UC Berkeley

Faces play a critical role in our everyday lives. The resulting lifelong experience with faces has made humans remarkable experts in face recognition. One of the main advantages of recognition of faces over other objects is that we process faces as a whole, a face-specific mechanism called holistic perception. Previous work suggests that there is significant variability in face recognition abilities among individuals and across different locations in the visual field. However, the origin of these observer- and location-specific differences remains unclear. In this dissertation, I hypothesized that these differences in face recognition are a result of idiosyncrasies in stimulus-specific holistic processing. To address this, three studies were conducted. First, I tested whether there are individual differences in holistic face recognition and whether these are stimulus-specific. The results indicated that there are substantial stimulus-specific individual differences: the faces that are recognized holistically by one observer are not the same for another observer. Second, I investigated whether holistic face perception is stronger in specific visual field locations and found that sensitivity to faces was relatively tuned toward the fovea. Given the stimulus-specific, observer-specific, and visual-field specific face recognition tuning found in Chapters 2 and 3, I hypothesized that the underlying mechanism may be observer-specific holistic templates, which humans use to detect and recognize faces. The goal of Chapter 4 was to measure and visualize these individual face templates using a classification image approach. The results revealed that there are idiosyncratic differences in face templates that support holistic face recognition and that these are specific to individual observers. Together the results shed light on the existence and nature of individual differences in face recognition, and emphasize the importance of taking into account individual differences in vision research more generally.