Pattern Separation and Generalization in the Hippocampus

Stanford University, Stanford CA

Investigators

Linked publications & trials

Abstract

Although an impressive amount of evidence has emerged supporting a fundamental role for the hippocampus in declarative memory, the manner in which individual structures within the hippocampus contribute to memory remains a current topic of debate. The central objective of this proposal is to evaluate how differences in circuitry among hippocampal subfields translate to functional differences in support of episodic memory formation. In forming new memories, the hippocampus is thought to separate related input patterns into distinct representations, thereby minimizing interference between similar memory traces. However, findings from neural-network models and animal studies suggest that tension exists within the hippocampus such that some subfields readily contribute to pattern separation, whereas others tend to generalize across similar input patterns. Using high-resolution functional magnetic resonance imaging (fMRI), the proposed research aims to evaluate the degree to which human hippocampal subfields support the formation of non-overlapping vs. shared memory representations in response to similarity manipulations at both the item and inter-item level. Furthermore, to complement this high spatial resolution fMRI approach, local field potentials will be directly recorded from the human hippocampus to better characterize the temporal profile of pattern separation and generalization. Delineating the manner in which hippocampal subfields mediate memory will advance mechanistic accounts of hippocampal function, as well as provide insights into brain regions affected in populations exhibiting pattern separation deficits, such as older adults and individuals with schizophrenia.