Conference on Cognitive Computational Neuroscience

Humans use a variety of cognitive capacities and strategies to learn from others, ranging from faithfully imitating other people’s behavior to inferring the mental states that produced the behavior. Prior work has suggested people flexibly arbitrate between imitating others’ choices (Imitation; I) and inferring the value of their chosen option (Value Inference; VI). However, it remains an open question how people balance these strategies against drawing ever richer social inferences about the structure of the environment (Model-Based Inference; MBI). Using a task designed to dissociate the three strategies, we find evidence for the adaptive use of imitation, as well as preliminary evidence for MBI-level performance. Our results provide a methodological framework to understand how humans learn from others, with future work using computational modeling of choices expected to provide important insights into the arbitration of social learning mechanisms.