they find online. Finally, researchers study how people trust systems that they use
for information or assistance.
Computer science studies how systems can make use of human input for many
different services. There are many algorithms that incorporate a notion of trust,
either implicitly or explicitly, to accomplish this. These algorithms often need
to consider both social and cognitive aspects that impact people's decisions. As
a result, understanding trust from these aspects is crucial to the development of
principled algorithms, which is what we tackle in the next chapter. We then follow
it with a review of trust in computing. However, we note that the study of trust in
networks must incorporate not only an understanding of trust in each field, but also
the dependencies between all the different components.
Defining Trust in Its Context
In the introduction, we defined the trust context as the system level description of
trust evaluation. In particular, the trust context incorporates a number of variables
that are crucial in determining how the trust evaluation unfolds. We will consider
these variables as the elements of trust context that must be defined to indicate
clearly what the context is.
So far, we have discussed a number of contextual differences in trust evaluation.
First of all, the trust evaluation may refer to either a human or a computational agent
trusting another entity ( cognitive trust vs. algorithmic trust ). In this brief, we use
the term cognition to refer to human cognition only, in other words cases in which
the trustor is a human. Clearly, for cognitive trust, one has to consider the impact of
cognitive and social processing on trust evaluation. In the case of algorithmic trust,
the trustor is a program and the algorithmic properties of the agent are a concern.
Ultimately, cognitive trust uses a different system than algorithmic trust. While
some algorithms are designed to mimic cognitive trust, others are not. Furthermore,
the trustor is part of a network that provides her with different institutions that help
or limit her actions in various ways.
Another important element of the trust context is the trustor's goals and the
trustee(s) she depends on for these goals. The difference in the goals is apparent
when the trust is for an action vs. when it is for information. For example, trusting
Bob to perform a task is different than trusting the information given by Bob ( trust
for actions vs. trust for information ). When discussing information trust, we note
that it is different than information credibility. In fact, information credibility is a
factor that may change the processing of information trust: credible information
may not be trusted and trusted information may not be credible. For example, a
highly trusted source may tell us that the building is on fire. The information is
not credible, but the source is trustworthy. When evaluating the information trust,
the trustor typically considers two things: the trust for the source of information
and her own evaluation of the credibility of the message. Both of these are part of