occasional errors by Bob. In fact, a well-studied problem in automation is to make
sure people continue to pay attention to the operation of a system considered to
be competent. Inability to monitor such systems appropriately may lead to major
problems in high-risk situations. It is also likely that competence is specific to a
topic. The generalizability of the belief concerning someone's competence depends
on the different problem domains in which someone's ability is observed. Similar
to trustworthiness, it is much more likely take notice of someone's competence if
there is a strong evidence. If Bob is surrounded by many other competent candidates,
Alice might not notice his abilities.
When judging the trust for a trustee, if sufficient information about his compe-
tence is not available, the trustor may use a heuristic and base her evaluation on his
trustworthiness. The reverse is not necessarily true. In fact, the decision to rely on
a trustee who is competent but not necessarily trustworthy may involve a rational
evaluation of risks and potential benefits. A decision based on the trustworthiness of
a trustee may be suboptimal from a rational perspective (as judged by utility theory).
However, when benefits are viewed in the long term, one can find reasons to deal
with a trustee who is trustworthy but not necessarily the most competent. Examples
of such benefits are the ease of communication due to shared vocabulary and past
experience with resolving conflicts, the possibility of benefitting from special favors
and so on.
In short, trustworthiness and competence beliefs are not the same. They consider
different inputs and evaluate evidence differently. In addition, the evaluation of one
belief may impact the other in many different ways.
The Trust Evaluation Process
As is clear from the description in the previous sections, trust evaluation is not a
single event, but a continuous process. On the smallest time scale, we consider
how the processing of different signals relevant to different beliefs interacts with
each other. For example, first impressions formed by the visual appearance of a site
are available before reading the content. In some cases, a trustor may even stop
processing of the content because she does not have enough cognitive resources or
time, or feels she has enough information. The first impressions may significantly
impact the trust evaluations formed by the later inputs.
We also discussed about many different interactions between the different
signals in cognitive processing: paying attention to the highest signal, considering
coherence among signals as a confirmation of their correctness, evaluating later
signals differently due to the way earlier signals created expectations, and so on.
This is studied in great detail in the polling literature [ 2 ] where questionnaires have
to take into account question order, wording of questions, and priming factors that
might impact the answers. People are known to make different decisions when faced