Poster Presentation

Poster A in Poster Session A: Tuesday, August 4, 9:30 – 11:15 am, Kimmel Center, Shorin & Rosenthal Rooms

Individual Variation in Non-Spatial Navigation is Related to Adaptive Learning and Use of Successor Representations

Alexa Booras¹, Theodoros Konstantinou Kapogianis¹, Elizabeth Chrastil¹, Aaron M Bornstein¹; ¹University of California, Irvine

Presenter: Alexa Booras

A crucial component of decision-making is the ability to learn from experience and form internal representations, or `cognitive graphs.’ Here we relate participant learning of different non-spatial associative networks to their subse- quent inferences about the latent graph. Participants first actively explored and implicitly learned stimulus-stimulus associations by viewing groups of images (nodes) rep- resenting potential paths (edges) and selecting images which triggered the presentation of the next image group. Next, participants were tested on their ability to navigate quickly through the graph. We analyzed how participant exploration related to their later navigation performance using a successor representation (SR) fit to exploration response times, capturing individual differences in plan- ning horizon as a function of graph structure. We found that people who changed their planning horizon to bet- ter align with the optimal planning horizon of a particular graph had better knowledge of that graph. Additionally, we found that individual differences in memory specificity con- strained the ability to adjust to the least-structured graph (longest planning horizon). Overall, our results suggest that humans adapt their learning and use of internal rep- resentations during associative network learning in ways that reflect their environmental regularities and memory constraints.

Topic Area: Memory, Learning & Knowledge Structures