Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Cellular representations of natural social interactions in humans
Poster Session B, Friday, October 25, 10:00 - 11:30 am, Great Hall 3 and 4
Mohsen Jamali1, Irene Caprara1, Analise Bottinger1, Elizabeth Mayer2, Jing Cai1, Douglas Kellar1, Ziv Williams1; 1MGH, Harvard Medical School, 2Northeastern University
Social interaction is an integral part of human life, enabling us to effectively communicate and form meaningful connections with others. Understanding the fundamental cellular processes that underlie natural social communication, however, has remained a major challenge. In this study, we leveraged a unique opportunity to perform semi-chronic recordings from frontotemporal neurons (n= 251) in participants (n = 5) engaged in natural dialogue to reveal intricate cellular representations of social interactions. To quantify and track these interactions, we mapped the participants dialogue to specific social constructs using natural language processing and word embedding techniques. We then identified neurons with selective responses to these constructs and modeled their ensemble activities. By further analyzing neuronal activities across numerous interaction events, we identified neuronal populations that selectively represented specific constructs such as social agency, emotionality, clout, allure, and tone, along with their real-time dynamics. We also found that distinct classes of neurons encoded the agency of interaction, dynamically transitioning across agents. Finally, we observed that the activity patterns of these neurons reliably predicted not only the types of interactions but also the upcoming responses of the social agents they were interacting with. Together, this research begins to illuminate the basic cellular mechanisms of human social communication and provides a tentative cellular framework for understanding social behavioral and communication in humans.
Topic Areas: Meaning: Discourse and Pragmatics, Computational Approaches