Dynamics of Trust - Trust Theory: A Socio-Cognitive and Computational Model - page 183

accept A 's tacit exploitation and adopt A 's goal, or negatively react to that. A can be aware
of and take into account the effect of its own decision in the very moment of that decision.
This make the decision of social trusting more complicated and 'circular' than a trivial
decision of relying or not on a chair.
b) How trust creates a reciprocal trust, and distrust elicits distrust ; for example because B
knows that A's is now dependent on it and it can sanction A in case of violation; or because
B believes that bad agents are suspicious and diffident (attributing to the other similar bad
intentions) and it interprets A's trust as a sign of lack of malevolence. We also argued that
the opposite is true: A 's trust in B could induce lack of trust or distrust in B towards A ,
while A 's diffidence can make B more trustful in A .
c) How diffuse trust diffuses trust ( trust atmosphere ), that is how A 's trusting B can influence
C trusting B or D , and so on. Usually, this is a macro-level phenomenon and the individual
agent does not calculate it. We focused on pseudo-transitivity arguing how indirected or
mediated trust always depends on trust in the mediating agent: my accepting X's evaluation
about Z or X's reporting of Z's information depends on my evaluation of Z's reliability as
evaluator and reporter. In other words this is not a logical or automatic process, but it is
cognitively mediated.

We also discussed a more basic form of trust contagion simply due to diffusion of behaviours
and to imitation because of a feeling of safety. Usually, these are macro-level phenomena and
the individual agents do not calculate it.

d) How trust can be transferred among agents on the basis of generalization of both tasks
and agent's features , that is how it is possible to predict how/when an agent who trusts
something/someone will therefore trust something/someone else, before and without a
direct experience. It should be clear that any theory of trust just based on or reduced to a
probability index or a simple measure of experience and frequency (of success and failure)
cannot account for this crucial phenomenon of a principled trust transfer from one agent to
another or from one task to another. Only an explicit attributional model of the 'qualities'
of the trustee that make her 'able' and 'willing' to (in the trustor's opinion), and of the
'requirements' of τ as related to the trustee's qualities, can provide such a theory in a

principled way.

Our cognitive model of trust, with its analytical power, seems able to account for the
inferential generalization of trustworthiness from task to task and from agent to agent not
just based on specific experience and/or learning. It should also be clear how important and
how productive this way of generating and propagating trust beyond direct experience and
reputation should be.

References

Baier, A. (1994) Trust and its vulnerabilities in Moral Prejudices , Cambridge, MA: Harvard University Press,
pp. 130-151.
Barber, S. and Kim, J. (2000) Belief Revision Process based on trust: agents evaluating reputation of information
sources, Autonomous Agents 2000 Workshop on 'Deception, Fraud and Trust in Agent Societies' , Barcelona, Spain,
June 4, pp. 15-26.