Date of Award
Doctor of Philosophy
Fred Helmstetter, Rodney Swain, Caleb Kemere
Bayesian decoding, Hidden Markov model, Hippocampus, Neuronal population code, Spatial memory
The hippocampus is essential for learning and memory. Neurons in the rat hippocampus selectively fire when the animal is at specific locations - place fields - within an environment. Place fields corresponding to such place cells tile the entire environment, forming a stable spatial map supporting navigation and planning. Remarkably, the same place cells reactivate together outside of their place fields and in coincidence with sharp-wave ripples (SWRs) - dominant electrical field oscillations (150-250 Hz) in the hippocampus. These offline SWR events frequently occur during quiet wake periods in the middle of exploration and the follow-up slow-wave sleep and are associated with spatial memory performance and stabilization of spatial maps. Therefore, deciphering the firing patterns during these events is essential to understanding offline memory processing.I provide two novel methods to analyze the SWRs firing patterns in this dissertation project. The first method uses hidden Markov models (HMM), in which I model the dynamics of neural activity during SWRs in terms of transitions between distinct states of neuronal ensemble activity. This method detects consistent temporal structures over many instances of SWRs and, in contrast to standard approaches, relaxes the dependence on positional data during the behavior to interpret temporal patterns during SWRs. To validate this method, I applied the method to quiet wake SWRs. In a simple spatial memory task in which the animal ran on a linear track or in an open arena, the individual states corresponded to the activation of distinct group of neurons with inter-state transitions that resembled the animal’s trajectories during the exploration. In other words, this method enabled us to identify the topology and spatial map of the explored environment by dissecting the firings occurring during the quiescence periods’ SWRs. This result indicated that downstream brain regions may rely only on SWRs to uncover hippocampal code as a substrate for memory processing. I developed a second analysis method based on the principles of Bayesian learning. This method enabled us to track the spatial tunings over the sleep following exploration of an environment by taking neurons’ place fields in the environment as the prior belief and updating it using dynamic ensemble firing patterns unfolding over time. This method introduces a neuronal-ensemble-based approach that calculates tunings to the position encoded by ensemble firings during sleep rather than the animal’s actual position during exploration. When I applied this method to several datasets, I found that during the early slow-wave sleep after an experience, but not during late hours of sleep or sleep before the exploration, the spatial tunings highly resembled the place fields on the track. Furthermore, the fidelity of the spatial tunings to the place fields predicted the place fields’ stability when the animal was re-exposed to the same environment after ~ 9h. Moreover, even for neurons with shifted place fields during re-exposure, the spatial tunings during early sleep were predictive of the place fields during the re-exposure. These results indicated that early sleep actively maintains or retunes the place fields of neurons, explaining the representational drift of place fields across multiple exposures.
Maboudi Ashmankamachali, Kourosh, "Deciphering the Firing Patterns of Hippocampal Neurons During Sharp-Wave Ripples" (2022). Theses and Dissertations. 3039.
Biomedical Engineering and Bioengineering Commons, Neuroscience and Neurobiology Commons