Date of Award
Doctor of Philosophy (PhD)
Nielson, Kristy A.
Reinforcement learning (RL) has been widely used as a model of animal and human learning and decision-making. The neural networks underlying RL involve many of the same structures primarily affected by Alzheimer’s disease (AD) such as the hippocampus. Yet, RL and non-invasive evaluation of its neural underpinnings have been underutilized as a framework for understanding disease pathology and its pre-clinical states. This study aimed to provide a novel approach for assessing subtle changes in asymptomatic apolipoprotein-E (APOE) carriers and non-carriers. Electroencephalography was collected from forty APOE genotyped older adults (Male n = 11; Mage = 79.30; Meducation = 14.88 years) during an RL task comprised of distinct phases (RL, implicit). Neural components associated with the error detection system involved in RL, the response error-related negativity (ERN) and the feedback error-related negativity (FRN), were examined for individuals at low (APOE ε4-; n=20) and high risk (APOE ε4+; n=20). RL task performance did not differ between risk groups. However, the high-risk group consistently elicited greater peak amplitudes than the low-risk group. The pattern of results indicated that the high-risk group deviated from typical RL processes such that peak amplitudes did not differ between early and late learning. Additionally, despite intact learning, latent hippocampal atrophy is believed to have disrupted the transfer and use of learned information to novel situations thus altering the hippocampal-frontostriatal circuit responsible for adaptive behavior and the corresponding neural signal. The results indicate that disease related changes can be identified prior to clinical diagnosis or functional decline using RL and a non-invasive assessment of neural function, which may prove to inform clinical conceptualization, assessment, and treatment.
Cognitive Psychology Commons, Genetics Commons, Geriatrics Commons