Sajjad Torabian¹ (), John A. Pyles², Yujia Peng³, Hongjing Lu⁴, Emily D. Grossman¹; ¹University of California, Irvine, ²University of Washington, ³Peking University, ⁴University of California, Los Angeles

Introduction. Humans share strong intuitions about physical events and also about the behavior of social beings. These two capacities have often been treated as distinct. For example, evidence has linked intuitive physics and reasoning about social behavior to separate neural systems that include the intraparietal sulcus (IPS) and the superior temporal sulcus (STS), respectively. Recent computational work (Shu et al. 2021), however, has modeled the perception of violations of physical laws and impressions of goal-directed movement within a unified psychological space, namely the physical social forces model (PSF). Here, we test whether this computational model can explain neural responses to physical and social events. Methods. In a rapid event-related fMRI design, participants viewed short animations (3s each) of two moving shapes within a square box. The movement dynamics were systematically manipulated such that they elicited a wide range of violations of physics and degrees of intentionality. Results. A searchlight representational similarity analysis revealed cortical maps that were best explained by the PSF model, a categorical model of physical vs social events, and models based on low-level visual cues. The categorical model appeared as a strong predictor of the neural similarity around the pSTS and TPJ. The fine-grained estimates of violation of physics and intentionality of the PSF model also robustly predicted neural similarity around the IPS and early visual cortex, which variance partitioning attributes to perceptual features of average speed and average distance between the shapes. Importantly, after removing animations with strong physical interactions, the intentionality estimates also predicted responses around the pSTS, with variance jointly explained with a model based on speed variability in the animations. Conclusion. These results reveal brain activities that not only encode a categorical representation of physical and social events, but also capture a gradient in a unified psychological space.

Acknowledgements: Funding: NSF BCS-1658560 to EG and BCS-1658078 to JP and NSF BCS-2142269 to HL