Anna Bognar^1,2 (), Ghazaleh Ghamkhari Nejad^1.2, Michael Stettler³, Albert Mukovskiy³, Martin Giese³, Rufin Vogels^1,2; ¹Department of Neurosciences, KU Leuven, Leuven, Belgium, ²Leuven Brain Institute, KU Leuven, Leuven, Belgium, ³Section Computational Sensomotorics, Department N3, Hertie Institute for Clinical Brain Research & Centre for Integrative Neurocience, University Clinic Tübingen, Tübingen, Germany

Primates rely on observing social interactions to infer intentions and hierarchical relationships. Although a macaque fMRI study has implicated the superior temporal sulcus (STS) in interaction processing (Sliwa & Freiwald, 2017), the neural mechanisms underlying the processing of social interactions remain poorly understood. To examine the encoding of natural social interactions versus altered interactions and facing versus non-facing scenarios (Papeo, 2000), we recorded units (single and multi-units) targeting the ventral mid-STS and anterior-STS body patch. Animals viewed 126 2.5 s videos depicting two monkey avatars performing fighting, following, or meeting actions either in their natural form or in versions where flipping one of the two avatars disrupted their interaction. Isolated one-body videos were included to dissociate social interaction-specific signals from responses to single body components. Videos were rendered from seven different views. Using responsive and reliable units (mid-STS: 402 & 237; anterior-STS: 377 & 241 in Monkey1 and Monkey2, respectively), we found in both monkeys and regions that average two-body responses were less than the corresponding summed single body responses, in line with studies on shape interactions showing divisive normalization. Multiple-regression analyses, with responses to the single bodies as continuous predictors, facing versus non-facing as categorical predictor and their interaction terms, showed that two-body responses were better predicted by the contralateral than ipsilateral avatar with this hemifield-specific contribution being stronger in mid-STS. In anterior-STS, facing fighting avatars produced significantly stronger averaged responses than non-facing ones, and 13–17% of units showed significant interaction terms for fighting. The stronger responses to facing versus non-facing fighting videos were also supported by other analyses that compared directly facing versus non-facing interactions. Less consistent social interaction effects were observed for the meeting and following actions and in mid-STS. These results suggest encoding of social content in a minority of ventral anterior-STS units.

Acknowledgements: This research was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant Agreement Number 856495).