Flexible hippocampal representation of abstract boundaries supports memory-guided choice

Abstract

AbstractCognitive maps in the hippocampus encode the relative locations of spatial cues in an environment and dynamically adapt their representation when boundaries geometrically change. In parallel, hippocampal cognitive maps can represent abstract knowledge, yet it’s unclear whether the hippocampus is sensitive to geometric changes to the borders, extreme coordinates, of abstract knowledge spaces. Here, we use a memory-guided decision making task to test whether the human hippocampus and medial prefrontal cortex(mPFC) flexibly learn abstract boundary representations in distinct two-dimensional(2D) knowledge spaces. Despite being unnecessary to accurately make decisions, participants conserve a 2D map-like representation of abstract boundaries after the task, where the precision of their representation relates to prior choice accuracy. Finding that the hippocampus and mPFC represent the Euclidean distance of a decision cue to the most proximal boundary during decision making, we then test whether there are brain regions sensitive to boundary-defined contextual changes in abstract spaces. We observe flexible hippocampal representation of abstract boundaries, where the fidelity of this representation relates to task performance. Taken together, our results highlight the importance of hippocampal boundary representations in facilitating flexible knowledge retrieval in dynamically changing abstract contexts.