Abstract
We examined the spatial limits of face processing by measuring the spatial spread of the face aftereffect (FAE) following adaptation to a single face and following simultaneous adaptation to a face and its anti-face at different spatial separations. In the adaptation phase, one of the following sets of stimuli was presented with each stimulus placed at locations around a 3 deg circle centered at fixation: 1) a single face; 2) a face and its anti-face on opposite sides of fixation; 3) two faces and two anti-faces evenly spaced; and 4) one to four ellipses evenly spaced with the average size and color of face stimuli (non-adapted condition). In the test phase, a face stimulus with various morphing levels between the face and the anti-face was presented at a random location around the display circle and subjects had to report whether the test stimulus was seen as face or anti-face. The adaptation effect was determined from the PSE in the psychometric function for all angles between adapt location and test location to provide a spatial map of the spread of the FAE. Results show non-retinotopic adaptation - an FAE of similar strength at all locations — following adaptation to a single face stimulus. Nevertheless, simultaneous adaptation to both the face and the anti-face, spaced at opposite locations across the fixation, did not cancel, producing instead two separate regions of opposite aftereffects. With the smaller separation of the 4 equally spaced stimuli, faces alternating with anti-faces, FAE was greatly reduced.
These results suggest: 1) the spatial extent of face analysis and subsequent FAE has a dynamic nature and depends on the number of objects presented in the visual field, 2) when spaced too closely (one per quadrant), the FAE from adjacent faces is pooled and cannot survive independently.