Functional Relationship Between the Left and Right Fusiform Face Areas

M. Shu,Z. Li,C. Cheng,M. Meng
DOI: https://doi.org/10.1167/13.9.176
2013-01-01
Journal of Vision
Abstract:The hierarchical relationship in the ventral visual pathway, e.g., from occipital face area (OFA) to fusiform face area (FFA) has been studied to analyze neural stages of face processing. Interestingly, studies have also shown the hemispherical asymmetry of face processing. The left FFA appears to be most heavily involved in analyzing image-level face-semblance, while activity in the right FFA correlates with categorical perceptual decision of whether a visual input is a face (Meng, Cherian, Singal, & Sinha, 2012). However, it remains unknown whether the processes of the left FFA and those of the right FFA occur in parallel or whether the two are serially dependent. Using a collection of 300 stimulus images (including 60 random nonface images, 180 false alarms from a computer face detection system, and 60 genuine face images), we evaluated the strengths of relationships between a feature-based measurement of face-semblance and the multivariate activation pattern correlations measured by fMRI of the left and right FFAs. The feature-based face-semblance metric was computed for each image by detecting the presence of 12 contrast polarity relationships between face areas specified by the Sinha model (Sinha, 2002), e.g. forehead is brighter than left eye and summing the number of relationships (face features) fulfilled by that image (Ohayon, Freiwald, & Tsao, 2012). Partial correlation analyses reveal that, removing the influence of right FFA, left FFA activity significantly correlates with the face feature measure for both nonfaces and the whole stimulus set. By contrast, when the influence of left FFA is removed, right FFA activity does not correlate with the face feature measure. These results suggest that left FFA activation may precede and influence right FFA activation. Unlike the right FFA that appears to only deal with faces, the left FFA may process image-level semblance in nonfaces in addition to faces. Meeting abstract presented at VSS 2013
What problem does this paper attempt to address?