Learning About Others Dynamically Changes Behavior And The Brain

Ariana Familiar, University of Pennsylvania

Abstract

Humans are social beings. The ability to interact socially requires associating perceptual and social information about other people. While prior work has elucidated the cognitive and neural basis of general social knowledge, less is known about how person-specific information is learned and remembered. The goal of this dissertation was to explore how learning associations between visual and abstract information could influence conceptual representations of specific individuals. Across three studies people learned social and reward values associated with different faces. Chapter 2 examined how the learned values influenced explicit judgments by measuring behavioral face similarity spaces before and after learning. While pre-learning spaces were structured by the visual similarity of the faces, social values selectively determined the post-learning spatial organization, and generalized to expectations of behavior in a future social context. Chapter 3 investigated the neural correlates of the face-value associations. Using functional magnetic resonance imaging (fMRI), brain activity patterns were measured while participants viewed faces and performed a task unrelated to the values, once before and once after learning. A region in the left anterior temporal lobe (ATL) had activity patterns that were biased by the social values after learning, such that faces of more similar social values evoked more similar activity patterns, and the magnitude of these learning-induced changes was directly related to an individual’s learning performance as a function of value type. Additionally, activity pattern similarity in the left inferior parietal lobe (IPL) tracked the spatial organization of individual behavioral similarity spaces after learning. Chapter 4 assessed whether there were perceptual consequences of such behavioral and neural modulations and whether effects were domain-general. A categorical perception paradigm was used to test whether learned values implicitly influenced face discrimination. Preliminary evidence indicated that both social and reward values affected discrimination performance for face and flower stimuli, however the effect of social values did not persist over a long-term delay and was susceptible to task order effects. Together, this work indicates that learned associations between visual and social attributes of other people can warp behavioral and neural representations, and such changes have downstream consequences on face perception and social preferences.