As we observe and listen to other people, we interpret their actions in terms of a rich causal structure, driven by underlying mental states and dispositional traits. The ability to rapidly extract abstract social information from perceptual input, termed social perception, is critical to human social behavior. This thesis investigates the cognitive architecture of social perception by studying the functional organization and development of brain regions implicated in this process. In the first Chapters 2-4, I focus on a region that has been strongly implicated in social perception: the superior temporal sulcus (STS). In Chapter 2, I assess the overall functional organization of STS responses to different types of social stimuli, using fMRI. I find that the STS comprises a number of functionally specific subregions that process certain types of social information, such as body movement, vocal sounds, linguistic input, and abstract mental states, suggesting a functional division of labor in social perception. I also identify a multimodal region that responds both to face movement and to vocal sounds. Chapters 3-4 further explore the properties of this region, termed the fSTS. Analyzing spatial patterns of response in this region to different types of face movement, I find evidence that fSTS contains a parts-based representation of perceived face movement type, suggesting a representation more tied to face movement kinematics than implied mental state, but which generalizes across low-level visual properties (Chapter 3). Next, assessing responses to a range of naturalistic face movements and vocal sounds, I find that the fSTS responds strongly to virtually any face movement or vocal sound, irrespective of social relevance or speech content (Chapter 4). However, patterns of response in this region distinguish more and less socially relevant inputs, in a manner that generalizes across facial and vocal stimuli. Taken together, these results point to the fSTS as a mid-level region in social perceptual inference, with representations that are still tied to perceptual features, but begin to integrate visual and auditory inputs and make explicit high-level social distinctions. Lastly, in Chapter 5 I address a broader, related question: how do functionally specific brain regions such as the fSTS develop? To address this question, I develop methods for fMRI in awake infants viewing visual stimuli. I find that regions preferring specific visual categories (faces and scenes) exist by 4-6 months of age, but that responses in these regions are less selective than in adults. This suggests that functionally specific brain regions exist in some form at an early age, but that they become increasingly specialized throughout development.

To read this thesis https://stellar.mit.edu/S/project/graduatethesis/courseMaterial/topics/…