Socio-Cognitive Model of Trust: Basic Ingredients - Trust Theory: A Socio-Cognitive and Computational Model - page 45

Moreover, this is also the deep meaning of the other component: willingness, intention,
disposition, motivation of Y that makes that prediction grounded.
In other words, in social trust the necessarily internal causal attribution of the trust requires
an intentional stance towards Y (Dennett, 1989). Trust is a disposition (and an affect) and a
social decision only possible in agents able to assume an 'intentional stance', that is to ascribe
a mind and mental stuff to the other agent . The prediction of the other's behavior is based on
some theory of mind or on some projection/ascription of internal factors (including moods
and emotions) playing a causal role in the activation and control of the behavior.
The fact is that in humans those qualities are internal in a psychic sense; thus - by definition -
they are unobservable . They can just be aduced by external signs. They are 'kripta' ((Bacharach
and Gambetta, 2000), (Bacharach and Gambetta, 2001)); but it is precisely in 'kripta' that we
trust and on 'kripta' that we count on when we 'trust' somebody; not the 'manifesta' (visible
signs). Trust is a hermeneutic and semiotic act. We trust the signs only for being reliable, for
informing us about the 'kripta', but at the end it is of /on/for the 'kripta' that we trust.
'Manifesta', signs, are not only the direct observable behaviors or markers of Y , but also all
the other sources of trust (of trust beliefs) - like experience , reputation , certificates , category ,
role , and so on. All of them are direct or indirect manifestations of how Y is or 'works', of
what makes him trustworthy.
Our favorite example of weak delegation (Section 2.6.1) makes our claim clear: I'm running
to take the bus, the driver and the people cannot see me, but I see that there are people at the
bus stop apparently 'waiting for the bus' (notice that this is just an mental ascription), on such
a basis - attributing them with the intention to take the bus and thus the intention to stop the
bus - I rely on them, I trust them for this.
While a simple prediction that somebody standing at the bus stop might raise their arm
would be based on simple observed frequency and probability, we deny that trust in Y to stop
the bus is just this. It is based on the interpretation of Y as 'waiting for the bus' and on the
ascription of the intention that will cause the action on which one relies. This is the - never
clarified - gap between mere subjective probability and trust.
In Chapters 3 and 8 we will criticize the reduction of trust to subjective probability, arguing
about the importance of explicitly distinguishing internal and external components of the
prediction; and also explaining how crucial and inherent for real trust are the bases (beliefs)
supporting the prediction. We do agree with Williamson (see Chapter 8): if trust is just a
euphemism for 'subjective probability' we do not need this term, we already have a strong
theory of it, and a new vague and merely evocative term is just a confusing layer. On the
contrary, we believe that trust is a specific, well-defined, mental and social construct .
On the basis of this 'internal attribution' ('design' or 'intentional' stance) foundation of
trust we are able to account for several things.
For example in Section 2.7.2 we argue that there is internal versus external trust and explain
why it is important to differentiate them. Consider a user working (collaborating, negotiating,
etc.) on the web with other users, she has to distinguish her trust in these other potential
partners from her trust in the actual context: the infrastructure with its internal constraints and
rules. We also show that trustee and context have different dynamics; etc.
However, it is important to make clear that a given trust is internal or external only 'relative'
to the assumed entity. While changing the target, again the distinction applies and again a form
of internality is needed. If we consider for example our trust in the technical infrastructure
(that relative to the partner Y , was an 'external' condition), we are now necessarily doing some