Abstract
Like face perception, body recognition is a fundamental social skill for many animals including primates. While several body-selective regions have already been identified in both humans and nonhuman primates, there is no comprehensive picture of which regions are specifically involved in dynamic body perception and whether they are species-specific. The current 7T fMRI study investigated the neural correlates of dynamic human and monkey bodies and faces in human participants. METHODS: Stimuli consisted of videos of dynamic faces, bodies and objects from both monkey and human recording. We used a block design and a fixation-change-based oddball task. We first conducted a univariate analysis to find the voxel-wise selectivity for dynamic bodies, then conducted ICA-based network analysis to find network-level modulations. Finally, we applied MVPA on ROIs to find pattern specificity of cross-species information. RESULTS: The univariate analysis found widely distributed cortical body patches including the EBA, FBA, STS, MFG, etc., and subcortical areas including pulvinar and amygdala. Further, two networks were found body selective: LOC network, including LOC, fusiform, SPL and pulvinar; and right STS network, including right STS, IPL, SFG, MFG, IFG, insula, and amygdala. The right STS network showed higher species selectivity for bodies than for faces. Similar response patterns for human and monkey bodies were found around LOC and fusiform but not in STS or frontal clusters. However, in STS and MFG, similar patterns between bodies and faces were found only for human videos but not for monkey videos. CONCLUSION: Our study extended the current findings on body perception by using dynamic stimuli and cross-species conditions. Widespread body modulations were found at both voxel and network levels. A more separated hierarchy was found for bodies than for faces, with the general categories represented by occipital areas and the semantic/social information represented by STS and frontal areas.