Learning about others and learning from others: Bayesian probabilistic models of intuitive psychology and social learning.

Abstract

How do infants and young children reason about other people? What inferences do they make when they learn from teachers and whom do they choose to learn from? Past research in developmental psychology has demonstrated infants' and young children's competence in making these inferences. However, the mechanisms underlying these inferences and how these mechanisms change across development are less clear. In this chapter, we review a growing body of Bayesian probabilistic models on intuitive psychology and social learning. We integrate these models with past and new empirical studies within the framework of rational constructivism. These models showed that infants and children have intuitive theories about others (agents, teachers, and informants). When given new evidence, they rationally update their beliefs about others and their beliefs about the world based on these intuitive theories. Developmental changes can be explained by advances in children's intuitive theories. Finally, we propose future directions for both empirical and modeling work in these domains.