Disentangling the representations of object shape and object category in the brain: the animate-inanimate distinction

Abstract

Objects belonging to different categories elicit reliably different response patterns in human ventral temporal cortex, with the most prominent distinction being that between animate and inanimate objects. However, it remains unclear whether these categorical distinctions genuinely reflect object category or, alternatively, category-associated visual properties. To address this question, we designed a stimulus set in which animate and inanimate stimuli were carefully matched for shape. Different aspects of visual similarity of the stimuli were measured in three behavioral visual search tasks. We then conducted an fMRI study to test the hypothesis that animate-inanimate organization in VTC response patterns can be explained by visual properties of these categories. We found that it was not the case: although visual dissimilarity predicted neural dissimilarity across the visual cortex, we also found regions in which category information was present even after regressing out visual dissimilarity. We then conducted an MEG study to examine the time course of shape- and category-related information in the brain. Following the analysis approach used in the fMRI study, neural dissimilarity of MEG sensor patterns was modeled using regression analysis, where visual dissimilarity and categorical dissimilarity served as predictors of neural dissimilarity. The results show that visual object properties were strongly contributing to MEG sensor patterns. Surprisingly, when regressing out the contribution of visual properties, no residual category information was present in MEG response patterns. These results suggest that MEG sensor patterns evoked by visually presented objects predominantly reflect visual object properties. Taken together, these findings suggest that MEG is less sensitive to object category information that is independent of shape information.