Content uploaded by Paul Dumouchel
Author content
All content in this area was uploaded by Paul Dumouchel on Oct 15, 2015
Content may be subject to copyright.
Trust as an Actioni
Paul Dumouchel
Ristumeikan University
Core Ethics and Frontier Sciences
56-1 Toji-in Kitamachi, Kita-ku, Kyoto 603 8577
JAPAN
Dumouchp@gr.ritsumei.ac.jp
2
Many scholars view trust as an expectation concerning the future action of other agents.
For example, in Trust a Sociological Theory (1999:25), Sztompka defines trust as “a bet about
the future contingent actions of others.” Sztompka insists on the fact that this bet, or
expectation, is trust only if it has some consequence upon the action of the person who makes
the bet. Suppose that we are watching a baseball match and I tell you: “I bet the pitcher is
going to throw one of his famous curve balls.” I am clearly making a bet about the pitcher’s
future action, but this expectation is not trust because it has no consequences whatsoever for
my own future action. “Trust”, Niklas Luhmann wrote, “is only involved when the trusting
expectation makes a difference to a decision.” (1979: 24) Similarly, when Diego Ganbetta
attempts to summarise the different conceptions of trust in the volume that he edited on that
topic, he tells us that “trust is a particular level of subjective probability with which an agent
assesses that another agent or group of agents will perform a particular action, both before he
can monitor such action… and in a context in which it affects his own action.” (1988:217)
Gambetta, like Sztompka and many others, views trust as an expectation concerning the
future action of some other agent(s) in a context where that expectation has an influence upon
the action of the person who has the expectation. The list of authors who view trust in this
way could be extended, but these three should be sufficient for our purpose. I propose to call
“cognitive theories of trust” theories that construe trust as an expectation about the future
behaviour of other agents. These theories are cognitive in two ways. First, they consider the
uncertain, incomplete and often probabilistic knowledge that an agent has of the future action
of another to be the central element of trust. Cognitive theories of trust also recognise a non-
cognitive element in trust: that the knowledge, expectation or subjective evaluation must be
related in some way with the agent’s decision to act. Sztompka argues that this second
element introduces an active, social and objective dimension to trust. As we have seen, we
should speak of trust only if the knowledge element it contains leads to action or at least to a
decision. Nonetheless, this second element also reasserts the importance of the cognitive
dimension. It says that what explains the agent’s action is the expectation. The uncertain and
imperfect knowledge that an agent has about the future behaviour of others occupies a central
place in the complex of beliefs and desires that explains the agent’s action and that either
causes or motivates him or her to act. Such theories are also cognitive for another reason.
They consider that what makes trust necessary is that we have imperfect knowledge of the
world and especially of the future behaviour of other persons. According to them, if we knew
in advance how others would act we would not need to trust. They see trust as a means
compensating for this lack of knowledge. From this point of view, the role of trust is cognitive
3
inasmuch as it replaces the knowledge that we do not have and allows us to act as if we had
that knowledge.
Despite of their success in many disciplines, including sociology, economics and
philosophy, cognitive theories misrepresent trust in many ways. Imagine that we are playing
tennis together. I have just hit the ball, sending it to the far right-hand corner of the court. I
now rush to the left end of the court, where I believe your backhand return will land. You are
faster than I thought. Having seen or perhaps anticipated my movement, you place the ball at
the right end on my side of the court. I have just lost the exchange. It seems strange to say that
when I ran to the left end of the court I trusted you. Yet this is what the cognitive definition
entails. I had an expectation about your future action and acted accordingly. No one would
think that as a consequence of this exchange, I have now lost trust in you or that I believe you
to be untrustworthy. Clearly, cognitive theories of trust cast their net too wide. There are
many circumstances in which action is guided by expectations about the behaviour of other
agents, most of which do not involve trust. Think of conventions, games of co-ordination in
general and innumerable everyday situations. Consider, for example, the following fictive
dialogue. You have just asked me why I did not phone to let you know I was coming back
earlier:
- I did not phone because I expected you to be out when I left. At 4:00 PM on Wednesday
you usually have your driving lesson.
- …
- Oh! You decided not to go today.
Neither trust nor distrust is involved in my decision not to phone or in my discovery that my
expectation was wrong. Cognitive theories of trust do not contain any criteria that allow us to
distinguish trust from other actions that we undertake on the basis of expectations about the
behaviour of other agents. It is therefore not surprising that in cognitive theories of trust, the
specificity of trust often tends to be lost. Trust comes to be equated with some ill-defined
“social capital” that is essential for co-operation,1 or with moral sentiments in general,2 or
with whatever it is that constitutes the social bond.3 This is true even of authors who go to
great lengths to distinguish trust from other apparently closely related phenomena like
confidence and familiarity.4 However, this lack of specificity and inability to differentiate
trust from other forms of behaviour that rest on expectations about the behaviour of other
agents is not the only or even the main problem of cognitive theories of trust.
1 For example, P. Dasgupta (1988); F. Fukuyama (1995); D. Gambetta (1988); R. Putnam (1992); P.
Sztompka (1999) or most of the authors K. Cook’s (2001) collection.
2 For example: M. Hollis (1998).
3 A. Seligman (1997).
4 Seligman, Op. Cit.
4
Conceived as a subjective probability or expectation, trust is more or less identified
with a judgement about the trustworthiness of another agent. It is this judgement that explains
the agent’s decision to act. The second difficulty that cognitive theories of trust face is
empirical evidence suggesting that this is not the case. It is not the way agents act. In
Autonomy and Trust in Bioethics, Onora O’Neil argues that agents very often trust people
whom they think untrustworthy.5 It is strange behaviour to be sure, but, when we think about
it, perhaps not infrequent. How should it be explained? One possibility is to suggest that such
agents are irrational and motivated by some obscure force to act against their better
judgement, which recommends that they not act co-operatively. However such agents are
incomprehensible. This way of dealing with the anomalous evidence is adequate only if
irrational agents, or circumstances in which agents act irrationally, are rare. Otherwise our
explanatory theories lose their utility. Another possibility is to suggest that it is the evidence
itself that is untrustworthy. It could be argued that even though people sometimes say that
they place their trust in individuals whom they think are untrustworthy, this should not be
taken at face value. Such people will usually add that they do this because they are forced to
or because they have no choice. What they mean is not that they act against their better
judgement but that if circumstances were different they would have acted in another way.
Their level of trust may be low, but that is not the only thing that explains their decision.
Theorists usually argue that when everything is taken into account, such as the cost involved
in not acting co-operatively, the expected benefits, the window of opportunity, and so on, the
agents’ decision to trust is not irrational.
Yet, as Gambetta suspected, this conventionalist strategy for avoiding anomalous
evidence threatens to make trust irrelevant. At first sight, it seems highly reasonable to
consider trust to be only one element in the decision-making process. In order to understand
an agent’s decision to trust we should take these other elements into account. When the stakes
are low and the probability is high that the other person will respond co-operatively, for
example, that you will bring back my book, perhaps because I am your professor and it is in
your interest to do so, I might trust you even if I think that you are not a very trustworthy
person.6 Clearly trust is not the only thing that explains our decisions to act co-operatively;
other elements enter into the process. However, making this insight operational in our model
of rational choice requires that trust be convertible into the currency of any other element that
determines the issue. This means that less trust can always be compensated by more interest,
as in the above example, or conversely that more trust is necessary when interest is a weak
incentive. When the desired outcome me is highly probable, trust has a smaller role to play
5 Onora O’Neil (2002), see also C. Heimer (2001), in particular p. 45.
6 This is essentially the structure of Russel Hardin’s “encapsulated interest” theory of trust, in Hardin
(2001).
5
and when it has a low probability, more trust is required in order for a person to act co-
operatively. All of this is very commonsensical, but the consequence is that in the decision-
making process, trust can always be replaced or compensated for by more interest or by
manipulating the environment in one way or another. In theory at least, trust is never
necessary. The goal of Gambetta’s (1988) paper, “Can we trust trust?”, is to argue that, in real
life, trust cannot be replaced as easily as it can be in our theoretical models.
Actually, the difficulty is more serious than Gambetta thought. It is not only that in
such a model trust can always be replaced or compensated for by any other element that
enters into the decision-making process, but that trust and distrust cannot explain anything:
neither a decision to act co-operatively, nor a decision to defect. According to this model, to
trust in the sense of placing trust, is to act co-operatively on the basis of an expectation
concerning another agent’s future action. Yet, as Gambetta noted (1988; 222), the decision to
act, i.e., to place one’s trust, must be independent of any specific level of subjective
probability concerning the other agent’s action. That is to say, there cannot be a probability
threshold below which the agent always decides not to act co-operatively, or any specific
probability that is identified with trust. This is precisely because trust can be replaced by any
other element in the decision-making process. In cases where our interests lead us to co-
operate, trust is not as important and a lower level of subjective probability will be enough for
me to co-operate. On the contrary, when the interests of agents diverge sharply and the stakes
are high a greater level of subjective probability will be required in order for me to trust.
However as Gambetta himself reminded us in the first paragraph of this paper, cognitive
theories of trust identify trust with the subjective probability that we use to assess whether an
agent will perform a specific action. It follows that to say that the decision to trust is
independent of any level of subjective probability, as Gambetta does (1988; 222), is to say
that the decision to trust is independent of trust! Cognitive theories of trust define trust as an
expectation concerning another agent’s action that is relevant to the decision to act. However
the decision model to which they resort simultaneously requires that the action of the agent
who has the expectation be independent of that expectation!
Finally, viewed as an expectation, trust is very strange indeed because in many cases
it is an expectation that cannot be represented, either by the agent who has the expectation or
by the theorist that attributes it. That is to say, it is not possible to assign any specific content
to the expectation. It is easy to come up with examples where the language of expectation
apparently works. “You told me you would be at the airport at 1:00 PM. I trusted you and on
the basis of that expectation I was there on time to meet you.” Here the language of
expectation works because we can give a definite content to the expectation, namely, “that
you will be at the airport at 1:00 PM”, and it is possible to assign to it a specific probability.
But in many cases, and perhaps in all the cases where trust is fundamental to a relationship,
6
this is not possible. Trust is usually considered a central element of most types of long-term
partnerships. However, long-term partnerships are by definition turned towards an unknown
and unknowable future. The meaning of trust in such contexts is that when faced with
unexpected circumstances our partner will act co-operatively, though we do not know how he
or she will act. How can we represent the future contingent action of another if it takes place
in an open-ended future that is unknown to us? How can we assign a subjective probability to
such an action? What is the content of: “He or she would never do a thing like that”? What is
“a thing like that”? How do you represent “that”? Such statements make sense only if there is
a previous statement that has determined the action that falls into the category of “things like
that”. Yet it is precisely against the ill-defined category of “things like that” events that trust
offers protection.
Let us summarise our efforts so far. Cognitive theories of trust offer a definition of
trust that lacks specificity and does not allow us to distinguish trust from other actions that we
undertake on the basis of expectations concerning the future actions of other agents. Such
theories also face anomalous evidence, namely, the fact that we often place trust in agents
who we believe to be untrustworthy. In their effort to accommodate this evidence cognitive
theories resort to a decision model in which trust cannot explain the agent’s decision to act,
and distrust cannot explain his or her decision not to act. Consequently the agent’s decision
must be seen as independent of trust, yet trust is defined as an expectation that is relevant to
the agent’s decision to act. Finally, in many cases trust cannot be represented as a definite
expectation with a specific content. The language of expectation then becomes metaphorical
to the point of absurdity. What does it mean to say that my action is explained by an
undefined expectation concerning an unknown future?
Faced with these difficulties two routes at least are open to us. One is to abandon
cognitive theories of trust, as I will soon argue that we should, the other is, so to speak, to
“bite the bullet” and to concede that trust cannot be distinguished from any other type of
cooperative phenomena. This last option is pretty much what Michael Bacharach and Diego
Gambetta propose in a recent article “Trust in Signs” (2001). Trust, they claim, is not in its
essential sui generis; it is “a complex phenomena (sic) but one whose main elements are met
in other domains of decision making.” (2001; 175) In other words, there is nothing specific
about trust; the word does not single out a homogeneous class of decision problems or of
cooperative endeavours. Nonetheless, they, somewhat surprisingly, argue that the problem of
trust does not in consequence disappear; rather, it is displaced. The problem of knowing
whether or not an agent will cooperate in a given situation is what our authors define as the
primary problem of trust. They take it for granted in the model they propose that this problem
is solved and advocate a move to what they call the secondary problem of trust. That
secondary problem is that of assessing whether the signs agents give to indicate that they will
7
cooperate are reliable or not. In other words, the central problem of trust is not of knowing
with what probability an agent will adopt a non-dominant strategy in a game of imperfect
cooperation, but that of knowing what type of game we are playing, one where the other
player’s dominant strategy is cooperation or one where it is defection.7 More precisely the
problem of trust, according to them, is that of knowing if the other agent is trying to deceive
us by signalling that he or she is playing a game of one type when in reality the agent is
engaged in a different type of game.
The real difficulty we face in life, according to them, is the reliability of signs. This
move to the secondary problem of trust, they argue, opens up a coherent domain of inquiry
that is common to both human and animal society. “The problem of the truster with respect to
trustworthiness is the problem of the employer with respect to productivity, of the dog with
respect to the belligerence of the hissing cat, and of the predator with respect to the toxicity of
the striped worm.” (2001: 174) In short, our authors propose to replace the problem of trust
with the problem of deception understood as being induced by an unreliable sign to perform
an action one would not otherwise have undertaken. However, as the examples mentioned
above suggest, the problems of trust and of deception do not really correspond to each other.
That is to say deception is a phenomenon with a much wider range than misplaced trust. As
various authors have argued, deception can be to everyone’s advantage,8 among other reasons
because a threat that is not acted upon because it is believed can leave every player involved
better off independently of whether or not the threat would have been honoured, that is to say,
even if some players have been deceived. Understood as the question of the reliability of
signs, the secondary problem of trust reaches far beyond what is usually understood by the
question of trust. Of course Bacharach and Gambetta can claim that this is not a difficulty.
Given that trust is so ill-defined and hard to pin point it is not surprising that a scientific
theory will encompass problems and questions that are not captured by the everyday term.
Perhaps is this true. However, as John Maynard Smith and David Harper (2003)
recently argued, in biology the question of signs among animals is not about deception but
about the fact that signals are on average reliable. The question is not to determine which
7 When they present their model (2001:150-152) the authors distinguish between the players “Raw-
Payoffs” and their “All-in Payoffs”. The first ones they tell us are motivated by “simple self-interest”
while the others are the players’ Payoffs “all things considered”. Given the values they assign to these,
when Raw Payoffs are taken into account, defection is the dominant strategy for both trustee and truster
but when they are replaced by All-in Payoffs cooperation is the dominant strategy for both players. By
definition, these different payoffs define two different games. Therefore the problem becomes one of
knowing which game we are playing: one that is defined by the Raw Payoffs where cooperation is
impossible or one that is defined by All-in Payoffs where cooperation is the only rational solution. It
immediately follows that the “All-in Payoffs” game postulates that the primary problem of trust simply
does not arise. The secondary problem of trust is then that of knowing if the signs that agents give to
indicate that they are trustworthy, i.e. that they are playing a game defined by their All-in Payoffs, and
reliable, given that agents have an incentive to lie.
8 P. Dumouchel (2005); Maynard Smith & Harper (2003).
8
signals are reliable and which are not but to understand why so many of them are reliable.
“Why are animal signals reliable? This is the central problem for an evolutionary biologist
interested in signals. Of course, not all signals are reliable: but most are, otherwise, receivers
of signals would ignore them.” (2003: v) Clearly Bacharach and Gambetta follow a different
agenda and in spite of their claim to the contrary there does not seem to be much difference
between the primary and the secondary problem of trust. “A signal”, they say, “is an action by
a player (the “signaller”) whose purpose is to raise the probability that another player (the
“receiver”) assigns to a certain state of affairs or “event”.” (2001: 159) Given this, it seems
hard to avoid the conclusion that trust in signs is an expectation concerning a future state of
affair and that this theory is just another cognitive theory of trust. It is therefore difficult to see
why this recent proposal should escape any of the difficulties listed above. In fact it does not.
Signs of trustworthiness are only one of the elements that enter into the decision to act
cooperatively. Their absence can be compensated by the fact that the interests of the players
converge strongly.9 In other words, there is no other specificity to the problem of signs
regarding trust, than that which is postulated by our authors as a premise of their model. There
is no secondary problem of trust that is distinct from the primary problem.
In view of these difficulties and in spite of such theories’ success in many disciplines,
I think that we should abandon cognitive theories of trust. Instead of viewing trust as an
internal cognitive or psychological element that explains or motivates an agent’s action, I
propose to start from the action of trusting itself, i.e., from the characteristics of trust as an
action. Unlike expectations or psychological dispositions, actions can be observed in the
world. I think that there are two characteristics of trust as an action that are particularly
important here. The first is that trust comes into play only in situations where the interests of
the agents partially diverge and partially converge. In the relatively rare circumstances where
the interests of the agents either perfectly converge or diverge, trust has no place. In zero-sum
games and in games of pure co-ordination, trust is neither necessary nor useful for reaching
equilibrium points. All that is needed is knowledge of our own interest. In circumstances in
which interests partially converge and partially diverge to trust is to opt for convergence, i.e.,
to chose co-operation. But, as we have already seen, this is insufficient to distinguish trust
from other types of co-operative action in circumstances where there is imperfect
convergence of interests. I prefer to cross the intersection before you because I am in a hurry,
but I would rather let you pass before me than crash into your car. If that happened, I would
get at my meeting even later. Choosing the co-operative strategy of yielding the right of way
does not require trust, simply the knowledge of the speed at which you are coming.
9 Maynard Smith & Harper (2003) report interesting examples of this trade-off in the animal world. See
especially chapter 3 “Strategic signals and minimal cost signals”, pp. 32-44.
9
The particularity of trust as a mode of co-operative action is that by trusting, a person
makes himself or herself vulnerable to the agent who is trusted in a way that would not exist
had the person refrained from trusting. To trust is to act in such a way as to give another agent
power over us. In other words, when I trust I increase my vulnerability to another agent
through an action of my own, and that action is precisely what trust is. If I had not acted I
would not be vulnerable, or at least not as vulnerable to the other agent. For example, when I
trust you with an important secret, I give you power over me, power that you did not have
before. The same is true when I pay now for something that you will deliver tomorrow. I give
you means of harming me that you would not have if I had decided not to pay before delivery.
I trusted you. To trust is to act and not simply to expect because it is the act, not the
expectation, that gives the other agent power over the person who trusts.
To trust is therefore to act in a very special way. It is to act in such a way that as a
result of one’s action another agent gains power over us. In general we do not act in that way
in order to give another agent power over us. I confide in you because I need support and
help. I pay before delivery in order to save 10%. In neither case do I act with the intention of
giving you power over me. That is to say, giving you power over me is not my goal. Rather, it
is something that I also do while doing something else and that I need to do in order to do that
something else. I cannot obtain your help and support unless I tell you my shameful secret. I
cannot get 10% off the price unless I pay before delivery. This is not true of every type of co-
operative behaviour. In a convention, for example, it is instead when I do not act in
accordance with the convention, i.e., when I act non co-operatively, that I expose myself to
sanctions. By acting in agreement with a convention I do not give the other parties in the
convention power over me.
It is important to see trust as action, e.g., an act of my own by which I give someone
else power over me. It is not simply that as a consequence of my action you have gained
power over me. Though this situation can also exist, as such, it does not involve trust because
the consequences of my actions are various and many are unintended. Unaware of the sharp
shooter’s presence, I stepped out into the open and gave him clear shot. It is not because I
trusted him that I died, but because I was ignorant of his presence and perhaps imprudent. To
say that trust is an action means that even if it is not my intention to give someone else power
over me it is nonetheless intentionally that I do it. I do not take a taxi with the intention of
giving the driver power over me, for example to drive me all over town before bringing me to
my destination. Yet, I intentionally put myself in his power. I trust him. The distinction
between doing something intentionally and doing something with an intention dates from
Anscombe and there is nothing mysterious about doing two or more actions simultaneously. I
can both walk in the rain and go visit you, and though it is not my intention to walk in the
rain, but to visit you, it is certainly intentionally that I walk in the rain.
10
In the case of trust, unlike that of walking in the rain, the action requires that the other
agent to whom it is directed recognise it as such. Because we are in the domain of human
interactions, actions are what they are only to the extent that they are perceived as what they
are. Some have argued that the most perfect gift is one given when the recipient not only does
not know who the donor is but is also unaware that what he has found has really been given to
him. Others may think that such a “gift” does not deserve to be called by that name. Whatever
the case may be, all agree that giving when the action remains unknown to the recipient is a
different action from a gift that is public and recognised as such because it has different
consequences in the world. Trust is a specific action that has some very particular
consequences. Among these is the creation of a bond. When I trust someone I impose a form
of obligation upon him or her, namely the obligation not to betray my trust. This obligation
exists only if the other recognises in some way that I have given him or her power over me.10
This means that trust transforms the dynamics of the situation. If I make myself more
vulnerable to you, I obtain in exchange a claim upon your future action. If you break my
trust, I can reproach you for your action. In a prisoners’ dilemma, I cannot reproach you your
defecting unless I have first trusted you. Your action is just what should be expected. Trust
creates a normative expectation in the sense that Niklas Luhmann gives to that expression.11
An expectation is normative if, when it is disappointed, the direction of correction is not to
change the expectation, as in the case of a descriptive or cognitive expectation, but to attempt
to change the world. Even if I was wrong to believe that you would honour my trust, I still
believe that trust should be honoured. Cognitive theories of trust propose that we trust
because we have descriptive expectations about the future behaviour of other agents. From the
point of view of trust as an action, it is because we trust that we have a normative expectation.
Trust is a very important type of action that plays a fundamental role in our lives.
There are many different circumstances where we must put ourselves in the hands of others.
We are brought to give doctors and lawyers power over us for example, but also to place
ourselves at the mercy of friends and lovers, colleagues at work, university administrators,
research consultants, stockbrokers, civil servants, travel agents and, yes, of course, taxi
drivers. We can at times refrain from trusting, which usually entails foregoing some
opportunity, but we cannot completely avoid trusting. Cognitive theories of trust consider that
what makes trust necessary is our limited knowledge of the world and especially of the future
contingent actions of other agents. Viewing trust as an action suggests that what makes trust
necessary is not so much our lack of knowledge about the world as the fact that we depend
10 Some authors like Bacharach & Ganbetta (2001:161) argue that we look people in the eyes to
discover signs of whether we can trust them or not. I think we look at them in the eyes because we want
them to acquiesce that they have been trusted and to recognize the responsibility that goes with it. Put
another way, we are not trying to learn something about them, we are trying to tell them something.
11 N. Luhmann (1985:31-40).
11
upon each other. It is true that we do not have perfect knowledge of the world, but even if I
did know in advance what you will do, it does not follow that I could always avoid giving you
power over me. A child who is confronted with an abusive parent or educator is often forced
to give power to a person whom he or she knows will take advantage of it. What pushes him
or her to do this is not lack of knowledge about the future behaviour of the adult but the
child’s relative helplessness. What motivates a person to give another power over him or her
is that the trusting agent believes that some good will follow from it. That good may be large
enough to compensate for the evil the agent also knows will follow from the action. When
cognitive theories try to accommodate this fact, they are lead to the absurdity that trusting (the
action) is independent of trust (the expectation or moral sentiment).
When trust is construed as an action rather than as a sentiment or internal state, such
absurdities are avoided. The specificity of trust among different forms of co-operative
behaviour is maintained. The evidence that we often place our trust in agents whom we “do
not trust” no longer appears anomalous. Finally, it suggests an interesting avenue of inquiry
into the relationship between trust and various forms of political relations.
Bibliography
Bacharach M. & D. Gambetta. “Trust in Signs” in Trust in Society. K.S. Cook, Ed. New York: Russell
Sage Foundation, 2001, pp. 148-184.
Cook, K. S. (Ed.) Trust in Society New York: Russell Sage Foundation, 2001.
Dumouchel P. “Rational Deception” in Deception in Markets. An Economic Analysis. C. Gerschlager,
Ed. London: Palgrave Macmillan, 2005, pp. 51-73.
Dasgupta, P. “Trust as a Commodity” in Trust Making and Breaking Cooperative Relations. D.
Gambetta, Ed. Oxford, Basil Blackwell. pp. 49-72;
Fukuyama, F. Trust: Social Virtues and the Creation of Prosperity. New York, Free Press, 1995.
Gambetta, D. “Can we trust trust?” in Trust Making and Breaking Cooperative Relations. D.
Gambetta, Ed. Oxford, Basil Blackwell. pp. 213-237.
Hardin, R. “Conceptions and Explanations of Trust” in Trust in Society. Op. cit., pp. 3-39.
Heimer, C. A. “Solving the Problem of Trust” in Trust in Society. Op. cit., pp. 40-88.
Hollis, M. Trust Within Reason. Cambridge, Cambridge University Press, 1998.
Luhmann, N. Trust and Power Trans. H. Davis, J. Raffan & K. Rooney. Chichester: John Wiley, 1979.
Luhmann, N. A Sociological Theory of Law Trans. E. King-Utz & M. Albrow. London: Routledge and
Kegan Paul, 1985.
Maynard Smith, J. & D. Harper. Animal Signals. Oxford: Oxford University Press, 2003.
O’Neil, O. Autonomy and Trust in Bioethics. Cambridge: Cambridge University Press, 2002.
12
Putnam, R. Making Democracy Work. Princeton: Princeton University Press, 1992.
Rivzi, S. A. T. “Deception and Game Theory” in Deception in Markets. An Economic Analysis. Op.
cit., pp. 25-49.
Seligman, A. The Problem of Trust. Princeton: Princeton University Press, 1997.
Sztompka, P. Trust A Sociological Theory. Cambridge, Cambridge University Press, 1999.
.
i Previous versions of this paper have been read at Kinjo University in Nagoya, at the University of
Kyoto and at the 48th Congress of the Canadian Philosophical Association held at the University of
Winnipeg. I wish to thank the participants at those meetings and others for useful comments and
criticism, in particular, Martin Aiken, Mary Baker, Rolf George, Harcel Hénaff, Steven Lukes, Mathieu
Marion, Kenneth Nickel, Masachi Ohsawa, Matthew Taylor, Hiromi Tomita and Kozo Watanabe.
