Abstract
The face inversion effect (FIE) is characterized by an important drop in recognition performance when facial stimuli are rotated by 180° in the picture plane. Pachai and coll. (2013) showed that inversion disrupts the processing of horizontal information (see also Goffaux & Dakin, 2010) and reported a significant positive correlation between horizontal tuning and the magnitude of the face inversion effect. Recently, Van Belle & Rossion (2015) showed that face inversion reduces the size of the perceptual field of view (PFV). This offers an elegant explanation for the performance drop with inverted faces since a small PFV restricts feature extraction to only a few (maybe one) at a time; a proposition reminiscent of the holistic hypothesis. To make the link between the lack of horizontal tuning with inverted faces and the PFV hypothesis, we measured orientation tuning in five participants for upright faces presented either through a small aperture (a gaze-contingent approach), or as a whole. First, the participants were asked to learn the face-name association for 10 identities. They practiced in each condition until they reached an accuracy of 95%. In the second phase, images were randomly filtered in the orientation domain with orientation bubbles (Duncan et al., 2014) to precisely reveal orientation utilization. Participants performed 400 trials per condition. The signal-to-noise ratio was adjusted so that the same performance level (55%) was obtained in both conditions. Congruently with what was observed for FIE, the signal-to-noise ratio was significantly higher when faces were presented through a small aperture than as a whole [t(4) = 12.9, p < 0.001]. Despite this large effect, the small aperture condition is not linked to a decrease in horizontal tuning. Our results show that the smaller PFV associated with the FIE cannot explain the lack of horizontal tuning with inverted faces.
Meeting abstract presented at VSS 2016