The (In)Flexible Social Brain: How Learning Dynamics Unfold in Our Uncertain Social World

Lamba, Amrita

Full Metadata

Overview

Year:: 2023
Contributor:: Lamba, Amrita (creator); FeldmanHall, Oriel (Advisor); Frank, Michael (Reader); Badre, David (Reader); Nassar, Matthew (Reader); Brown University. Department of Cognitive, Linguistic, and Psychological Sciences (sponsor)
Genre:: theses
Subject:: Reinforcement learning; Social learning; Computational neuroscience; Psychiatry
Extent:: 12, 196 p.

Files

Description

Abstract:: How do humans learn to adaptively function in a fundamentally uncertain social world? Since we cannot directly observe the contents of other people’s minds, the key challenge of social learning is to find latent features that enable us to tailor our behavior to specific individuals despite uncertainty. To date, very little is known about the algorithmic or neural mechanisms that support social learning. In this thesis, I use a neurocomputational approach to examine how humans learn when the causal structure of the world is unknown, and how humans adjust their behavior when the world suddenly changes. Using a Reinforcement Learning (RL) model that evaluates how people assign credit (i.e., attributing outcomes to the correct cause) combined with multivariate neuroimaging, I show that credit assignment substantially differs across individuals. Furthermore, people are more precise in crediting social agents compared to nonsocial objects, a process that is mediated by high-fidelity (i.e., distinct and consistent) neural representations in the prefrontal cortex. Next, I investigate how people learn in noisy and nonstationary environments. Prior work shows that uncertainty speeds learning by prioritizing the impact of recent outcomes on one’s beliefs. However, previous work typically treats uncertainty as a homogenous construct without distinguishing between the influence of different sources of uncertainty. Using a dynamic trust task and a novel eye-tracking method, I deconstruct uncertainty into two distinct components—policy and epistemic uncertainty—to show that humans adaptively switch between optimizing outcomes (reducing policy uncertainty) and gathering general world knowledge (minimizing epistemic uncertainty) to learn about others. I then investigate how individuals with increased generalized anxiety, a disorder characterized by increased uncertainty sensitivity, learn in changing environments. Anxious individuals exhibit slowed learning, particularly in the social domain and when engaging with untrustworthy partners. A novel Bayesian-RL model attributes maladaptive learning to increased difficulty resolving policy uncertainty and a bias towards epistemic uncertainty. Across this thesis, I demonstrate that several critical factors, namely how we implement credit assignment and how we use uncertainty to scaffold learning, impact how we build and exploit our mental models of others.
Notes:: Thesis (Ph. D.)--Brown University, 2023

Content

Citation

Lamba, Amrita, "The (In)Flexible Social Brain: How Learning Dynamics Unfold in Our Uncertain Social World" (2023). Cognitive, Linguistic, and Psychological Sciences Theses and Dissertations. Brown Digital Repository. Brown University Library. https://repository.library.brown.edu/studio/item/bdr:n762nxcq/

Relations

Collection:

Cognitive, Linguistic, and Psychological Sciences Theses and Dissertations

Theses and Dissertations for the Cognitive, Linguistic, and Psychological Sciences department.
...