Abstract
Biased exposure to faces belonging to distinct categories (e.g., race) and global visual deprivation can both impact face recognition. More rich experience with faces leads to more robust performance. Visual recognition is tuned to facial appearance by experience and experience (or a lack thereof) constrains observers’ processing abilities. Here we examined face processing in a population that occupies a unique place between biased exposure and deprivation: Observers from depopulated regions. Does an overall lack of unique faces in the environment lead to measurable differences in behavioral and neural responses to faces? We recruited participants from small (< 1000 persons, N=19) and large (>30,000 persons, N=18) communities to determine how face exposure during development affected face memory and ERP responses to faces. We characterized face recognition abilities using two measures: (1) The Cambridge Face Memory Test (Duchaine & Nakayama, 2006) and (2) N170 responses to upright and inverted images of faces and non-faces (chairs). We anticipated that impoverished experience with faces would lead to poorer behavioral performance and N170 responses that were less face-specific (e.g. a reduced face inversion effect). Accuracy in the CMFT was significantly worse (t(35)=1.98, p=0.028) for small town participants (72.9%) compared to their large-town counterparts (79%), suggesting that behavioral performance does depend on the sheer number of faces available in the environment. Furthermore, N170 amplitudes also exhibited a two-way interaction between stimulus category and group (F(1,32)=4.05, p=0.05) driven by a reduced difference between face and non-face amplitudes in participants from small towns. Besides typical main effects of stimulus category, orientation, and hemisphere, we also found a history of limited face experience has consequences for later neural processing of faces. Face processing is therefore constrained by the richness of the visual environment in terms of facial appearance, even when visual function and the overall visual environment are normal.
Meeting abstract presented at VSS 2015