Dynamics of Trust - Trust Theory: A Socio-Cognitive and Computational Model - page 165

Z 's potential influence in the task ev( is performed by Y (following X 's opinion).

This factor gives account of the different potential relationships among the (evaluating and
evaluated) agents that can have an influence on the objectivity of the judgment of evaluation.
Notice that this pseudo-transitivity depends on subtle cognitive conditions. In fact, it is not
enough that X trusts Z for adopting its trusting attitude towards Y ; for us there is no real
transitivity in trust: that X trusts Z and Z trusts Y does not imply that X trusts Y . Don't forget
that trust is 'about' something; X 's trust for Y is to do with some power, action, service. It has
four arguments: not only X and Y but the task/goal and the context. About what does X trust
Z ? And about what does Z trust Y ?
Suppose for example that X trusts Z as a medical doctor but considers Z a very impractical
person as for business, and X knows that Z trusts Y as a stock-market broker agent; X has no
reason to trust Y . On the contrary, if following XZ is a good doctor and he trusts Y as a good
nurse, X can be learning to trust Y as a good nurse. What does this mean? This mean that X
trusts Z as for a given competence in a given domain (Castelfranchi and Falcone, 1998b): Z is
considered by X an 'authority' (or at least a good evaluator, expert) in this domain, this is why
X considers Z 's evaluation as highly reliable ( Z is a trustworthy 'source' of evaluation).
So, since X trusts Y on the task

τ ) when τ

τ we should have:

DoT XY τ = Inf X ( ZYe v ( τ )) ∗ DoT XZe v ( τ ) >σ 1 ,

where .

We have a pseudo-transitivity (we consider it as cognitively-mediated transitivity) when:

σ 1 is the X 's reasonable trust threshold for delegating τ

1) Inf X (Z Y ev( τ )) = 1, in other words, there is no influence on Z by Y performing τ (following

X ) that could produce a non-objective judgment; and
2) DoT XZev( τ ) = DoT XZ τ , in words, X has the same degree of trust in Z both on the task τ and

on the task ev( τ ) ; in other terms, X attributes the same DoT to Z both performing τ and

trusting another agent about . This is a very common case in human activities, very often

due to a superficial analysis of the trust phenomenon.

τ

If the two above conditions are satisfied we can say that each time

DoT ZY τ >σ 2 and DoT XZ τ >σ 1 we will also have DoT XY τ >σ 1 .

In Table 6.5 we show the various possibilities of X 's trusting or not of Y and how the two
DoTs are combined.

Conformism

This mechanism is not based on a special expertise or authority of the 'models': they are not
particularly expert, they must be numerous and just have experience and trust: since they do,
I do; since they trust, I trust.
The greater the number of people that trust, the greater my feeling of safety and trust; (less
perceived risk) the greater the perceived risk, the greater the necessary number of 'models'
(Figure 6.12).
A good example of this is the use of credit cards for electronic payment or similar use of
electronic money. It is a rather unsafe procedure, but it is not perceived as such, and this is