Conference on Cognitive Computational Neuroscience

Recent literature has shown that episodic memory (EM) and working memory (WM) contribute to learning along with reinforcement learning (RL) processes. These studies typically consider one other process in addition to RL. In this study, we extended a learning task previously used to study WM contributions to RL behavior, to ask whether EM also plays a role. In a deterministic instrumental RL task, we attempt to elicit and reveal EM contributions by using trial-unique image markers and investigating whether probing previously viewed images changes performance compared to previously unseen ones. We found minimal evidence for EM in this task as participants made more accurate choices for repeated compared to new images during learning while they showed opposite results during testing. Our results show that EM’s role in WM dependent RL learning is complex, opening ways for further investigation of how multiple memory systems interact.