Similar considerations exist when the trustee is a computational agent or a
robot. Alice needs to trust the robot, the specific piece of hardware that she is
interacting with, to be in good operating condition. Furthermore, she needs to trust
the underlying programming of the robot to be capable and trustworthy.
In short, similar to the case of information trust, for a specific trusting choice,
there might be multiple goals and trustees. In such a case, the trustor depends on
each trustee to some degree. Each trustee has some level of control over the outcome
of the goal through a specific subgoal.
Trust Constructs: Trustworthiness and Competence
As we have seen in the discussion of information and action based trust, the
trustor typically has a complex goal involving subgoals. Each subgoal defines
the evaluation of specific aspects of the trustee. Trust research has generated a
long list of keywords used in this effort, such as goodness, morality, benevolence,
expertness, credibility, predictability, dependability, etc. There have been many
efforts to categorize these considerations into distinct types [ 11 , 17 ]. We will review
these in the next chapter.
In the case of trusting actions, an often-used categorization separates trustwor-
thiness from ability.
Trustworthiness refers to our expectation that a person will do as they say. This
term incorporates dimensions like the trustee's integrity and good intentions.
Ability is the trustee's competence in accomplishing a specific task (Fig. 2.6 ).
A trustworthy person may be dependable for completing a task, but their
performance may not be satisfactory if they are not capable.
In reputation management, a system designed to force people to act in a
trustworthy manner may sanction bad behavior [ 2 ]. However, sanctioning alone is
not sufficient to separate competent but untrustworthy people from incompetent but
trustworthy people. To assess the ability of people, online reviews have been found
to be more useful [ 3 ]. These are called signaling mechanisms. For trustworthiness,
tracking bad behavior is considered more diagnostic, while for ability, tracking and
accumulation of good behavior is more meaningful. This agrees with research that
explains how people form opinions of others [ 5 ]. Trustworthiness is a general
measure of the trustee's reliability and is not dependent on a specific topic. If
someone is our friend, we expect that he will do what he says and will help us
when we ask for it. However, ability is dependent on a topic: a doctor's competence
is in medicine and a mechanic's is in car repair.
Some research also concentrates on a third type of trust construct. In this case,
the intentions of the trustees are considered [ 14 ]. Even if Bob does not act in a
way that Alice prefers, he may still have Alice's best interests at heart. Sometimes
these are called good intentions . An adversary with bad intentions may act in a
trustworthy way to achieve advantages that he may later use against the trustor. In
cognitive psychology, the evaluations of another's intent are studied as part of the
theory of mind. According to this theory, people learn about other's beliefs and