Abstract
Models of face processing propose that the fusiform face area (FFA) plays an important role in processing invariant aspects of faces such as identity. Recent studies have shown that patterns of response in the FFA can successfully decode individual identities across different viewpoints. However, other studies have reported that patterns of response in the FFA can successfully decode different viewpoints across different identities. Thus, the relative role of identity and viewpoint on the pattern of neural activity in the FFA remain unclear. Here, we used fMRI and a correlation-based MVPA to explore how identity and viewpoint of faces is represented in the fusiform gyrus. A 3 x 3 design was used with faces from three different identities shown from three different viewpoints. The faces from each identity were familiar to participants across the different viewpoints. Images from each condition were presented in a blocked design. To test the contribution of identity and viewpoint to the neural responses, we used a representational similarity analysis in which model correlation matrices were generated that represented the extreme cases where the patterns of response are entirely predicted by the identity or by the viewpoint. These models were then used in a multiple regression analysis of the fMRI data. The results showed that viewpoint explained significantly more variance than identity in the fusiform gyrus. Next, we asked whether low level visual properties could explain these patterns of neural response. A strong positive correlation between the neural patterns and the underlying low-level image properties was evident in the fusiform gyrus. Our results suggest that the differential role of viewpoint and identity might reflect differences in the image properties conveyed by these facial cues.
Meeting abstract presented at VSS 2016