Monday November 3rd 2003, 4:00pm
GENDER FROM GAIT REVISITED
It is widely reported that point-light displays of human gait provide information sufficient to recognize the gender of the walker; and this result is taken as evidence of the exquisite tuning of the human visual system to the perception of biological motion. In this paper we revisit the issue of gender recognition from gait with the goal of establishing a baseline for human performance and an estimate of the efficiency with which gender is recognized. The results of a meta-analysis of 20 separate experiments investigating gender recognition revealed 67% correct as an estimate of human accuracy at gender recognition for a side view of walking movements and 71% for views other than the side view. These results can be contrasted with theoretical estimates of optimal human performance based on Center of Moment (Cutting, Proffitt & Kozlowski, 1978) and anthropometric data which showed, consistent across different populations, an accuracy of around 79% correct at identifying gender. Ideal observer analysis provides a means of comparing human and theoretical performance to obtain estimates of human efficiency and results of these calculations revealed efficiencies of 26% for the side view and 48% for the front view (Pollick F.E., Kay J., Heim K. & Stringer R. (2003))
PRESENTED BY
Frank E. Pollick