Neuronal Oscillations: In Hippocampal Functions and in Simulations

Abstract

The overall theme of this dissertation is how the brain uses rhythms in spike timing to carry out computation, and secondarily about developing tools for studying these rhythms. The bulk of the dissertation concerns the different roles played by hippocampal cells in coding spatial-reward information and in the hippocampal replay. Previous experiments have provided strong evidence that, accompanied by theta and gamma oscillations (4–10 Hz and 25-140 Hz, respectively) as partial network synchrony, hippocampal cells form a spatial representation of animal location in the environment. During fast oscillations called sharp-wave-ripple (SWR, 140-220 Hz), hippocampal cells exhibit synchronized-firing bursts at a population level, replaying previous firing patterns related to memory. This phenomenon, termed “hippocampal replay,” has been regarded as the prime mechanism for system-wide memory consolidation during sleep and planning for future movement during the awake state. While attractive, these theories may be oversimplifying. Many studies have revealed the conjunctive-coding nature of many hippocampal cells as a possible strategy to indicate the integration of different types of information into spatial navigation. So far, how cells with different coding properties emerge from (and contribute to) the hippocampal oscillations at the network level, including theta, gamma, and SWR oscillations, remain unclear. As a preliminary attempt to address this question, in this dissertation, I present our contribution in three parts by collocating three manuscripts. We first characterize hippocampal neurons in a set of complex goal-directed spatial navigation experiments with their coding properties and report a population of dorsal CA1 cells ranging continuously from “place cells” to “reward cells.” With a novel hierarchy of quantitative measures, I then proceed to investigate how they contribute to SWR-mediated hippocampal firing patterns. Furthermore, to assist the design of large-scale network models concerning the emergence of coding properties from neural oscillations, I develop a Multi-Level Monte Carlo method (MLMC) to speed up time-consuming simulations. Remarkably, I find that the applicability of MLMC is highly related to the network synchrony. Overall, this dissertation provides a novel view of reward-place conjunctive representation in the hippocampus and how it is coupled with hippocampal replay during the high-frequency neuronal oscillations and synchronized firing patterns. The synchrony-dependent applicability of MLMC method, on the other hand, suggests a unique status of synchronized dynamics during simulations of hippocampal network models, which should be handled more cautiously.