The effect of object and event orientation on perception of biological motion

Detection and recognition of point-light walking is re-duced when the display is inverted, or turned upside down. This indi-cates that past experience influences biological motion perception. The effect could be the result of either presenting the human form in a novel orientation or presenting the event of walking in a novel orien-tation, as the two are confounded in the case of walking on feet. This study teased apart the effects of object and event orientation by exam-ining detection accuracy for upright and inverted displays of a point-light figure walking on his hands. Detection of this walker was greater in the upright display, which had a familiar event orientation and an unfamiliar object orientation, than in the inverted display, which had a familiar object orientation and an unfamiliar event orientation. This finding supports accounts of event perception and recognition that are based on spatiotemporal patterns of motion associated with the dy-namics of an event. In point-light displays of human actions, in which only points marking the location of the joints are visible, both events and actors are recognizable. A swirling set of 26 points can quickly resolve itself into a couple dancing a polka. The phenomenal ease of this perceptual achievement belies the sparse structural details in these displays. Each static frame may appear as a constellation of points, bearing no rela-tionship to a human, but when even a few frames are presented in se-quence, a robust form is evident (Johansson, 1973). Interest in this phenomenon may reflect the contrast between the richness of what can be seen and the poverty of the static stimulus. The displays provide sufficient information to distinguish characteristics of the actor, in-cluding his or her identity and sex (Cutting & Kozlowski, 1977; Kozlowski & Cutting, 1977), and of the action (e.g., observers can identify how far an invisible object has been thrown; Runeson & Frykholm, 1983). Explanations of how observers extract form and action from these displays fall into two classes: event-from-for