Schedule of Events | Symposia

Attention compensates for reduced reinforcement learning in older adults

Poster Session A - Saturday, March 29, 2025, 3:00 – 5:00 pm EDT, Back Bay Ballroom/Republic Ballroom

Daniel Oppermann Peixoto¹ (dpeix001@ucr.edu), Ian C Ballard¹; ¹University of California, Riverside

Adaptive learning requires tracking the value of conjunctions of features in the environment, as when one learns that cinnamon and oatmeal can be combined into a delicious dish, even though cinnamon is unpleasant on its own. Previous work has shown that conjunctive representations in the hippocampus support such conjunctive reinforcement learning. Given age-dependent declines in hippocampal function, we hypothesized that older adults would exhibit reduced conjunctive reward learning. We tested younger (n=194) and older (n=196) adults on a speeded target detection task in which four stimulus configurations differentially predicted the appearance of a target (70% for target-positive stimuli AB+ and B+ and 30% for target-negative stimuli AC- and B-). Because each feature alone (A, B, C) was not target-predictive, forming conjunctions (e.g., “AB”) was necessary to maximize task performance. We also included a recognition memory task to assess whether episodic memory mediated age-related differences in conjunctive learning performance. As predicted, older adults exhibited reduced conjunction-based learning. Surprisingly, older adults also exhibited reduced feature-based learning, indicating an overall reduction in reinforcement learning. The degree of conjunction and feature learning correlated with recognition memory performance, consistent with a model in which hippocampal representations of task features support reinforcement learning. Strikingly, older adults earned as many points in the task as younger adults. Indices of sustained attention were correlated with the number of points earned, and this effect was more pronounced in older adults, suggesting that attention compensates for age-related differences in reinforcement learning.

Topic Area: EXECUTIVE PROCESSES: Other