.

When we have the thick relationships of a small, close community, we may find that our interactions are governed by norms of cooperativeness that are collective rather than dyadic. I behave well toward you because the community will sanction me if I do not. In such a case, the norms of cooperativeness may be sufficiently effective that we do not so readily develop dyadic trust relationships over many things, because those things are governed by the communal norms (Cook and Hardin 2000).

XREF: Connects to sociological concepts of social capital (Putnam) and institutional trust replacing interpersonal trust.

In Shizuko Go's Requiem, a painfully beautiful novel about the destruction of a vast web of social relationships in Japan through wartime deaths in the last months of World War II, the heroine Setsuko is entertained by an older woman whom she has never met before and is unlikely ever to see again. She recalls "the familiar precept of perfect hospitality: 'We meet but once'" (Go 1985, 107). There is strategic subtlety in this bit of popular wisdom. If I know we meet but once, my hospitality is not an initial move in a potential trust or exchange relationship. It does not encapsulate your interest in reciprocity over the long run. It is purely a gift or an expression of my hospitable character.

DEFINE: Clarifies that 'meet but once' hospitality is strategic, removing the possibility of long-term reciprocity from the exchange.

One could construct arguments for the rationality of developing strong normative commitments: for example, following a norm saves on the burden of making decisions anew in many contexts. Alternatively, one might be able to give a rational reconstruction of the rise of such a norm, in which case the individual's behavior might simply be normative outright. Setsuko's older woman has been taught to be kind in certain circumstances of hospitality, and she might behave that way more or less independently of broad incentives to vary her degree of kindness.

XREF: Connects to behavioral economics and social norm theory — how automatic norm-following economizes on cognitive effort versus deliberating each choice.

If you and I are part of a fairly rich web of interactions in a relatively small, close-knit community, we might imagine that we would therefore grow up to be relatively trustworthy. Although the data do not yield definitive conclusions, some studies suggest that small, close communities are governed more by norms than by trust relationships (Amato 1993; Fischer 1982; see also Cook and Hardin 2000). In urban contexts, people find their ways into multiple networks for varied purposes, such as work, recreation, close friendship, and so forth. In one of these networks, they are likely to develop ongoing relationships that enable them to develop trust in one another with respect to the particular issues around which that network has formed. If I violate your trust with respect to the issues at stake in our network, your sanction is simply to withdraw from further interactions with me, and if my reputation gets around, I might even be dropped from the entire network. I would then have to find other people with whom to deal on the relevant issues.

XREF: Connects to sociology of embeddedness and network theory traditions, and to ideas about reputation as social currency within closed groups.

In small communities, instead, general norms of cooperativeness govern many behaviors. Anthropologists and sociologists sometimes call this generalized reciprocity. These norms are enforced at the group or communal level and not in reciprocal one-on-one relationships. For example, if one of us has a serious illness or death in the family or some other crisis, others help in various ways according to their own particular capacities. I might help you but then we might never be in positions in which you should, according to our norm, help me. Hence the norm is not reciprocal but is general—it is “universal” within our community.

Although members of a small community might never develop a sense of reciprocal trust such as that in the encapsulated-interest account, they could all or almost all develop trustworthy behavior. These members would then be somewhat like the infant whose parents are supportive and whose experiences are benign, so that the child develops optimistic expectations of the trustworthiness of people and is therefore relatively ready to take the risks associated with cooperating with people. When a member of such a community enters a broader society without the protection of the sanctions of the communal norm of cooperativeness, this optimism should lead to attempts to cooperate with others and therefore mastery of assessing trustworthiness.

Of course, the person who is trustworthy while in the ambit of the community and its norms and sanctions might be completely untrustworthy outside that community—where its norms do not reach. It is especially problematic for trustworthiness as moral or other commitment if these are group specific. Ethnic and even merely neighborhood conflict is commonly exacerbated by the attitude of a group’s members that nonmembers are proper subjects for exploitation and abuse. Indeed, they sometimes think it actively a good thing to abuse certain outsiders.

XREF: Connects to moral relativism and in-group/out-group psychology literature, plus the 'bad apple' vs 'bad barrel' debate about situational ethics.

It is the genius of the market to organize exchange in ways that commonly do not require high levels of trust. Its workings recall Barak’s response when asked whether he trusted Arafat (quoted in chapter 1). His answer was that trust was not at issue because each sought to push the interests of his own people. Similarly, in the market, I do not trust Ford or Microsoft, but I do rely on them to follow their own interests in being disciplined by market incentives, perhaps with a bit of help from public regulatory agencies.

My “trust” in the market may be like my trust in simple facts of nature—it is merely confidence from induction. I will correct specific details of my confidence when any dealer out there violates it, but I will otherwise treat each dealer as benign, at least in the sense of not malign. If I share Adam Smith’s view that most dealers are likely to share my interests (because they must serve my interests to serve their own), I may even think of them as actively benign. I can do this because dealers are typically in competition with other dealers and they must live by their reputations and by the development of ongoing customer bases.

XREF: Connects to Adam Smith's concept of self-interest serving mutual benefit, and broader epistemic questions about induction from Hume.

As Hume (1978 [1739–40], 3.2.8, 546) says of contracts, if they “had only their own proper obligation, without the separate sanction of government, they wou’d have but little efficacy in [all large and civilized] societies. This separates the boundaries of our public and private duties, and shews that the latter are more dependant on the former, than the former on the latter.” Hobbes may have exaggerated the extent to which powerful institutional sanctions are required for grounding trust and promises, but he was not radically mistaken.

XREF: Connects to contract theory debates in political philosophy — echoes the Hobbes-Hume line on the institutional grounding of promises and trust. QUESTION: Worth probing how far private obligations really rely on state enforcement versus social norms operating without institutional sanctions.

After the revolt of 1647, the Habsburgs deliberately worked to destroy trust relations in order merely to maintain control (until Naples passed to Austria under Bourbon rule in 1738). Three Neapolitan political economists of the eighteenth century, Paolo Mattia Doria, Antonio Genovesi, and Gaetano Filangeri, attempted to explain how a working economy could be created on the ruins of the distrusting Spanish order.

XREF: Connects to broader themes of how imperial powers sacrifice social capital for control.

If a producer's reputation with potential buyers is of great interest, the producer has reason to produce higher-quality goods even though there might be no easily specified contractual definition of quality. Hence, as in the literature on extracontractual aspects of business dealings (for example, Macauley 1963; Bohnet, Frey, and Huck 2001), producers and buyers would be in a partially cooperative relationship rather than the nearly adversarial relationship of the Soviet economy.

XREF: Connects to reputation economics and the broader literature on extracontractual business dealings, including Macauley, Bohnet, Frey, and Huck.

This general background or atmosphere of trustworthiness makes not only the market but also social life more generally go much better than it would without such an atmosphere. Hence life in a harsh ghetto or in a society that has broken down into violence and rabid self-seeking, as in Somalia at the end of the twentieth century, is hampered by the prudent lack of trust and by the disastrous lack of institutions to enable joint enterprises to proceed even without much trust. Introducing trust in such a context would be pointless. What is required for a constructive atmosphere is trustworthiness, which cannot easily be established by individuals across a whole society that has destroyed it. Again, before trustworthiness can be established there must first be institutional safeguards against the potential for disastrous consequences of dealing with others so that people can begin to take the risk of cooperating in ways that, if successful, would lead to trust relationships.

XREF: Connects to the chicken-and-egg problem in institutional economics — how trust and institutions bootstrap each other in development contexts.

In a society in which the background standards are those of the con artist, stances toward others will generally be defensive, and wariness might block most cooperative dealings, as in contexts of pervasive distrust and untrustworthiness. In a much more cooperative society, which is likely to be either a very close society or one with relatively good institutional structures to protect various dealings, the typically correct stance will be one of openness rather than wariness. Science has traditionally seemed to be such a society. Scientists generally seem to think that fraud is rare in science, but there are few data to establish that claim.

XREF: Connects to discussions of scientific norms of trust and replication crises in science.

When your cousins now do play a strong role in mediating our relations, I have reason for distrust. This is, of course, the point of Max Weber's thesis (1981, 228) on the rise of the modern business corporation and the invention of the idea of corporate capital of fourteenth century Florence: the resources and accounts of the individual get separated from those of the firm. It was also the original point of rules against nepotism that are now increasingly in abeyance when the relative who might work for a firm is not a brother or male cousin but a woman, especially a spouse.

XREF: Connects to Weber's thesis on the corporate form and the 14th-century Florentine invention of separating individual from firm accounts. QUESTION: Worth probing how modern nepotism rules have asymmetrically relaxed around women kin — what does that reveal about gendered trust assumptions?

Finally, note that for a real-world society in which there must be substantial variance on most social, intentional, cognitive, and other such measures, average levels of anything may be far less important than actual distributions, especially if individuals and societies have ways of selecting on the relevant dimension to found social structures on differences in trustworthiness. So, for example, if we are in a society that is hierarchically organized for various purposes, it may often be more important that the most competent or trustworthy people be in certain hierarchical roles than that people of average competence or trustworthiness be in those roles. Even in a strictly egalitarian context it may be more important that there be some highly trustworthy individuals in the system than that everyone be close to the average in trustworthiness. For example, in a market one exceptionally trustworthy dealer may force others to be more trustworthy in order to stay in competition, as John Mueller (1999, 79–80) argues for the case of John Wanamaker and the modern department store. On the other hand, in certain contexts, in a society in which personal relationships are generally not deviously exploitative, one can take a lot for granted on first meeting a new associate. Con artists, sexual exploiters, and many others have a field of play only because the background standards and expectations of behavior are relatively high.

In the academic literature, there are four main theories or models of trust that are actually brought to bear in empirical claims and research. Three of these are based on the kinds of reasons for judging the trustworthiness of the potentially trusted, and we could as sensibly say that these are three different theories of trustworthiness as that they are theories of trust. These kinds of reasons are encapsulated interest, moral commitment, and commitment from character. Two of these—moral commitment and character—are dispositional reasons, and the other—encapsulated interest—is a reason from interests. The fundamentally important common feature of these three theories or models of trust is that they require cognitive assessments of the trustworthiness of the potentially trusted. The fourth theory is purely dispositional trust that is not grounded in the assessment of the trustworthiness of the trusted and therefore is not at all a theory

DEFINE: Distinguishes three reason-based trustworthiness theories from a fourth purely dispositional trust model.

Most of the discussion of a disposition to trust (for example, Rotter 1980) could easily be read as supposing that people can be relatively optimistic about others generally, perhaps for reasons of early experience, as discussed in chapter 5. If this is the meaning of trust as dispositional, then it is not generally a theory of trust but merely an explanation of why some take risks more readily than others do or why some read the risks as less serious than others do. Hence the dispositional view is about marginal variations in trusting in the face of whatever evidence of trustworthiness is available. The view that trust is purely a disposition and that it is therefore unrelated in any specific instance to the object of the trust (trusting B to do X) is surely irrelevant to the experience of trust for most of us most of the time—

DEFINE: Clarifies what dispositional trust actually means — not a theory of trust itself but an account of marginal variation in trust behavior.

As noted in chapter 1, most people, when asked whom they trust, speak immediately of people with whom they have ongoing relationships. For them and for the encapsulated-interest view, trust is relational. My trust of you grows out of my relationship with you; it is not independent of that relationship (or of reputational proxies for such a relationship).

DEFINE: Introduces the 'encapsulated-interest view' where trust is inherently relational and stems from personal relationships or reputational proxies.

In chapter 6, I discuss falling in love and into friendship as ways in which trust comes quickly, virtually before any interaction. What often happens, however, is that people grow into a richer relationship of friendship or at least cooperativeness with respect to some range of matters. At that point, trust may often be overdetermined because trustworthiness is overdetermined. You may be trustworthy out of interest, morality, and friendship. You may even develop a simple disposition to be trustworthy to me, and you might develop a disposition to trust me. (Such dispositions toward particular persons could be shattered by misunderstanding or bad faith.)

XREF: Connects to the relationship between trust, friendship, and morality — parallels other accounts of trustworthiness being reinforced by multiple independent grounds.

Trust typically arises at the level of, and is grounded in, relatively small-scale interactions. It is not restricted to merely dyadic relationships, but it cannot be grounded in interactions involving very large numbers of people. It is generally a cognitive notion, and it faces epistemological and temporal limits on just how many relationships one person can master. Because most of us live in large-scale societies, we need devices other than trusting and trustworthiness to make many of our more or less cooperative activities go well. Typically, we can overcome the scale limits of direct trusting by relying on social constructions such as intermediary guarantors and reputations, although even these devices are commonly limited. To go much further, we must have strong institutional and especially legal backing that displaces the need for trust and reliance on others' doubtful trustworthiness. Unfortunately, such backing cannot be made to work efficiently—or, in some contexts, even at all.

XREF: Connects to Ostrom's work on commons governance and collective action, and to Elinor's scale-of-trust findings.

It may still be true that trust and trustworthiness are fundamentally important in making large-scale activities and, especially, large social institutions function. To show how they do this, however, requires substantial unpacking of the relationships within those institutions to understand how trust plays a role at the micro level. Trust is inherently a micro-level phenomenon. It is individuals who trust, and it is individuals who, under institutional and other constraints, are trustworthy to some, perhaps limited, extent in particular contexts. Trust and trustworthiness may permeate the social structure, but they do so bit by bit. And much of that structure is a response to the difficulties of relying on trust and trustworthiness to motivate cooperation.

One might argue that the first two of these questions are not really about trust but about the elemental decency of “most people,” not about their actual—trusting or nontrusting—relationship to anyone in particular.

Here, of course, the vagueness of the instrument, which may be necessitated in part by the nature of survey research, is at fault in that it does not give respondents much leeway. Sensible respondents must read this question as not being about substantial issues, such as whether one could trust another to donate a kidney or to repay a large, unsecured personal loan, or even much less substantial issues. Hence although it is unspecified, the range of matters on which “most people can be trusted” is not likely to be unrestricted.

DEFINE: Clarifies how 'most people can be trusted' survey items really measure limited, context-bound trust rather than generalized trust, despite lacking explicit scope.

Again, only one of these questions—the first—actually uses the term trust. The third question might elicit a judgment of policies more than of trust. The second and fourth questions address motivations that might bear on the trustworthiness of officials. As is true also for the General Social Survey questions, answers to these questions tend to be strongly correlated, so that it is sensible to claim that they are tapping something coherent even though it is arguably bold to assert that they tap a sense of trust in any particular conception of trust. Reputedly, when they were first added to the National Elections Studies battery, these questions were intended to tap political cynicism, not trust.

QUESTION: Worth exploring whether widely cited trust measures systematically conflate trust with cynicism, skewing research findings. XREF: Connects to methodological critiques of measurement validity in political science survey research.

1. Adam Seligman (1997, 13–15) claims that trust as we know it is a strictly modern concept because it arises out of strictly modern conditions. This is a perhaps Platonic stipulation in its own right. Niklas Luhmann (1988, 96) makes a similar claim: “Trust comes with the discovery or reconceptualization of life as involving risk in early modern times.”

QUESTION: What specifically changed in early modern conditions that made trust possible or necessary? The risk-reconceptualization angle is worth chasing.

Aristotle (Eudemian Ethics 8.2.123b.12–16) notes that “there is no stable friendship [philia] without trust [pistis], and there is no trust without time.” We may suppose that the point of time in this claim is to have opportunity for enough interactions—including interactions into the future—on which to ground trust. … According to Paul Bullen (personal communication, 26 February 1998), Aristotle uses the word “philia” to cover more than just what we would call friendship. It can include any harmonious relationship, even impersonal ones based purely on selfinterest. Depending on the context, “pistis” can mean belief, confidence, and rhetorical proof, as well as trust. If pistis is trust for Aristotle, then evidently it is a cognitive term.

DEFINE: Clarifies that Aristotle's 'pistis' is a cognitive term encompassing belief, confidence, and rhetorical proof, not merely emotional trust; also notes 'philia' spans all harmonious relationships including self-interested ones.

Baier (1986, 251) thinks the prisoner’s dilemma is overemphasized in discussions of moral philosophy and that this is especially a mistake for discussion of trust. However, she has an unduly formal view of the prisoner’s dilemma as inherently fitted to contracts and fixed payoffs. It is because many relations have the prisoner’s dilemma structure that trust is at issue in them.

XREF: Connects to game theory and social contract literature on trust, and the author's defense of the prisoner's dilemma against Baier's critique.

Schelling (1989) canvasses peculiar devices for securing compliance with promises in such difficult contexts as those between strangers. Establishing reliability in such contexts requires strong measures, such as subjecting oneself to risk of real, often unrelated, harm if one fails to comply with one’s promise.

XREF: Connects to game theory and commitment devices literature (Schelling's strategy of commitment), and to hostage-based trust-building in anthropology and economics.

It is sometimes supposed, on the contrary, that promising is typically used to regulate relations with strangers. Baier (1986, 246) says that exchange of promises typically requires “one to rely on strangers over a period of time.” Unless my worldly experience is extraordinary, this view is prima facie false. Promises to genuine strangers are rare, not least perhaps because a stranger would not trust one’s promise (see further Good 1988). … When we have to rely on strangers in important matters, we commonly prefer to bind them through contracts under law.

XREF: Contradicts Baier's view (1986, 246) that promising regulates relations with strangers; the author explicitly engages her claim.

One might suppose that the first of these branches is incoherent almost by inspection of the nature of our lives, in which action of moral significance is most often interaction. Social life is relational, not individual. I cannot cause outcomes by my actions but must rely on our joint causation of outcomes as merely a partial function of my own actions. To focus on rules for behavior or the individual's action taken separately from others' actions therefore grossly oversimplifies the nature of our problem (see further Hardin in press b, chapter 6; Hardin 1988b, 68–70). For example, such deontological rule following is fundamentally irrelevant to most public policy. Indeed, when it is intruded into public policy, it is often harmful to the purposes of the policy, as Jonathan Baron (1998) argues forcefully. He notes that we should not be surprised when policies produce suboptimal outcomes, such as organ shortages and even world poverty, when the policies result from reliance on deontological

XREF: Connects to consequentialist vs. deontological ethics debates, and to the broader literature on collective action problems and responsibility.

Baier (1985, 61) says trusting someone is always a risk "given the partial opaqueness to us of the reasoning and motivation of those we trust and with whom we cooperate." There is an "expected gain which comes from a climate of trust"—hence, in trusting we give up security for greater security. But, again, it is not trusting that is risky, it is acting on trust that is risky. Trusting is merely a bit of knowledge, and, of course, knowledge is fallible. Fallible trust can seem to give us grounds for acting when the acting will turn out to be harmful to our interests.

DEFINE: Distinguishes trusting (a knowledge state, fallible but not risky) from acting on trust (where risk actually resides).

Becker supposes that an abused child who clings to an abusing parent, or an abused spouse who clings to the abuser, trusts the abuser—hence noncognitively trusts. I am not qualified to discuss the psychological literature on such relationships, but this seems unlikely. Rather, the child or the abused spouse has no better place to go. The citizens abused by Papa Doc did not reveal noncognitive trust by the mere fact of staying on in Haiti.

XREF: Challenges Becker's notion of noncognitive trust; connects to theories of learned helplessness and structural coercion.

If "trust" is selectively directed at only certain other people but not based on grounds for selecting which others, then it must be capricious—unrelated to its objects and not a consistent expression of character. Such "trust"—which should be characterized as acting as though one trusts—seems neither sensible nor meritorious.

DEFINE: Distinguishes genuine trust from 'acting as though one trusts' — a conceptual clarification of what counts as real trust.

It is true that where there is no room for choice on the part of another, trust cannot be at issue. Yet it can be at issue when I am quite confident of your choice of action. In a particular context, it makes sense to say I trust most the person I think most likely to act in a certain way, and I trust least the person I think least likely to act that way.

QUESTION: Distinguishes trust from reliance on predictability—worth probing where the boundary actually lies.

Karamazov's lieutenant colonel has a very strange morality and sense of honor. He can act illegally and irresponsibly toward the army and the nation in which he is an officer. Yet he expects Trifonov to act in a trustworthy way with respect to him in their criminal dealings. Morality and trust often do not correlate well.

XREF: Connects to themes from Dostoevsky's Brothers Karamazov about hypocrisy and moral double standards in criminal dealings. SEED: Potential piece on how people impose stricter moral expectations on partners than on themselves, especially in illicit or grey-area arrangements.

In a classic study of military performance, Samuel Stouffer and colleagues (1949, 142–49) argue that great familiarity among members of a small combat unit often leads to great courage under fire because the members of the unit are loyal to one another rather than merely to the larger national purpose. In this case, all members incorporate the interests of the others in their own to some extent (see further chapter 6).

Thresholds for behavior may be important and asymmetric here. We may tip from trust to distrust only when big enough news comes in. Additionally, it may be generally harder to tip from distrust to trust.

As Gambetta (1988, 172–73) notes, the Mafia might help someone sell fraudulently. This implicitly would make clear that without Mafia protection distrust of one's fellows is rational and would therefore undercut competition that would benefit almost everyone.

Other factors further burden Sicilian society. For example, economic backwardness produces little opportunity for advancement. A standard way to advance is to prevail over others in one's own society—hence, advancement is a positional good (Gambetta 1988, 163).

XREF: Connects to economic sociology concepts of positional goods and status competition, echoing Frank's work on positional externalities.

Perhaps this is the sense of Luhmann’s (1980, 4) claim that distrust is self-reinforcing: It does not generate enough information for the distruster to correct his or her view of possibilities.

XREF: Relates to confirmation bias and echo chamber dynamics — a distrusting stance filters out corrective evidence.

The worst payoffs are the (1, 4) and (4, 1) payoffs in the prisoner’s dilemma of game 2 in chapter 1. The first of these is worst for the column player, and the second is worst for the row player. These represent the outcome of one player’s taking the other’s holdings without anything in trade—hence theft rather than exchange.

DEFINE: Clarifies the distinction between theft and exchange through game-theory payoff structure: theft is the extreme where one player takes without giving anything in return.

Hardin 1980) should perhaps say "functional explanation" rather than "functionalist explanation," because the latter suggests the blind assumption that whatever is good for a society must have happened and whatever behaviors a society develops must be good for it.

DEFINE: Clarifies the terminological distinction between 'functional' and 'functionalist' explanation in social theory.

Hart says, in what is for him an unusual sociological claim, that without the voluntary cooperation of many, "the coercive power of law and government cannot be established."

QUESTION: How far does this cooperation extend — elites, officials, or the whole population? Worth probing Hart's distinction between internal and external points of view.

In general, it is important to note that one may neither trust nor distrust another or a government with respect to some issue. As noted in chapter 4, this is contrary to the vernacular sense of not trusting. A person who says, "I do not trust him," very likely means that he or she actively distrusts him. However, we are often ignorant of another's intentions or likely behavior, and therefore we are in a state of neither trust nor distrust toward that other.

DEFINE: Distinguishes the vernacular 'not trusting' (active distrust) from the formal sense of a neutral third state of neither trust nor distrust, often arising from ignorance.

I once stood on line at the stand of a German street vendor who dispensed wurst, sandwiches, drinks, and many other things with seemingly choreographed efficiency of movement that was remarkably graceful. When complimented by the person before me, the vendor replied, no doubt truthfully, "I know my limits."

This is a claim that should be testable experimentally. To my knowledge it has not been subjected to any tests, although one might draw inferences from some experiments surveyed by John Ledyard (1995).

QUESTION: The untested nature of this claim suggests a gap in the experimental literature worth exploring — what experiments might settle it?

You might extend hospitality or act as though you trust me in order to demonstrate to me that you have faith in my morality or character or to give me an opportunity to live up to your hopes even though I may have no incentive to reciprocate your action. Alternatively, you may act according to a rule of reciprocity, doing unto others what they do unto you, even when it is not strictly in your interest to do so. Hence you may be trustworthy toward me because I am trustworthy toward you or because you may not wish to be the kind of person who acts toward another as though out of distrust that is not based on solid evidence. Such motivations are apt to lead to disappointment in many contexts, but they might be statistically justified in certain milieus. In particular, they might be justified in contexts in which there are rich possibilities of further interactions. In such contexts, however, interest is likely to conspire with your hopes in getting me to reciprocate.

XREF: Connects to game theory and iterated prisoner's dilemma logic — cooperation emerges when future interactions create stakes.

There are other possibilities. You might extend hospitality or act as though you trust me in order to demonstrate to me that you have faith in my morality or character or to give me an opportunity to live up to your hopes even though I may have no incentive to reciprocate your action.

You might extend hospitality or act as though you trust me in order to demonstrate to me that you have faith in my morality or character or to give me an opportunity to live up to your hopes even though I may have no incentive to reciprocate your action. Alternatively, you may act according to a rule of reciprocity, doing unto others what they do unto you, even when it is not strictly in your interest to do so. Hence you may be trustworthy toward me because I am trustworthy toward you or because you may not wish to be the kind of person who acts toward another as though out of distrust that is not based on solid evidence.

DEFINE: Distinguishes multiple distinct motives for behaving trustworthily: expressing faith in another's character, granting them opportunity to live up to expectations, and acting on a reciprocity rule.

There are other possibilities. You might extend hospitality or act as though you trust me in order to demonstrate to me that you have faith in my morality or character or to give me an opportunity to live up to your hopes even though I may have no incentive to reciprocate your action. Alternatively, you may act according to a rule of reciprocity, doing unto others what they do unto you, even when it is not strictly in your interest to do so. Hence you may be trustworthy toward me because I am trustworthy toward you or because you may not wish to be the kind of person who acts toward another as though out of distrust that is not based on solid evidence. Such motivations are apt to lead to disappointment in many contexts, but they might be statistically justified in certain milieus. In particular, they might be justified in contexts in which there are rich possibilities of further interactions. In such contexts, however, interest is likely to conspire with your hopes in getting me to reciprocate.

XREF: Connects to game theory and repeated-interaction models of cooperation (e.g., Axelrod's tit-for-tat), where reciprocity emerges from ongoing relationships.

Coleman (1990, 177–80) implicitly includes the trusted’s incentives when he notes that a reciprocal trusting relationship, as in mutual trust in the iterated prisoner’s dilemma, is mutually reinforcing for each truster. Why? Because each person now has additional incentive to be trustworthy. I trust you because it is in your interest to do what I trust you to do so long as you want me to do what you trust me to do. If there is some residue beyond rational expectations in one-way trust, there is less role for that residue in this straight, likely self-interested, exchange.

I was once cheated by a small shop owner in downtown Chicago, who sold me a watch battery that was already dead, although it had recovered just enough to make my watch run for long enough to let me think the battery was okay. That shop owner’s success was parasitic on the larger system of shops and stores that were more trustworthy, because my behavior in his shop was merely a generalization of my normal behavior in shops, most of which in my experience had been trustworthy.

I was once cheated by a small shop owner in downtown Chicago, who sold me a watch battery that was already dead, although it had recovered just enough to make my watch run for long enough to let me think the battery was okay. That shop owner’s success was parasitic on the larger system of shops and stores that were more trustworthy, because my behavior in his shop was merely a generalization of my normal behavior in shops, most of which in my experience had been trustworthy. In a particular shop that I might have used in my own neighborhood, I would not have needed to bother to check whether the shopkeeper opened a sealed package to take out a new battery. That shopkeeper depended on reputation for business by repeat customers. In the downtown shop, I should have kept a more diligent eye on the shopkeeper because his business depended on casual traffic and not on repeat business or reputation.

XREF: Connects to the concepts of reputation economics, trust, and institutions sustaining cooperative markets.

Weber argued that "the great achievement of the ethical religions, above all the ethical and asceticist sects of Protestantism . . ., [was] to shatter the fetters of the kin" and that, therefore, China did not develop a modern economy on its own.

XREF: This directly evokes Weber's Protestant Ethic thesis and its comparison with China's kin-based social structure — connects to debates about why capitalism emerged in the West.

Weber argued that “the great achievement of the ethical religions, above all the ethical and asceticist sects of Protestantism . . ., [was] to shatter the fetters of the kin” and that, therefore, China did not develop a modern economy on its own.

XREF: Weber's Protestant Ethic thesis connecting religious ethics to economic development — contrasts with Chinese kinship systems.

As Charles Peirce (1935, 3) expresses this commonplace view, “The scientific man is above all things desirous of learning the truth and, in order to do so, ardently desires to have his present provisional beliefs (and all his beliefs are merely provisional) swept away, and will work hard to accomplish that object.”

DEFINE: Peirce defines the scientific attitude as a desire to have one's provisional beliefs swept away, framing it as a core attribute rather than a method.

This was also true of the eighteenth-century data of Gregor Mendel on the inheritance of color among his peas. Mendel was essentially right in his views of inheritance, but he lacked insight of the full genetic structure of the process, in which the occasional plant might inherit two recessive genes and therefore might be the "wrong" color for its parents. A scientist today, facing such an anomaly as Mendel thought he faced, might be delighted at the implication that theory must be wrong and that the anomaly would itself be of great and original interest to fellow scientists.

XREF: Connects to Thomas Kuhn's work on anomalies and paradigm shifts in scientific revolutions.

This was also true of the eighteenth-century data of Gregor Mendel on the inheritance of color among his peas. Mendel was essentially right in his views of inheritance, but he lacked insight of the full genetic structure of the process, in which the occasional plant might inherit two recessive genes and therefore might be the “wrong” color for its parents. A scientist today, facing such an anomaly as Mendel thought he faced, might be delighted at the implication that theory must be wrong and that the anomaly would itself be of great and original interest to fellow scientists.