with two different ways to phrase the same question. They are more likely to avoid
something framed as a loss than pursue something framed as a gain [ 1 ].
This type of processing happens continuously at many time scales.
In Chap. 1 , Alice read several different online reviews, but the first review she read
formed a baseline for her. As a result, she did not trust any review that contradicted
the first review. Correctly or incorrectly, she had formed an opinion and was
much more willing to trust information that confirmed her opinions. Overall, trust
evaluations are continuously changed as different inputs become available, and
new inputs are assessed as more processing time becomes available. Often, trust
evaluation is viewed as a simple process, but this is far from the truth. As the
trustor explores the information space, her knowledge of the problem domain may
change considerably. This may impact her ability to process the credibility of new
information she encounters.
Viewing trust evaluation as a continuous process instead of a single event brings
new modeling considerations. By this, we do not mean the usual trust belief
formation process. Typically, trust beliefs evolve slowly as new information about
the trustee becomes available. These beliefs serve as a baseline in trust evaluation.
At the same time, one can form impressions of the other person fairly quickly based
on the available evidence. Impressions are situational and can change rapidly as
the trustor gets new information about the trustee. When modeling these situational
components, we must remember that the current trust impressions may be an input
to the next phase of trust evaluation.
In this chapter, we outlined a framework for describing important components of
trust context: those elements of trust that are situational and impact how trust is
computed. Clearly, this definition does not capture all possible elements of the
trust context. Some elements are almost undefinable or unknowable except through
proxies: the intentions of the trustor, their knowledge of the world and so on.
These can be modeled through different proxies. Our emphasis in this brief has
been on the elements of context that have been studied in social and cognitive
psychology, and in computer science and engineering. These can serve as a starting
point for introducing new modeling methodologies to trust context and trust models
in particular. The framework is given as an outline in keeping with the survey nature
of this brief. The next step is to define the appropriate formalisms to incorporate
these modeling concerns into different interpretations of trust, and to test them in
new applications through different means (Fig. 5.1 ).