Abstract
There is a clear perceptual component to Congenital Prosopagnosia (CP). Given a triangular array of three faces in a minimal match-to-sample test (Fig. 1), those who have difficulty in matching which of the two highly similar lower (test) faces is an exact match to the upper face (the sample), also score at the prosopagnosia end on all the standard tests of face recognition proficiency, such as the CFMT. But there is another perceptual component to face perception: invariance to viewpoint. Is it the case that those who have greater difficulty in discriminating small metric variations of faces at the same orientation, also have greater difficulty in matching faces when their orientations in depth differ? (Fig. 1, right panel.) Subjects discriminated triangular arrays of either three faces or three geons, one of which was an exact match to the sample (Fig. 1). Sample and matching stimuli could be at the same orientation in depth or differ by 13°. The underlying differences in the stimuli were metric variations of the spacing between face parts and cheekbone height. Geons varied in the curvature of the axis or convergence of the sides. Dissimilarity of the foil to the matching stimulus was equated by Gabor similarity for faces and geons. Across subjects (n=14), RTs at a 0° difference in face orientations correlated positively with the increase in RTs at a 13° orientation difference, indicating that those who have difficulty in identifying subtle differences in the physical appearance of faces all at the same orientation, demonstrate less invariance at identifying faces at different orientations (Fig. 2). Importantly, this effect was only present for the faces, not for the geons. These deficits in discriminating faces thus do not reflect a general deficit in discriminating metric differences between stimuli.
Meeting abstract presented at VSS 2018