Socio-Cognitive Model of Trust: Basic Ingredients - Trust Theory: A Socio-Cognitive and Computational Model - page 34

', it is also a set of beliefs about the
needed qualities or 'virtues' of Y . Trust, as evaluation, is in fact a model of Y 's qualities and
defects (which define his trustworthiness dimensions ).
As for any kind of explicit evaluation, with trust we are not satisfied by the mere belief
that Y is OK, is 'good for', has the power to achieve goal g X , to execute the action/task

However, trust is not only a belief that Y 'is good for τ

τ
(delegated to him). We try to understand why Y is good for this (while Z is not); to have a
theory of what makes Y able . 8 In other terms, we try to know which kind of characteristics
are useful or required for effectively performing

α /

or achieving g X . And many of them are just
hidden, internal (and mental): kripta (Section 2.2.7).

τ

Qualities and Standards

This applies in general to the Evaluation theory (Miceli and Castelfranchi, 2000).
Given that Y is 'good for' (for example, this knife Y is good for cutting the bread), to

which features of Y should this be ascribed? In the case of the knife: To its plastic or wooden
handle? To its sharpening? To its color? To its being serrate and long? And so on. In several
cases this implies a true causal model (although naive and commonsensical) of

τ

,of Y , and of
what effects it produces. In the example of the knife, it is clear that the plastic handle or the
color are irrelevant, while a good knife for bread needs to be long, serrate, and sharpened.
Those features ( F ) to whom the 'goodness of Y for

τ

' is ascribed are Y 's qualities ( Q ).
Defects ( D ) are those features of Y 's to which is attributed the lack of power, the inadequacy
or dangerousness of Y for

...

.
Notice that a feature of Y that is a 'quality' relative to task

α / τ

τ , can be a 'defect' relative to

another task ' , and vice versa (see Figure 2.2).

Let us also take note of how this theory of 'qualities' and 'defects' is cognitive and prag-
matically very crucial and effective. In fact, while buying a new knife in a store we are not
allowed to have with us a piece of bread and experimentally evaluate which knife is 'good for'
it. Thus, how can we choose a 'good' knife without trying it? Just because we have a theory
of the needed qualities, of what makes a knife a good knife for bread. We just look for these
characteristics; we compare that knife with the 'standards' for a good bread-knife.
Standards are in fact just qualities generalized to the class of the object; the ideal properties
that such a kind of object Omust possess in order to be good as an O .
Qualities (and standards) are necessary not only for recognizing and choosing a good
instance of object O ; that is they are signs of its being good, reliable for

τ

τ (trustworthy) (see

Section 2.2.7 on Signs); but they:

are fundamental also for generalizing trust from one task to another (are for τ needed more
or less the same qualities than for ' ?), or from one agent to another: 'has Z the relevant

qualities remarked in Y ?' (see Chapter 6); and thus they;
are also fundamental for predictions based on general models and classes.

τ

is both 'able' and 'in condition' for realizing g X . For example, in J. J. Meyer's logic (Meyer, 1992) Y 'CanDo' when
both Y is 'Able' and 'In Condition'.
8 While we put the other dimension (the evaluation of the conditions and external resources) in the 'environmental
trust' (see Section 2.10).