Proceedings of the Annual Meeting of the Cognitive Science Society

A recent approach based on Bayesian inverse planning for the“theory of mind” has shown good performance in modelinghuman cognition. However, perfect inverse planning differsfrom human cognition during one kind of complex tasks dueto human bounded rationality. One example is an environmentin which there are many available plans for achieving a specificgoal. We propose a “plan predictability oriented model” as amodel of inferring other peoples’ goals in complex environ-ments. This model adds the bias that people prefer predictableplans. This bias is calculated with simple plan prediction. Wetested this model with a behavioral experiment in which hu-mans observed the partial path of goal-directed actions. Ourmodel had a higher correlation with human inference. We alsoconfirmed the robustness of our model with complex tasks anddetermined that it can be improved by taking account of indi-vidual differences in “bounded rationality”.