Exploring the structure, dynamics, and developmental trajectory of person models

Abstract

Effective social interaction requires reasoning about people as generative models. In our day-to-day experience, we come across a remarkable amount of social information, often in the form of other people’s behaviors. Observed behaviors are used to infer agents’ unobservable mental states and traits – the latent causes that drive their behavior. These inferences are stored in person models, which allow us to interpret patterns of observed behaviors across multiple instances and contexts by attributing a common cause to those behaviors, and also allow us to predict people’s future actions, so that we may navigate interactions smoothly and choose our social partners wisely. This dissertation pursued several open questions on flexible trait reasoning. In Paper 1, we found that the relative contributions of different traits to overall impressions may vary depending on what we know about a person. In Paper 2, we found increased neural activity in Theory of Mind regions following the violation of strong and positive prior impressions. In Paper 3, we found that 6-9-year-olds exhibit a negativity bias in impression updating, and older children are sensitive to the strength of behavioral evidence. Overall, we found evidence for flexible trait reasoning – both children and adults were sensitive to the strength and valence of available behavioral evidence, and to the overall inference context. These studies help shed light on how children and adults reason about person models and respond to new social information, and we suggest multiple avenues for further research in this arena.