Modeling Other Minds: A Computational Account of Social Cognition and Its Development

This article reviews how humans come to understand other minds from a computational perspective. We propose that social development is structured around three abilities: ( a ) building representations of agents and minds from a small set of abstract primitives, ( b ) embedding these representations into a probabilistic causal model of rational action, and ( c ) using this model to interpret everyday behavior. For this third ability, we argue that using a full model of other minds is too computationally demanding. To manage this, people learn how to build simplified context-specific models that balance computational efficiency with explanatory power. Learning how to build these restricted scope models may be a central but understudied aspect of development, shaped in part through everyday conversation. All together, our framework offers a formal account of social development and highlights open questions about how this capacity emerges and develops.