Learning & Memory/Cognitive

Abstract:  Human observers are able to perform a variety of perceptual tasks using only the relative motion of a series of points placed upon another person's body (e.g., determining a person's walking direction, gender, identity, emotional state). A pervasive assumption that is made about such 'point-light' stimuli is that they are highly informationally impoverished relative to their original, full-figure counterparts. That is, transforming a full body's motion into a small series of moving points is assumed to discard massive amounts of discriminative information from the stimulus. A second assumption that builds upon the first is that our ability to perform various tasks using only point light motion implies that we must be highly efficient at using information in such displays, because so little information is (assumed to be) available to us. In this talk, I will describe some surprising results from a series of experiments in which I used ideal observer analysis to directly address these ubiquitous--but previously untested--assumptions.