are framed makes a big difference. For example, people value gains and losses
differently. Framing the same situation as a potential gain or a potential loss that
is averted may impact the decision greatly. The other well-known factor is the
endowment effect: we will value something that we own much more than an
equivalent item that is not ours. When multiple options are presented, the desirability
of an option may also change when an irrelevant option is added. All of this points
to the inevitable conclusion that utility is not a simple value evaluation: the framing
effects must be considered in trust evaluation [ 9 ].
Almost all trust computation is based on a set of cues that are used to assess the
trust for the trustees with respect to the specific goals. People use these cues daily,
and computational models aim to mimic the human cognitive process. The social
cognition of others is based on many different cues, including their faces, their social
position, the stories we hear about them (social reputation) and our own experience
with them. A similar set of cues exists for text, ranging from the appearance of the
content to the authority of the source. We also form opinions of other entities, such
as computer systems, information systems, or intelligent agents helping us using
many similar cues. We will examine these in detail in the next chapter. We note
that at the cognitive level, some of these cues are very easy to execute while others
require effort. Kahneman [ 9 ] argues that cues that are less costly to process are more
frequently used than those that are more costly. Furthermore, the order in which
we evaluate cues may impact the trust evaluation; for example, when interacting
with a person with an untrustworthy face, it may take us longer to trust them since
trustworthiness of faces is processed much more quickly, and the resulting first
impressions may impact our subsequent evaluations of trust.
Where the trustor's past experience of the trustee is concerned, some models
consider all past experience, while some models consider that the trustor can
forget or forgive various past events. The other important factor to consider is
priming. For example, people who have been reading about social justice may
view an article about wealth distribution differently than those who have been
reading about problems with the welfare system. Priming is a well-known effect and
used frequently by those involved in advertising or political propaganda. Different
external cues may alter the evaluation of information credibility for the same article.
The other thing to remember is that the processing capability of the trustor plays a
role in how much of the relevant input will be considered when making a decision.
When we are busy or tired, we have fewer cognitive resources and may rely on
simpler cues to evaluate trust. For example, relying on Bob's information about the
restaurant's location was quite simple. It did not require Alice to think at all about
whether the location is indeed incorrect by retrieving information about past visits
from her own memory. It was much easier to acquiesce to Bob's opinion, especially
at the end of the day when Alice was tired and hungry.
When considering networking effects, the nature of social relations can also be
relevant. For example, an often debated issue is whether or not trust is transitive or
not. Ultimately, it depends on the underlying social and trust context. For example,
transitive closure is often found in social relations involving close friends. This is
because two of Alice's friends end up hanging out together and become friends.