Intentions and intention recognition in intelligent agents

PerMIS 2009 included a special session that explored R&D work using the Theory of Mind (ToM) concept. Simply stated the ToM hypothesis is that intelligent agents attribute mental states to other agents in order to reason in a theory-like fashion about the causal relation between these unobservable mental states and the agents' subsequent behavior [53]. Such theories grow in part out of the consideration of the richness and complexity of primate social interactions, which have long been seen as a driver for the evolution of primate intelligence [34, 33]. Child research also suggests that as human infants develop they use knowledge of their own mental function as a model for how other agents function. When infants see others acting 'like me,' they construct and test a representational correspondence hypothesis that others have the same mental experience generating their behavior [44]. This is enhanced by the regularities of perceptions and actions of social interaction, where others act as if they are governed by a similar type of mind. Having a ToM is readily useful because it affords the possibility of profitably applying judgments, originally made about one's self, to others. The PerMIS session explored whether the ToM hypothesis can be testing and if the concept is useful to the goal of highly competent systems able to achieve goals in a relatively autonomous way [6].