Abstract
Purpose: What defines visual sensitivity to human movement? Perceptual learning theories emphasize observers' lifetime of experience watching other people move. Motor theories posit that visual sensitivity depends upon input from the observer's own motor system. To what extent does visual sensitivity to human movement depend upon these factors? To address this question, we examined identity perception with point-light displays of human action. Method: Six naïve participants viewed point-light depictions of themselves, friends, and strangers performing various actions. Actors were matched for gender, age, and body size so that these cues could not be used for actor identification. Following a 3AFC naming procedure, participants in Experiment 1 identified the actor in each 5 second movie (self, assigned friend or assigned stranger). In Experiment 2, these same participants performed a 2AFC identity discrimination task. On each trial, two different movies depicting two different actions were displayed. On half of the trials, the same actor performed both actions. On the remaining trials, two different actors performed the two different actions. Participants reported whether the two displays showed the same actor or two different actors. Thus, no explicit naming was required. In Experiment 3, this task was administered using inverted displays. Results: Participants in Experiments 1 and 2 were most accurate in recognizing their own movements. Participants also demonstrated greater visual sensitivity to the movements of their friends than to the movements of strangers. Identity perception was action dependent. Performance dropped significantly with display inversion. Conclusion: These results suggest that both motor and visual experience define visual sensitivity to human movement. Furthermore, attempts to build automated identification systems based on the analysis of human gait may be ill fated since identity perception is least accurate for walking