Learning from other minds: An optimistic critique of reinforcement learning models of social learning

In the past decade, reinforcement learning models have been productively applied to examine neural signatures that track the value of social information over repeated observations. However, by operationalizing social information as a lean, reward-predictive cue, this literature underestimates the richness of human social learning: Humans readily go beyond action-outcome mappings and can draw flexible inferences even from a single observation. We argue that reinforcement learning models need minds, i.e, a generative model of how other agents’ unobservable mental states cause their observable actions. Recent advances in inferential social learning suggest that even young children learn from others via a generative model of other minds. Bridging these perspectives can enrich our understanding of the neural bases of distinctively human social learning.