A Fuzzy Implementation for the Socio-Cognitive Approach to Trust - Trust Theory: A Socio-Cognitive and Computational Model - page 323

less accurate but saves the time of analyzing all the incoming messages. Furthermore, if
some busy agent accepts only a limited number of tasks, or if there is a limited time span
for performing the maximum number of tasks, it is important to be quick to delegate them.
11.14.4 Performing a Task
When an agent receives a delegation, it tries to perform the assigned task. Performing a task
involves three elements: two are features of the agent (task specific ability and willingness).
The third is the (possible) external influence of the environment. In order to be successful, an
agent has to score a certain number of hits (e.g. 3 ); a hit is scored if a random real number
in (0, 1) is rolled that is less than its ability score. The agent has a number of tries that is
equal to ten times its willingness value, rounded up (i.e. from 1 to 10 essays). The environment
can interfere with an agent's activity giving a positive or negative modifier to each roll (so it
interferes with ability but not with willingness). If the number of scored hits is sufficient, the
task is performed; otherwise it is not.
The rationale of this method of task resolution is that, even if the tasks are abstract in our
simulations, they semantically represent concrete ones: they involve a certain (controllable)
amount of time; they are 'cumulative', in the sense that the total success depends on the success
of its components; they can be achieved at different degrees (in our experiments the number of
hits is used as a threshold of minimum performance and after being successful the agent can
skip the other essays). Moreover, and most importantly for our theoretical model, the contribute
of willingness (persistence) is clearly separated from ability; for each task 'attempting' is a
prerequisite of 'doing' and an agent can fail in either. The contribution (positive or negative)
of the environment is limited to the second phase, representing a favorable or unfavorable
location for executing the task. The duration of the task is used to introduce another crucial
factor that is monitoring : a delegator has to be able to control the activity of the delegee and
possibly retire the delegation if it is performing badly; this aspect will be introduced in one of
the experiments.
In the simulations we used 3 hits as a default ; however, all the effects are stable and do not
depend on the number of hits. We have performed experiments with a different number of hits,
obtaining similar results: choosing higher values leads to less tasks performed on average, but
the effects remain the same.
This kind of task performing highlights both the role of ability and of willingness
(persistence).
11.14.5 FCMs for Trust
For the sake of simplicity we have assumed that each cognitive agent has access to all true
data of the other agents and of the environment; these data involve their task specific ability ,
their willingness and their current environment . All these data are useful for many strategies:
for example, the best ability trustor always delegates to the agent with the higher ability value
for that task.
The cognitive trustor , following our socio-cognitive model of trust, builds an elaborated
mind model of all the agents; this is done with Fuzzy Cognitive Maps, as described in