Abstract
A face varies in its retinal properties due to fluctuations in orientation, surface lighting and viewing distance. A person's facial appearance can be altered by makeup or changes associated with age and health status. Successful face recognition therefore requires that the range of face instances be correctly categorized as belonging to the same person. However, the perceptual factors governing this visual categorization process are not well understood. We hypothesized that person representations can be abstracted from a rapid and structured stream of face presentations. Using the RSVP method, participants passively viewed a continuous sequence of 160 different grey scale photographs of four Dutch female celebrities (40 photographs per celebrity). The photographs were centrally presented unmasked every 500 ms. In the blocked condition, the photographs were grouped by celebrity (e.g., 40 images of Celebrity A, 40 images of Celebrity B, etc.). In the mixed condition, the photographs were presented in random order. After two rounds of presentations, participants completed a "same/different" test in which two celebrity photographs were sequentially presented for 500 ms, each followed by a visual mask. Participants responded "same" if the photographs depicted the same celebrity or "different" if the photographs depicted two different celebrities. The test faces were novel and not seen during the presentation phase of the experiment. The main finding was that participants in the blocked presentation condition performed reliably better on the same/different task (d' = 2.21) than participants in the mixed presentation condition (d' = 1.74, p < .05) and participants in a no-presentation control condition (d' = 1.41, p < .001). Performance of participants in mixed and control conditions did not reliably differ, p > .10. These results suggest that the rapid presentation of many face instances, grouped by category promotes the efficient formation of person representations.
Meeting abstract presented at VSS 2016