Definitions of Trust: From Conceptual Components to the General Core - Trust Theory: A Socio-Cognitive and Computational Model - page 11

already achieved, i.e. you will not damage me, but rather let me continue to have what I have.
(This is why we call 'passive' this kind of expectation).
One of the clusters of definitions found by Castaldo is the definition of trust as a 'Belief
about future actions' (27, 8%) that 'makes no reference to the concept of Expectations'. This
is in our interpretation either due to the fact that the authors adopt a very weak notion of
'expectation' and thus simply analyze it in terms of 'a belief about the future'; or, instead, they
have a richer and stronger notion of 'expectation' but expunge this notion from the definition
of trust, ultimately reducing trust to some grounded prediction . In both cases, this position is
unsatisfactory; trust, even when considered as a pure mental attitude before and without any
decision and act, cannot be reduced to a simple forecast (although it contains this element of
prediction within its kernel). Computers, if adequately instructed, can make excellent weather
forecasts, but they do not have 'expectations' about the weather, nor do they put their 'trust'
in it - indeed, they remain incapable of trusting anything or anyone, as long as they are mere
forecasting machines. In what follows, we will take Andrew Jones' analysis of trust (Jones,
2002) as a good example of this tendency to reduce trust to an epistemic attitude, to grounded
prediction, and discuss it in order to show why such an analysis necessarily misses some very
important trait of the notion of trust (see Section 1.4 and Chapter 2).
Some of the most frequent terms highlighted by Castaldo's content analysis, like Coop-
erate, Mutually, Exchange, Honesty, Commitment, Shared Values , are clearly valid for
describing trust only in specific domains and situations, e.g. commerce and organization.
Mutuality , for instance, is not necessary at all: in most contexts, trust can be just unilateral -
and in fact later on we will criticize this same bias in philosophical, economic and game theo-
retic theories of trust. Meanwhile, it is a real distortion of the game theoretic perspective to use
'trusting' as a synonym of 'cooperating' (see below and Chapter 8). Analogously, the terms
Customer, Company, Salesperson, Firm (gathered by Castaldo in the category of 'Subjects')
are clearly domain-specific. The same would hold for the term Security in the growing domain
of 'Trust and Security' in Information and Communication Technologies (see Chapter 12).
1.4 Two Intertwined Notions of Trust: Trust as Attitude vs.
Trust as Act
Although we are aiming for a unified, covering, general and possibly shared definition of trust,
this result will not be achieved by looking for just one unique monolithic definition. We also
do not want to gather just a list of different meanings and uses that share with each other only
some features; we can accept this 'family resemblance' as a possible result of the conceptual
analysis, but not as its starting assumption and its ultimate objective.
Ideally, what we will try to identify is a kernel concept : few common and truly fundamental
features . In doing so, what is needed - as is often the case (see for instance Castelfranchi's
definition of 'agent' in (Castelfranchi, 1998; Castelfranchi, 2000a; Castelfranchi and Fal-
cone, 2003)) - is a layered definition or a hierarchy of definitions, with an explicit account
of the conceptual relationships between the different but connected meanings of the same
notion.
The common sense term trust (at least in many of the languages used by the scientific com-
munity, like English, German, French, Spanish, Italian) covers various phenomena structurally
connected with each other. As we said, it is crucial to distinguish at least between two kinds