Achievement

Trainees Pantelis, Cholewiak, Sanik, Weinstein, and Affiliate Wu (Psychology and Computer Sciences) studied judgments of intentionality in a virtual environment created by the students. Understanding the perception of intentions is important, particularly as humans do more of their interacting within virtual worlds. The virtual agents navigate, eat, and initiate or avoid attack. Human subjects accurately classified the agents' mental states (52% correct, twice chance level), which is quite good given that the state of the agent was inherently hidden. When allowed to invent behavioral categories, the perceived state transitions were correlated with the actual transitions of the agent. The results show that fundamental rule-based perceptual systems underlie the perception of intentions, and these can and should be exploited by designers of effective virtual worlds. ruccs.rutgers.edu/~jacob/demos/imps.html. (2011 Proceedings of the Cognitive Science Society).