Abstract
The spatial relationships between the features of a face (known as second-order configuration) have been thought to play an important role in the recognition of identity. Despite such claims, however, behavioural studies have shown that geometric distortions that markedly affect this spatial configuration of a face image often have only a limited impact on judgements of identity. The aim of this study was to use fMRI to probe the sensitivity of face selective regions to geometric distortions that affect the spatial configuration of faces. In Experiment 1, participants viewed blocks of face images with same or different identity. There were 4 image manipulations: (1) unchanged, (2) linearly scaled, (3) linearly distorted (vertical stretch) or (4) non-linearly distorted (top half - unchanged, bottom half - vertical stretch). In the FFA, significant adaptation (different > same) was evident to unchanged, linearly scaled and linearly distorted faces. However, there was no adaptation to faces that were non-linearly distorted. These results suggest that linear distortions, which significantly affect the configuration of the face, continue to activate overlapping populations of neurons within face-selective regions. In Experiment 2, we explored sensitivity of face-selective regions to the more subtle types of non-linear change in the spatial configuration of faces that occur naturally across different individuals. We measured the response to blocks of composite images with the surface texture of one face, combined with the spatial configurations of different faces. Despite the smaller magnitude of these non-linear changes to the spatial configuration, no significant adaptation was evident in the FFA. This suggests that faces that have small non-linear differences in their spatial configuration activate non-overlapping populations of neurons in the face processing network. This differential sensitivity of face-selective regions to linear and non-linear changes in the spatial configuration of faces provides an important insight into the neural mechanisms underlying recognition.
Meeting abstract presented at VSS 2013