Reinforcement Learning in Individuals at Risk for Alzheimer's Disease

Christina Marie Figueroa, Marquette University


Explicit memory is the hallmark of impairment in Alzheimer’s disease (AD) while implicit memory has mixed task-dependent results. Models of memory processes have posited that hippocampal function is sensitive to reinforcement learning (RL), which involves both explicit and implicit memory. The hippocampus is also vital for the transfer of learned associations to novel situations. Nevertheless, RL paradigms have been underutilized in assessing memory processes in individuals at risk for AD, which may aid in early identification of cognitive decline. Thirty-six apolipoprotein-E (APOE) genotyped older adults (Male n=8; Mage=80; Meducation=15 years) performed word stem completion, word recognition, and RL tasks. The RL task was comprised of an RL phase, an implicit testing phase, and explicit recognition component. Group comparisons were made based on low risk (APOE ε4-; n=16) vs. high risk (APOE ε4+; n=20) for AD. A series of mixed ANOVAs based on task performance indicated that risk groups did not differ on EM measures (RL, word recognition, and RL recognition). However, high risk participants exhibited significantly poorer IM performance (RL testing and word stem) than the low risk group, p = .03. The pattern of results in the present study was counter to prediction in that risk groups did not differ on explicit memory measures, which was strongly supported by existing literature. However, the exhibited performance of poorer implicit memory in the high risk group is consistent with results implicating the hippocampus in the application learned associations to novel environments. RL paradigms may offer high sensitivity for assessing preclinical decline.