Proceedings of the Annual Meeting of the Cognitive Science Society

We don’t learn about each person we meet from scratch: Ourknowledge of social groups (e.g., cognitive scientists) shapesour expectations about new individuals (e.g., the reader). Herewe explore how 4- and 5-year-old children and adults use min-imal statistical evidence about social groups to support induc-tive inferences about individuals. Overall, we find that bothchildren and adults readily infer the preferences and groupmembership of new individuals when they have appropriateevidence to support these inferences. However, our resultsalso suggest that children and adults interpret this informa-tion in different ways. Adults’ responses align closely witha Bayesian model that assumes that each group’s preferencesare independent of one another. By contrast, we find prelimi-nary evidence that children’s inferences about the preferencesof new group members are sensitive to the composition (Exper-iment 1) and size (Experiment 2) of the opposing group. Ourwork provides insights into how people form structured, gen-eralizable representations of social groups from sparse data.