Abstract
The surface properties that constitute 3D facial structures are key for identity recognition. Johnston, Hill & Carman (1992, Perception 21(3)) showed that inversion effects on identity recognition, where error rates for identifying inverted compared to upright faces are greater, are less marked when inverted faces are illuminated from below. The authors suggest that illumination from below disrupts 3D shape information processing, while spatial inversion disrupts configural processing, such that both manipulations do not have an additive effect on facial recognition impairment. The present study utilises a gauge figure task to measure 3D face representation across four conditions (face lit from above vs below, and rotated vs normal face) in 10 participants. Participants adjusted the slant and tilt of a gauge figure so that it appeared flush against surface of a face, at locations on a regular triangular grid. These were then used to create a 3D map of the perceived 3D structure for each observer and condition, which were compared to the ground truth structure of the face. There was a significant difference in perceived shape for the lit below conditions, but for not the rotation. However, after conducting an affine transformation on the data, there were no remaining differences in residual errors across conditions. These results suggest that the disruption of recognition as a result of lighting from below reflects not a change in the surface-based representation of the face, but higher-order recognition processes.