Abstract
Reinforcement learning deficits have long been associated with schizophrenia. However, tasks traditionally used to assess these deficits often rely on multiple processing streams leaving the etiology of these task deficits unclear. In the current study, we borrowed a recent framework from computational neuroscience, which separates reinforcement-learning into two distinct systems, model-based and model-free. Under this framework, the model-free system learns about the value of actions in the immediate context, while the model-based system learns about the value of actions in both immediate and subsequent states that may be encountered as a result of their actions. Using a decision task that has been previously validated to assess relative reliance on each system we showed that individuals with schizophrenia demonstrated decreased model-based but intact model-free learning estimates. Furthermore, parameter estimates of model-based behavior correlated positively with IQ, suggesting that model-based deficits in schizophrenia may relate to reduced intellectual function. These findings specify reinforcement-learning deficits in schizophrenia by showing both intact and disturbed components. Such findings and computational frameworks provide meaningful insights as researchers continue to characterize decision-making circuitry in schizophrenia as a means to discover new pathways for interventions.
Committee Chair
Deanna Barch
Committee Members
Thomas Oltmanns
Degree
Master of Arts (AM/MA)
Author's Department
Psychology
Document Type
Thesis
Date of Award
Summer 8-15-2015
Language
English (en)
DOI
https://doi.org/10.7936/K73T9F8T
Recommended Citation
Culbreth, Adam, "Breaking Apart the Reinforcement Learning Deficit in Schizophrenia" (2015). Arts & Sciences Theses and Dissertations. 589.
The definitive version is available at https://doi.org/10.7936/K73T9F8T
Comments
Permanent URL: https://doi.org/10.7936/K73T9F8T