Proceedings of the Annual Meeting of the Cognitive Science Society

Bayesian inverse planning models have had considerable success in accounting for how humans understand others’ goal-directed behavior. To date, however, this approach has relied on a pre-specified distribution of possible goals, and it isnot clear where knowledge of this goal space comes from. We present an alternative, connectionist model for whichpossible goals are not specified a priori; instead, action predictions is derived from statistical regularities across pastvisual experiences. The model was evaluated by comparing its prediction performance to mouse-tracking data fromhuman subjects in a novel trajectory prediction task. Like humans, the model showed an initial bias for efficient motion,but rapidly adjusted its predictions based on observed trajectories. This pattern of adjustment indicated sensitivity tocontinuously varying ”sub-goals” that were not explicitly provided to the model and could not be attributed to participantsa priori.