Abstract
Finding one's way through a crowd of people is a computationally demanding task. To navigate through crowded environments the brain must segment motion signals generated by the one's own translational movement from those generated by other moving individuals or objects in the scene. Despite the complexity of this problem, humans are able to rapidly and accurately assess dynamic crowds during self-motion, allowing them to move about safely on a day-to-day basis. The purpose of the current series of experiments was to investigate how people estimate their direction of travel in crowds based on visual information alone. Observers were shown scenes in which movement was simulated through crowds of intact or spatially scrambled point light walkers that articulated in place. They were then asked to estimate their heading direction. We found that heading estimates were significantly more accurate when crowds were comprised of intact as opposed to scrambled point light walkers. In a second experiment individual translational components were added to the walkers in the crowd, and the degree of walker articulation was manipulated. Though the addition of walker translation increased heading errors, the finding that heading estimation was more accurate for intact walker crowds than scrambled crowds was replicated. Additionally, we showed that walker articulation produced a reduction in heading errors independently of walker type. Together the results from these experiments suggest that, during locomotion through crowds, the visual system may take advantage of biological motion information in order to reduce spurious motion in the visual scene. Importantly, our results also suggest that the improved heading accuracy observed in the presence of intact walker crowds as compared to scrambled crowds is largely driven by global form cues.
Meeting abstract presented at VSS 2016