Abstract
Visual cues to gender are abundant in the physical properties of faces and bodies. We used fMRI and multivariate pattern analysis (MVPA), combined with a rapid-event stimulus procedure, to explore the neural basis of perceived gender. Unlike previous MVPA studies of face gender perception, we included bodies and common names in our study. We predicted that MVPA could classify the gender of faces and bodies, both in early and category-selective visual cortex. In contrast, we predicted that common gender-specific names (e.g. Ann and Mary versus Bob and Mike) would not be successfully classified using MVPA in early visual cortex. We predicted this because then gender of names is not directly indicated by the physical properties of name stimuli per se. We did, however, predict that name gender could be accurately classified in more anterior brain areas. Additionally, we expected faces and bodies to be better classified in the right hemisphere than in the left, with the opposite expectation for common names. Contrary to our expected results, we found little evidence of accurate MVPA gender classification in early visual cortex for any of our stimulus categories. Instead, successful MVPA gender classification was limited to occipitotemporal cortex, primarily in the left hemisphere, in addition to more anterior brain regions. Maximal MVPA gender classification accuracy was observed in left occipitotemporal cortex, for all three categories of stimuli. Our results are consistent with the role of left occipitotemporal cortex as a visuo-semantic hub involved in gender classification, and perhaps in person perception more generally.
Meeting abstract presented at VSS 2018