Social learning, how we acquire knowledge from, for, and about others, emerges from multiple interacting dimensions: whom we interact with, how we draw inferences, and how we make decisions. This talk outlines a series of studies connecting these dimensions across cognitive, neural, and computational levels. I will present evidence that humans do not simply copy others but integrate actions and outcomes through structured and value-based inference. Neuroimaging findings show that social prediction and vicarious value rely on both domain-general learning mechanisms and socially specific neural computations. I will also discuss how social learning may go awry in individuals with mental health risks. Finally, I will present recent work on how these principles can inform machine intelligence by tuning large language models with a “social chain of thought”, which may offer promising steps toward more human-like social inference in AI. Together, these insights map key dimensions of social learning across mind, brain, and machine.
Dr. Lei Zhang is an Associate Professor at the Centre for Human Brain Health, School of Psychology, University of Birmingham. He directs the Adaptive Learning Psychology & Neuroscience (ALPN) Lab, where he addresses the fundamental question of the “adaptive brain” by studying the cognitive, computational, and neurobiological basis of social learning and decision-making in healthy individuals and in psychiatric disorders. Dr. Zhang received multiple international recognitions, including the Credibility Prizes from the British Neuroscience Association and the Rising Star Award from the Association for Psychological Science. He is an Editorial Board Member for Communications Biology and PLOS Mental Health, and has served as a reviewing expert for multiple journals and international research funding agencies (e.g., ERC, NIH, UKRI).
Host: Prof Jingyi WANG, Assistant Professor, Division of Social Science, HKUST