Rafael Heitor Bordini has written
>The problem is that a basic principle in MAS is that agent's mental state are
>private, so to speak. That is, agents do not have direct access to other
>agents' minds. Besides, they may all have incomplete knowledge of the
>environment etc., and therefore may have different views on things.
That all is very true. However, many of us have also modelled the
thought processes of groups of agents and of individual agents. In
such models, one tries to include the models of _other_ agents being
resident in agent minds. These models are to be updated dynamically
(N.B. learning agents) and, in spite of their possible shortcomings,
can be made good use of in, for example, predicting agent actions in
response to the environment.
Nick Findler
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|