Some Forms of Trust

Abstract

Three forms of trust: topic-focused trust, general trust, and personal trust are distinguished. Personal trust is argued to be the most fundamental form of trust, deeply connected with the construction of one's self. Information technology has posed new problems for us in assessing and developing appropriate forms of the trust that is central to our personhood.

1. Introduction: Trust and the Challenge of IT

Trust is one of the foundations of human action. Every well-functioning person counts on relationships of trust with other people, social institutions, and even objects or artifacts. To be thrust into a world where none can be trusted is either a horrible punishment or an outright tragedy; it is dehumanizing. To be surrounded by trusted people and institutions, in contrast, is to be feel fully at home in the world, secure yet open to novelty and innovation. Most of us live neither in a hell of trustlessness nor a heaven where trust is ubiquitous and unstintingly available. So people are constantly faced with questions about whom to trust how far. And most people do a decent job of apportioning their trust and winning the trust of others.

As social conditions change, however, so do the conditions appropriate for giving and receiving trust. Knowing whom to trust in a small, homogenous agrarian community before the advent of modern transportation and communication was quite a different affair from knowing whom to trust in an industrial, urban society. The rapid growth of modern information technology in a global society promises to force us to learn new ways of creating and apportioning trust.

This poses a problem akin to what Jim Moor [24] has called a “policy vacuum.” Moor points out that the development of new technology often opens up new possibilities for action — not just the possibility of actions that couldn't have been successful before, but even forms of action we could not have tried. A hundred years ago, we could not have tried to split an atom or clone an organism; thought of such things was mere science fiction. And we did not then have to make real decisions about whether to try these things or how far we might take these technologies. When these became real possibilities, we had to make some difficult decisions about whether, when, and how such activities could be undertaken, and our past experience and policies, we found, did not settle the ethical and social issues that emerge in this new context.

Judgments about whom to trust, how far to trust them, how to validate one's trust or create trust where none has existed are vital questions in any society, but the rapid growth of information technology, the speed with which it operates, the complexity of the systems employed, its global reach, and the anonymity it often affords pose a real challenge to our practiced ways of apportioning trust. This essay, however, does not attempt to fill in the vacuum that IT has created in our practices of trust. It examines trust itself in preparation for that task, arguing that there are disparate forms of trust that need to be treated in different ways. The primary points of the essay are that trust is, most fundamentally, a relationship that holds between people or between people and their social institutions, things we often think of as agents with distinctive values and motives, and second, that it is a relationship that is crucial to the constitution of one's personality. The close relationship between trust and the constitution of personality also helps explain why trust is tightly linked to issues of privacy. If this is right, then all the new problems engendered by modern information technology are still, in the final wash, problems concerning human relationships and what kind of people we want to make of ourselves.¹

2. Methodological Preliminaries

I start from an intuition about trust, namely that the fundamental form of trust is that of a deep relation between persons, all other forms of trust being derivative in some way from that fundamental relation. By calling it a “deep” relation, I want to indicate that it reaches to the very heart of our identities as persons in ways that I hope to say interesting things about. Intuitions are at best highly defeasible – that is, subject to being defeated or overridden by other considerations – starting points for philosophical investigations and arguments. They need both explanation and defense. My arguments here cannot go much beyond the opening moves of a thorough attempt to explain and defend this intuition about trust, but one has to start somewhere.

Trust has become the focus of a good deal of discussion in both ethics [2,3,4,13-16,17,21-23] and epistemology [1,5-9,11,12,18-19] of late. Though there is not yet sufficient connection between work on trust in these two fields, this work has contributed significantly to a deeper understanding of trust and its fundamental role in our lives as rational beings. But we do not yet have anything like a definitive analysis of trust or a thorough categorization of its varieties and subspecies, and this has made discussions of trust murkier and more confusing than they need be.

On the way to defending my claim that trust is fundamentally a deep relation between persons, I will distinguish three species of trust, and argue that the differences between them are quite significant. Clarifying these distinctions should illuminate the status of claims about trust and the arguments offered on their behalf, for points made about one species of trust need not hold true of the other, and one must be careful which examples one offers to back up or illustrate one's claims.

Methodologically, I begin with what I consider to be a dangerous move. I use the surface grammar of my language as a clue that leads us to some deeper, philosophically interesting conclusions about something. Such attempts are certainly not new: Plato, Aristotle, and Kant (inter alia) all draw distinctions and construct categorizations inspired by the surface grammar of their languages. And what they do is often very illuminating. But we have to be aware, of course, that there is no simple, good inference from the surface structure of any natural language to any clear and interesting philosophical conclusion. The most we can hope for is good inspiration that can then be filled out with independent arguments. That's what I hope for here.

3. The Grammar of Trust

3.1. Some Syntactic Contexts

There are numerous syntactic contexts in which ‘trust’ appears.

I trust that your stay has been a pleasant one.

I trust Ellen Goodman on women's issues.

In God we trust.

John does not trust the local garage to fix the electronic timing on his car.

Mary trusted me with her jewelry while on vacation.

Harry trusted his old Honda.

I trust my son.

These and sentences like them provide the material for the opening moves of the argument.

3.2. Propositional Trust

First, I want to claim that the propositional form that heads the above list ought to be seen as derivative from the others. Trust is an attitude of mind, but I have doubts that it is first and foremost a propositional attitude. Some attitudes, such as hope and belief, seem most naturally ensconced in a context that takes a propositional complement (though even in these cases there are other forms available, such as hoping for something or believing in something). Some, such as fear, seem equally at home with a propositional complement or an object complement: I fear that my retirement fund may have lost its value long-term; I fear the neighbor's dog. But trust is not like either of these. Trust seems most comfortable when it takes an object complement of some kind and acts like a relation, albeit an intentional relation, to an object, not a proposition.²

Here are some reasons to believe this. First, a fairly weak argument: suppose someone begins to say “I hope …” and then hesitates. Appropriate prompts for a completion of the thought or sentence are “What do you hope?” or “You hope what?” Although I don't know any significant amount of linguistics, these both seem to be straightforward wh-transformations, from the declarative ‘that’ to the interrogative ‘what’.³ Now suppose someone starts off “I trust …” Again, an interlocutor can prompt for completion with either “What do you trust?” or “You trust what?” But it seems slightly jarring to me if the response is “I trust that p.” The natural response to “What do you trust?” seems to me to name some object of trust, not some state of affairs. The context can, of course, influence these matters deeply, and in many contexts “What do you trust?” itself would have to give way to “Whom do you trust?” or “You trust whom?”, which obviously beg for completion by a person's name, not a proposition. We seem to think of trust, like love, as primarily an object-related attitude, not a proposition-related attitude.

Second, when someone uses ‘trust’ with a propositional complement, it seems to me to connote a very thin sense of trust. In “I trust that your stay was pleasant,” or “I trust that you'll remember to bring your license next time” the speaker is signaling that her confidence in a particular truth is so high, she assumes she really need not raise the issue, but will nonetheless raise it to make sure everyone else (especially the interlocutor it is directed at) knows that it has been raised. “You have to trust that it will all come out right in the end” is an expression of a fairly vapid faith that might console some, but hardly satisfies if there are no specific objects of the recommended trust.

Such assertions are indeed complex speech acts, but they are far from constituting reports of core trust. “I trust that you'll remember your license next time” (in the circumstances most commonly envisioned as appropriate for such a remark) is equivalent to “I am sure that …” or “I am certain that …” or even “You'll remember to bring your license next time, won't you?” (perhaps accompanied by a knowing rise of the eyebrows). “I trust that …, “it seems to me, is naturally given an epistemic reading. This does not seem a central case of trust: other terms or concepts can easily substitute for it, and it can neither accommodate nor explain the moral weight that often attaches to trust.

My third reason may not be an independent reason at all; it is even more of a pure intuition than my two previous points, and like all intuitions should be taken with a grain of salt. But I find that it moves me, so I will pass it on. Imagine a person who trusts that p and trusts that q, etc., for a wide range of ps and qs, etc., but trusts nothing nor any person (and I don't think that is easy to imagine). I want to say that that person is untrusting and is, in fact, seriously deficient in trust. If, further, that person also could not recognize object-directed (as opposed to state-of-affairs-directed) trust, I would also want to say that he or she lacks a real concept of trust. At the very least, that person would have a difficult time applying the concept of trust anywhere, since it is only abstract and general for her, and without specific object.

Having now demoted propositional trust to a secondary role, at least provisionally, let me focus on the remaining usages on the above list. These, I claim, fall into two groups syntactically. First, there are the usages in which there is a prepositional phrase modifying the trust relation, and second, there are the usages in which a trust relation is asserted without any modifier. These are, I claim, clues to a distinction between kinds or species of trust that need to be recognized, but have not been heretofore. But the clues are not straightforward, for I will argue further that there are two distinct forms of trust relations that are described without using a prepositional modifier.

4. Three Forms of Trust

4.1. Topic-focused Trust

Let me now focus on those forms of trust that we express using a prepositional phrase as a modifier. One can trust another to do something: I trust my barber to cut my hair well. I trust the police to keep order at the demonstration. One can also trust someone with something: I trust the dry cleaner with my clothes. I trust my teenager with the car. One can trust someone on some issue or question: I trust Feynman on quantum electrodynamics and Ellen Goodman on women's issues. Trust in someone or something, I think, works differently, so I exclude it from my treatment here.⁴ In each case, the trust is focused in some way on a particular topic, action, or item, and the only matters that are relevant to the evaluation of that trust, whether it is justified or unjustified, deserved or undeserved, risky or safe, concern the relation between the trusted one and that topic and the truster's evidence concerning that relation. I want to group all such forms of trust together, and I shall call them ‘topic-focused trust.’

Topic-focused trust is certainly both a familiar and a relatively well-behaved animal. We commonly make such attributions of trust to ourselves and to others, and we have a pretty clear sense of how they are to be evaluated. There are several different dimensions of evaluation relevant to attributions of topic-focused trust. First, the truth conditions of attribution: what are the conditions for a topic-focused trust to exist? Second, assuming the trust (or the distrust) is there, we can evaluate it, again along several different dimensions: Is the trust justified? Is it deserved? (These two need not receive the same answer.) How much trust should be given?

What counts as relevant evidence for these evaluations is fairly well delimited. In each case, however, the standard employed is vague: It is not clear just what either my behavior or my other attitudes towards the barber, the dry cleaner, or Richard Feynman must be for it to be true that they are the object of my topic-focused trust. Nor is it clear how many times the dry cleaner or the barber must do a good job before I am justified in trusting them or they deserving of my trust. When one tries a new dry cleaner, is one already trusting the establishment by entrusting it with one's clothes? It isn't clear how many illuminating Ellen Goodman columns one needs to read before being justified in trusting her on women's issues, nor, were she to fall from grace, how many columns that fall might take. Does the fact that on an icy, snowy night, my teenager put the car in a ditch mean I can no longer trust him with it? All this vagueness granted, it is nonetheless the case, I claim, that topic-focused trust is relatively well-behaved in that it remains pretty clear throughout that only certain kinds of acts are relevant to the evaluation of such trust, even if the quantity requisite is not clear.⁵

The trust discussed by ethicists and epistemologists is often of this topic-focused variety. Trusting people to teach one's children, trusting people to keep their promises, trusting institutions with one's life savings, trusting people to tell the truth, trusting experts on the subjects of their expertise are all situations of ethical and/or epistemological import.

4.2. General Trust

But not all trust takes the form of topic-focused trust. We can attribute trust via expressions that utilize no prepositional modifiers. I gave two examples:

Harry trusted his old Honda.

I trust my son.

Syntactically, of course, these are on a par. The attributed trust is directed at an object designated by a noun phrase. How do such attributions of trust differ from prepositionally-modified attributions? It seems obvious that they are not topic-focused in the same way. It is true that in all forms of trust attributions there is an object referred to, but in attributions of topic-focused trust we are also told which of the object's properties is relevant to this attribution of trust. In that sense the trust attributed in our current examples is unqualified. Clearly, however, the lack of qualification is somewhat shallow; if I trust my car or my son, I need not trust them in every way, on every matter. But there is at least a general trust. If Harry trusts his old Honda, he will not worry about its starting on cold or rainy days, about its holding up under ordinary usage or suddenly falling apart. But Harry might be unwilling to take the Honda on a cross-country drive through the Rockies or the Alps.

This general trust is trickier to assess than the topic-focused trust I just discussed. It is less clear both what counts as relevant evidence and what the relevant standards ought to be. If the old battery in Harry's Honda gives out and the car fails to start, ought Harry to cease trusting the car? What if rust is found in the front-end assembly or corrosion in the cylinder walls?

I want to take a moment now to distinguish trust from another concept it might get confused with: reliance. In common parlance, it is true, we often do not distinguish between relying on someone and trusting them, but I'd like to suggest that there is an important distinction there that we ought to pay attention to. Reliability also has prepositionally modified forms as well as a more general, unmodified form, and I'd like to keep them both distinct from trust relations. Reliability has recently become an important concept in epistemology, but I think it is a big mistake to conflate that more or less technical concept of reliability with the concept of trust. Reliability as it occurs in recent epistemology is a concept defined in terms of statistical relations between well-defined (or at least (one hopes) well-definable) inputs to and outputs from some belief-producing process. There are related uses of ‘reliability’ in non-epistemic contexts: the relative reliability of FedEx vs. UPS can be ascertained by a study of the percent of undamaged, on-time deliveries they make. The reliability of a device, process, or system ultimately rests on its structure, the laws of nature, and other regularities in the world; it could be, with sufficient research, precisely quantified.

Trust, as I understand it, is not quantifiable. Empirical research is relevant to it in a different way, I believe, and though I don't believe that the laws of nature are ever violated, they are not the important ingredients in trust relations.⁶ Reliability is often the basis of our confidence, and we often use the language of trust to report or attribute such confidence. But this is, I think, dangerous, creating an erroneous impression that there may be nothing more to trust than reliance. But there certainly can be. Many Londoners rely on the underground; few trust it. There are clearly times when we rely on people in a fundamentally statistical sense: in this sense of ‘rely’, good industrial designers rely on people to do stupid things, and then work hard to make such stupidity difficult. But what if old Mrs. Johnson relies on her neighbors to drive her to the grocery every week? The line between relying on someone and trusting him or her is vague, but nonetheless important.

In the case of Harry's Honda, being able to attribute and assess his general trust are complex tasks, but not usually very daunting. I want to go on now to argue that there is a difference between general, non-topic-focused trust aimed at inanimate or impersonal things and such trust aimed at persons and their social institutions. Trusting persons is, in my view, what trust is really all about, the real focal point of the concept, and when the object of trust is a person and the kind of trust is not topic-focused, there are some peculiarities that we need to be aware of.

4.3. Personal Trust

4.3.1. Differentiated from General Trust

I can say simply that I trust my son. Period. Sometimes, maybe even often, when I say that I trust someone, it is short for some topic-focused form of trust, and the focus can be inferred from the context. But this is not always the case. I can say that I trust my son, period, with no context-dependent intended or elliptical completion or modification. There is a use of ‘trust’ in which it stands alone. If I trust my son, period, then surely there are many things I will trust him to do, many things I will trust him with, many things I will trust him on. But my trust is in him, without specific qualification. Let me call this personal trust.

Personal trust seems to be a lot like general trust. Like general trust, it would be tempting, but a bit misleading, to call this unqualified trust, for there can be reservations and exceptions in such a trust. I trust my son—but not when it comes to picking up his clothes. I trust my wife—but not to be on time. Personal trust makes provision for the foibles and eccentricities humans are prone to. But I want to resist the idea that personal trust is just general trust aimed at a person. There are important differences between the trust Harry has in his Honda and the trust he has (I hope) in his wife. At the very least, aiming general trust at a person transforms it in important ways.

Perhaps the most important symptom of the fact that general trust aimed at inanimate things is a weakened form of trust different from true personal trust is that the concept of trust is deeply tied to the notion of betrayal. It sounds weird to me to say that, if Harry's Honda doesn't start, it has betrayed his trust—or, rather, it sounds metaphorical to me, because to betray a trust, one must comprehend the trust in some way, one must understand that being trusted often imposes an obligation or responsibility, and then act in a way to violate that responsibility. And if that's the case, being animate isn't enough either. In most cases, when a dog disobeys its master, he doesn't betray her, even though we say she trusts the dog.⁷

Betrayal is an interesting phenomenon I wish we had more time to explore, but it is significant that betrayal may well be essentially more cognitively complex than trust itself. For it doesn't seem required for trust that the object of trust comprehend that trust and understand that it imposes an obligation, etc. Think here of trust between children. Yet, if real trust is tied to the possibility of betrayal, then trust itself may require a more complex cognitive environment than one would be led to believe on the basis of an examination of trust alone. As in many other cases, the full complexity of the phenomenon of trust can be appreciated only by looking at cases gone wrong.

Another indication that personal trust differs from general trust in some important ways is that general trust aimed at inanimate objects, especially artifacts, is focused on the function of the object or its role in the context. If Harry's trust in his Honda extends beyond starting reliably, running well, being safe and fun to drive, and other such driving-related tasks, Harry has a psychological problem. We also say such things as “I don't trust the weather.” The weather itself has no intrinsic function, but it certainly has a role to play in our lives, and this use of trust (or is it really mere reliability?) surely exploits or refers to that. A sheep dog that gnaws the furniture probably hasn't betrayed a trust, but one that savages the sheep he's intended to protect arguably has.

But when I say that I trust my son, I do not believe that it is always the case that a function or role for my son, sons in general, or human beings needs to be hovering in the background. True, there are many occasions when assertions or attributions of trust sans phrase are shorthand for assertions or attributions of some more topic-focused trust, and we read the topic off of the context. When what's in question is whether he'll remember to mow the lawn on the weekend, and I assert that I trust my son, it is immediately understood that I trust him to mow the lawn. But not all assertions or attributions of trust operate like that. If one understands all norms in terms of functions and fulfillments, so that being a good person is literally a matter of fulfilling well the function of a person, then trustworthiness and trust itself must also follow suit and be tied to functionality, but I see little reason at this point to push every normative property into this Procrustean bed.

Despite the flood of new work dealing with trust, the distinction between topic-focused, general, and personal trust seems not to have been noticed or taken seriously before. But recognizing this distinction can help us make better sense out of some of the (already good) work that has been done. For instance, much of what I have to say about trust echoes themes also sounded in Trudy Govier's seminal works on trust. But Govier does not distinguish between topic-focused and personal trust, and this leads to some muddling of distinct phenomena.

Let me take a specific example. In section I (“Preliminary Remarks”) in “Trust, Distrust, and Feminist Theory” [14], Govier asserts that,

“the expectations entailed by trust are open-ended: We expect that the trusted person will do the right sorts of things, and will not do the wrong sorts of things, in whatever situations may arise … We have an open-ended expectation that cannot be understood in terms of doing all acts specified by a given social role or set of rules” and that “in judging trustworthiness, in deciding whether to trust, we reach a sense of the whole person and his or her integrity or competence” (p. 18).

But Govier also stipulates that “Both trust and distrust are often relativized to role or context: we may, for example, trust someone as our dentist but not as an accurate commentator on political affairs” (p. 18). But the holistic and totalizing aspects of trust pointed to in the former passages do not cohere easily with the context- or role-relative aspects brought out in the latter passage. If I trust the clerk or the waiter to add up the bill (another of her examples), it certainly doesn't seem to me that I have had to “reach a sense of the whole person and his or her integrity or competence,” nor that I have implicitly judged this person to do the right thing “in whatever situations may arise,” independently of “a given social role or set of rules.” Govier wants to take personal trust as the model (or ideal type?) of all trust, and accommodate topic-focused trust with contextual limitations. But the descriptions of the two don't cohere well. I believe we have to posit at least two different species of trust to accommodate the phenomena.⁸

4.3.2. The Nature of Personal Trust

Let me try to spell out further how I believe personal trust differs from the other forms of trust. Personal trust is specifically different from topic-focused and general trust at least in part because it operates at a different level. It will take me a while to explain what I mean by describing it as a different level, so bear with me a bit.

In some forms of topic-focused trust it is easy to specify just what behavior is pertinent to evaluating and to fulfilling or betraying the trust. If I trust someone to be on time, then his or her general record of punctuality (adjusted perhaps for the seriousness of the occasion) is pretty much the only thing germane to the justification or evaluation of my trust, and whether the arrival is punctual (again allowing for the relative seriousness of the occasion and other competitors) is the only thing relevant to deciding whether that trust has been fulfilled or betrayed. Similarly, trusting someone on a topic allows for relatively clear specification of the pertinent behavior. Trusting with seems to me less easily specified. Trusting someone with your car certainly means that you expect them to take good care of it, not to wreck it, but the actual behaviors that count as taking good care of the car can be very context-dependent.

General trust is not, as I have said, so easily confined. But as I conceive of it, it is, indeed, a generalization of topic-focused trust. It is tied to a role or function; it covers, in general, the range of behaviors specified by the function or role, and the norms by which it is assessed are derived therefrom. Although I distinguish general trust from personal trust, it is probably the case that a great deal of the trust we have in other people is general trust, not full-fledged personal trust as I conceive of it. When, for instance, I trust the waiter, I trust that a certain role will be fulfilled adequately, perhaps even well. But I don't commit myself beyond that; I do not undertake to trust the whole person, and there need be no implicit judgment about the waiter's personal integrity.

Now I'm in a position to say why I want to describe personal trust as operating at a different level from either topic-focused or general trust. First, the expectations it entails cannot be specified in terms of particular behaviors (even particular-behaviors-in-context) because it is not really about the behaviors or dispositions to behave themselves. It is significantly more open-ended than standard cases of topic-focused trust.⁹ It is also more open-ended than general trust, for it is not tied to a role or function, it is tied to a person. Personal trust operates at a different level, a deeper level, where the concern is not with specific behaviors, tasks, functions, or roles, but with the motives and values exemplified in and controlling the person's behavior. Thus, it is concerned with a person's fundamental character. In the passages quoted above, Govier signals this by asserting that trusting a person entails a (perhaps merely implicit) judgment about that person's integrity. And there's something right about this, but we have to be clear about what that really amounts to. Trust does not entail judging the other to be a good person, to be morally upright, etc. There could be “honor (and trust) among thieves.”

The right way to put it, I think, is that in a case of personal trust it is the values and motives of the other that are endorsed, so that across a broad set of situations that other would be expected to perform much as one would want another to. Of course, we cannot estimate or evaluate the values and motives of others independently of their behavior, so the behavior of the other is always relevant to personal trust. But it is not the central focus of the trust; that remains on the other's values and motives.

In saying I trust my son, I am not just making or implying certain predictions about his behavior. I can predict the behavior of those I do not trust. ¹⁰ That personal trust involves more than our expectations implies several things. For one, it points to the fact that trust is more than a purely cognitive attitude. And it is doubly so, I believe. Personal trust is both an affective attitude (a point to which I will soon return) and a conative attitude. Let me focus for a moment on the interesting sense in which personal trust involves our wills. When I say I trust my son, I am endorsing his values and motivations and, in a sense, staking my own self on his behavior. In trusting someone, I am willing to let him or her be my proxy; I am willing to accept his or her words and actions as if they were my own. I am willing to accept his or her will as my own.¹¹

Personal trust, however, is not a mere recognition of like-mindedness. Two people could be of like minds on a wide range of issues and recognize that fact without trusting each other. (This might be a common occurrence among politicians.) In such a case, these people might rely on each other, knowing that the constraints of the system will keep the other in line, but it is not real trust—they suspect that there is a deeper divergence of their values or motives. Even if their actions in current circumstances coincide or reinforce each other, they suspect or recognize that, were circumstances (not that much) different, they would no longer be allies or like-minded.

This is why I think of trust as a very deep relation between people; in trust one puts oneself at risk, not just by exposing oneself to some external danger, but by allowing another to “carry the banner” for one and thereby putting not just one's body but one's will, even one's identity in the hands of another. Who I am will, in the end, be determined in part by whom I trust. If my trust is ill-placed, I will not be at all the person I was striving to be.¹²

Among the noncognitive (or perhaps better, extracognitive) dimensions of trust are affective commitments (See Karen Jones [21] and Lawrence C. Becker [4]). Trusting someone or something entails having certain beliefs and other propositional attitudes, but it also commits one to certain behaviors, patterns of behavior, motivational structures, and affective attitudes.¹³ I don't see that personal trust and topic-focused trust need differ in kind in terms of their affective dimensions. However, personal trust will almost always be more strongly and/or more deeply tinged with affect than topic-focused trust. After all, if I am right, a violation of personal trust is always more than a violation of my plans, which then need rethinking; it is a violation of myself, which must then be rebuilt.¹⁴ Topic-focused trust can also carry with it extensive and soul-engaging commitments, but it certainly need not. I can trust the dry cleaners to do a good job on the clothes or the check person at the museum with my bag and coat without entering into a complex network of commitments to behaviors and affects that carry the potential to rock my very soul or put my identity and the kind of person I am into question. (But someone I trust with my children holds my life and happiness in her hands.)

Those one trusts are, in a sense, extensions of oneself. Though we are all aware of the dangers of guilt by association, it is nonetheless true that one is known by the company one keeps: but the company one trusts is not just that by which one is known, it is in part that by which one is. Inappropriate trust prevents one from achieving oneself as intended. I can't do it here, but I believe that an argument can be constructed that relations of personal trust are partially constitutive of persons: anyone constitutionally incapable of relations of personal trust is a severely damaged person at best.

4.3.3. Trust and Privacy

We can see here why there is usually taken to be a close relationship between trust and privacy. One is open with others to the degree that one trusts them. Tavani [25] distinguishes three kinds of privacy: accessibility privacy, or freedom from intrusion; decisional privacy or freedom to make decisions about one's own affairs; and informational privacy or control over the flow & use of one's personal information. In each case we can see a link to the notion of a person's autonomy, for lack of or invasions of privacy tend to decrease one's ability to control one's decisions, actions, and appearance. (See also the references in Deborah Johnson's discussion of privacy in [20].) In personal trust, as I've described it, one puts one's self at risk by opening oneself up to another and placing one's self in that other's hands. One's autonomy is thereby put at risk, and one classic form of betrayal is a violation of privacy.

In an information-rich, globally networked world, informational privacy is a particularly significant issue. Who has what kinds of information about one, who else has access to it under what conditions, and what can be done with that information? These questions become increasingly important as more and more of our activities are mediated by information technology, which is becoming increasingly sophisticated at mining each interaction for further information. How much information about oneself should one put on a social networking site? To whom are different pieces of such information made available? Can one trust the website itself to implement its own stated policies reliably? And what about all the information about our interests, our preferences, and our decisions that can be gleaned from on-line activity, often without our knowing of such monitoring? The Beacon system on Facebook collected information concerning Facebook client activities on other websites (e.g., e-Bay, Travelocity, Overstock.com) ostensibly in order to be able better to target advertising at those clients. Some of that information was even published on Facebook against the clients' wishes. One of the most disturbing aspects of modern life is that we must constantly rely on large, complex, and faceless systems and bureaucracies, even when we find it hard to trust them. It is a constant challenge to us to ensure that our systems are not just reliable but trustworthy.

4.3.4. Developing and Extending Trust

The development and cultivation of trust is a complex and important topic. Children may begin with an unsophisticated and natural trust (of sorts) in their caregivers, but whatever initial form it may have, it must be nurtured to flourish and can easily be perverted.

So trust of the kind I am trying to articulate must sometimes be faithful or projective: if his parents and elders do not trust a young child to some degree in the anticipation of his growing into trustworthiness, the child will probably never learn either trust or trustworthiness. Trust in this sense is a norm-laden commitment we make, not a condition that we are either in or not. Personal trust is also pretty clearly never merely an epistemic affair, that is, confined solely to the truth of what one says or the probity of the evidence one offers. Personal trust clearly goes much deeper and must range more widely than the epistemic: as I said earlier, personal trust involves an assessment of the other's values and motivations.¹⁵ The other's epistemic values and motivations are part of this, but certainly do not exhaust it. Personal trust involves an assessment of the other's character and virtues. (To the extent that there can be trust among thieves, trust among the dishonorable, this assessment of character need not be absolute. The question is whether the other's values and motivations are consonant with mine, not directly whether his values and motivations are good.)

Furthermore, personal trust need not be strictly personal. One can have this kind of deeply committed, unspecific trust in things other than persons. For instance, one can trust or distrust the government, or one's company, or Amnesty International. These are all at least legal persons. But one could also have this attitude towards something amorphous enough not to qualify as even a legal person. For instance, one could trust the Western Tradition. If one can have personal trust in something that isn't a person, however, why can't one have personal trust in one's car or the daily horoscope? Personal trust as I have characterized it, however, engages its object at the level of values and motives. A car or the horoscope don't have values or motives – though they may represent someone's values and motives, so I don't believe they are appropriate objects of personal trust. Something like the Western Tradition, on the other hand, does have values, though it sounds weird to attribute motives to it. Trusting the Western Tradition would mean that one sees this tradition as sharing one's values and believes that, at least on the whole, this tradition or line of cultural communities either gets things right or at least is moving in the right directions. This seems quite possible to me. Indeed, loss of such trust can be a devastating experience, leaving one without support in matters of great import. Even the cynic will seek some safe harbor of people, institutions, or traditions he trusts. If we had nothing that we trusted in this personal way, we would be woeful creatures indeed.

Inanimate objects and artifacts are not proper objects of personal trust, I've argued. But if we were to develop truly autonomous artificial agents, artifacts that can be said to have values and motives, we would squarely face a significant policy vacuum. Should we insist that these creations are still our creations with values and motives derivative from our efforts, or should we happily welcome these beings into the community of persons as fully autonomous beings? I am wary of giving an apriori pronouncement on this, for it seems inevitable that the specific details of the development and implementation of the autonomous agents would be relevant to deciding how to treat it.

4.3.5. Pathologies of Trust

Personal trust should also be distinguished from what I'll call ‘blind faith.’ Blind faith can be similar to personal trust in that it operates also at the level of the person, not his or her behavior or performance, and in that it can involve unspecific endorsement (or perhaps better, acceptance) of the fundamental values and motivations of the other. But blind faith is blind, that is to say, uncritical. Notably, it does not have the flexibility built into personal trust. Blind faith is unreserved, making no allowances for the foibles or eccentricities of the other. In this, blind faith involves not just letting the other have one's proxy, but indeed surrendering oneself to the other, giving the other command over oneself. Blind faith may indeed be the developmental forebear of personal trust: young children begin with a natural and unreserved reliance on their caregivers that normally (unless the caregivers consistently betray their charge) develops into an equally natural, unreserved, blind faith or trust in those caregivers, and ultimately into the rich personal trust I've been concerned to point out. Some people may not get that far; they retain their blind faith in their caregivers or seek someone new in whom to trust blindly. But that is, in my view, developmentally arrested or neurotic. (For interesting reflection on developing trust, see Victoria McGeer [23].)

That blind faith in a supposedly mature person is a case of either developmental arrest or neurosis may not be a truth of empirical developmental psychology or psychiatry. Personal trust, construed as a normative commitment that involves an endorsement of the values (the normative commitments) of another, presupposes the presence of at least some critical abilities, that is, abilities to recognize and evaluate normative commitments. Thus, in my view, very young children do not begin with personal trust, but have to grow into it as their critical abilities develop. In adult blind faith, however, these critical abilities are systematically kept distant from one's faith. And this is at best odd. I don't think I can show that it is necessarily irrational nor even that it is usually unmotivated. Perhaps the best I can do is say that it seems to fall short of an ideal of personal integration that looms large at least in Western Ideology, and that I am perfectly happy at the moment to endorse. So, in this view, full-fledged personal trust lives in a context in which one's critical abilities, if not actually exercised universally, at least have no arbitrary or artificial constraints put on them.

4.3.6. Personal Trust and IT

Some of the issues that arise under the banner of trust in the context of IT turn out to concern matters of reliability: which programs, service providers, or agencies perform their functions properly how often? But to the extent that modern IT opens up new possibilities for knowledge, action, and interaction, it not only extends our reach but also exposes us, making us vulnerable in novel ways to misdirection, deception, abuse, and betrayal. Phishing emails and spoof websites are dangers against which one must guard, perhaps not unlike watching out for pickpockets and con men on the streets. But there is another level of risk involved in social networking, where one can lay one's identity open to public view, not just by members of one's immediate community, but the whole, sometimes dangerous, world. The intimate details of one's life found in medical and financial records, diaries kept on one's computer or PDA, personal conversations over the phone or maybe Skype, and the patterns that can be spotted in one's internet activity are sometimes things one does not want even one's best friend to know. I've argued here that establishing appropriate levels and kinds of trust with others is an important factor in personal self-realization. Modern IT forces us to cope with new risks as well as new opportunities, and issues of trust will always be near the center of our concerns.

I have already mentioned the connection between trust and privacy and argued that this is especially significant because privacy has significant connections to the notion of autonomy. Full-fledged persons can exist only in a context in which their autonomy is capable of being recognized and respected, and thus only in a context in which relations of personal trust are at least possible. But IT is particularly relevant to questions of trust for at least one other reason as well. Persons are not and cannot be isolated beings: we are essentially communicative. We have to learn to be persons, and we have to learn it from other persons. The recognition and respect that are partly constitutive of our personhood themselves have an essential communicative dimension: An uncommunicated recognition of another is as good as none. Relations of trust are essential to the constitution of effective communicative systems, which in turn affect the further possibilities and forms of trust. A revolution in information technology drastically changes the possibilities and forms of communication, and these can have unforeseeable effects on the ways in which personhood is constituted and developed.

5. Conclusion

As this paper has progressed, you may have noticed that it has gotten less argumentative and more speculative. My basic claim has been that there is a special kind of trust that is reserved for persons or person-like agencies and is open-ended in its scope. I have made the claim that this form of trust is specifically different from other, more restricted forms, but I cannot claim that I have bullet-proof arguments. It may be that what I have called personal trust is a kind of ideal type of trust, and other forms of trust are understood to be restricted, sometimes even hobbled versions of that ideal.

I have also argued that this form of trust is deeply connected to two crucial aspects of personhood: persons are autonomous yet communicative beings. Relationships of trust help form the scaffold that permits us to hold these two aspects of personhood in balance. I hope I have at least convinced that there is something special in such relations, and that without the richness of personal trust, we would be diminished, lesser beings.

I have not addressed particular issues concerning IT and trust; this essay is intended to help sharpen our thinking about the nature of trust and to show why the trust issues that do arise in the context of IT should not be treated lightly.