The position of visual word forms in the anatomical and representational space of visual categories in occipitotemporal cortex

Abstract

Abstract Recent reviews emphasized the need for investigating the complexity of multiple subareas of word selectivity and how this relates to selectivity for other visual categories, at the individual level at a high spatial resolution (without normalization or smoothing). To investigate this, both on the brain surface and in the representational space of the occipitotemporal cortex, we presented 19 participants with images of 20 different categories during 7T fMRI. These categories included several word-like conditions, and in addition cover many of the dimensions that have been suggested to define object space, such as animacy and real-world size. In the left hemisphere, we found three subareas of the visual word form area (VWFA) and one extra subarea around the pFus face-selective area. We also observed several areas of selectivity to hands that could consistently guide the localization of word and face areas. No clear predictive anatomical landmarks were found. Results of the right hemisphere were less clear, in part due to weaker word selectivity. In the representational space, word selectivity stood out from other categories. It had multiple neighboring categories at a similar distance (e.g., faces, bodies, hands, cars), so no special relationship was found with, for example, faces. These results enable a consistent and reliable way to locate subareas of word selectivity and may inspire future research into words in the representational space of the occipitotemporal cortex.