Angela Langdon, Ph.D.

Research Topics

Learning to predict outcomes and act accordingly is the cornerstone of adaptive behavior. The aim of the Unit on the Neural Computations in Learning is to understand the computational processes involved in reward prediction and learning in the brain and in behavior and how these processes are disrupted in a range of psychiatric disorders. Research in the lab is focused on how neural activity in midbrain, striatal and cortical circuits support trial-and-error learning in animals and humans, and how timing processes dynamically shape expectations during reward-guided behaviors. We are particularly interested in how ‘prediction errors’—the mismatch between predicted and actual outcomes—are computed and signaled in the brain, how these error signals are used to update reward predictions during experience, and how these reward predictions are organized into ‘task representations’ that guide goal-directed behaviors.

Research in the lab exploits various theoretical tools from reinforcement learning, Bayesian inference, dynamical systems theory and machine learning to build models of learning at both the neural and behavioral levels. We then test our models against neural recordings from animals collected by our collaborators and human behavioral data collected during novel reward-guided tasks designed within the lab to parallel the animal paradigms. We collaborate widely, with systems, behavioral, cognitive and clinical neuroscientists, in order to extend and test our theories to understand maladaptive learning and achieve translational insight into altered reward-guided behaviors in states of stress, compulsive behaviors and disorders of mental health.