Poster C63, Sunday, March 25, 1:00-3:00 pm, Exhibit Hall C
Neural representations of temporal statistics can predict subsequent reasoning
Athula Pudhiyidath1, Anna C. Schapiro2, Alison R. Preston1; 1The University of Texas at Austin, 2Harvard Medical School
Life never stops moving forward, and yet we segment events into discrete moments in time, with a beginning and an end. Here, we test how event segmentation guides the formation of neural representations that code the temporal statistics of the environment, and how these temporal statistics promote reasoning about the relationships among memory elements. We used a temporal community structure paradigm developed by Schapiro et al. (2013) to generate an iterative sequence of novel, 3D objects that participants viewed while we measured their neural responses with fMRI. Unbeknownst to the participants, the sequence was structured so that a given object belonged to one of three highly structured temporal clusters. After exposure to the temporal community structure, participants completed a series of reasoning tasks outside of the scanner, including inductive generalization. We hypothesized that participants’ knowledge of the temporal community structure would influence inductive reasoning. Behaviorally, participants were more likely to induce that members of the same temporal community shared non-temporal properties (e.g., preferred habitat) than members of different communities. Moreover, using representational similarity analysis, we found that that neural representations of temporal regularities in the hippocampus predicted subsequent reasoning behavior. Objects that were represented more similarly in hippocampus after exposure to the temporal communities were associated with higher levels of induction. Collectively, these findings add to a growing body of literature demonstrating spatial and temporal codes within the hippocampus influence decision-making beyond the domain of memory.
Topic Area: LONG-TERM MEMORY: Episodic