Abstract
Behavioral studies have shown that in addition to three-dimensional (3D) shape, two-dimensional (2D) surface reflectance information is important for perception of facial identity (e.g., O'Toole et al., 1999; Lee & Perrett, 2000; Jiang et al., 2006; Russell et al., 2006, 2007). In this context, it has been recently reported that inversion equally affects face individualization based on shape or surface reflectance information (Russell et al. 2007). While this can be taken to support the view that inversion affects face processing quantitatively, 3D global shape information variation was relatively minimized in this latter study by cropping of the face contour. Here we reinvestigated this question by means of face stimuli in which natural global shape variations were preserved. To do so, we used a 3D morphable model (Blanz & Vetter 1999) to create paired face stimuli that differed in 3D shape, 2D reflectance, or both. Twenty-four participants performed a delayed face matching task with both upright and inverted faces, in which discrimination between the target and distractor faces was possible based on shape information alone (shape-different), reflectance information alone (reflectance-different), or both (different). Whereas inversion impaired performance for all conditions, the effect was significantly larger when discrimination was on the sole basis of 3D shape information. A second experiment with upright faces showed that a patient with acquired prosopagnosia (PS, Rossion et al., 2003) suffering from holistic face perception impairment performed significantly better in the reflectance-different condition than in the shape-different condition. Altogether, these observations suggest that 3D shape information is more dependent on holistic face processing than 2D surface reflectance. They also support a qualitative view of face inversion (Rossion, 2008) and concur with recent findings of a right hemispheric dominance in processing diagnostic 3D shape cues from faces (Jiang et al., submitted).