UC San Diego

Episodic memories unfold in space and time, typically in a sequential manner. The hippocampus, being an effective sequence generator, critically supports the acquisition and recall of episodic memories. Further, it is thought to accomplish this by implementing a “cognitive map”—and internal representation of relations among external events. However, several aspects of how the encoding, maintenance, and retrieval of sequences are mediated at the hippocampal neuronal level are not well-understood. Moreover, it is unclear how the hippocampus continues to process established cognitive maps to use them for guiding behavior at a later time. My thesis consists of two studies combining experimental and computational approaches in which I demonstrate how the hippocampus implements sequential computations to support episodic memory. To understand how the hippocampus might act as a cognitive map, first I examine how animals display stable memory-guided behaviors despite substantial reorganization of representations in the hippocampus code for space (i.e., change in the correspondence of neural activity to constant external/internal variables). In this study, I uncover several aspects of neural dynamics in terms of spatial and temporal co-firing of neural ensembles which support memory. I also investigate how subpopulations representing specific memories are reactivated after learning to support future memory retrieval. In the second study, I investigate the phenomenon of phase precession which is thought to constitute the “low-level” basis for forming and orchestrating multi-neuronal sequential activity that is repeatedly manifested in behavior. By independently controlling converging inputs (DG and MEC) onto the same neural network (hippocampal CA3), I elucidate the distinct roles of these inputs in organizing the temporal codes of CA3. I propose a simple computational model that explains the observations and makes quantitative predictions for future testing. I conclude by discussing how the phenomena discovered in the two studies are related, and propose a framework for testing my predictions in future experiments. Together, these investigations advance our understanding of hippocampal dynamics that support encoding, maintenance, and retrieval processes by dissecting multiple circuits and distinct neuronal populations within them.