15 October - Clark Barrett UCLA Anthropology The Evolutionary Origins of Mindreading     One thing that humans are good at is predicting the behavior of other living things. We do it so often and so effortlessly that we scarcely notice ourselves “mindreading”: automatically computing the goals, motivations, and intentions of others. Why are we so good at this, and how do we manage to do it?
    In recent years there has been an explosion of interest in this phenomenon. Many studies have focused on one component of mindreading – the ability to understand true and false beliefs, a sophisticated skill that emerges relatively late in childhood – but there have been fewer investigations of more fundamental building blocks of behavior prediction on which such advanced skills may rely.
    In this talk I argue that the first step in behavior prediction is solving a particular kind of frame problem: knowing what kind of interaction one is engaged in. This is a crucial step because it prunes the enormous tree of possible predicted behaviors to a single, manageable branch. It is made possible because interactions can be sorted into natural kinds, each with its own set of predictive principles and diagnostic cues. I briefly sketch the results of several studies of the conceptual and perceptual machinery that helps us to solve this frame problem, with particular reference to predator-prey interactions, and discuss implications for the understanding of human cognitive evolution.