Estudos eletrofisiológicos no hipocampo - desenvolvendo novas técnicas de análise e investigando a codificação neural

Abstract

Extracellular electrophysiology is among the main tools used for studying neural activity. In addition to having a high temporal resolution, this technique is stable enough to allow for recording freely-moving animals. In the last decades, the electrophysiology of the hippocampal formation has received particular attention due to the discovery of its relationship with memory and spatial navigation. Specifically, many spatial correlates have been investigated to understand the basis of neural encoding, such as place cells, which increase their firing rate when the animal is at particular locations of the environment. In this thesis, we present three studies exploring hippocampal electrophysiology through different levels of analysis, which aim at better understanding the different coding mechanisms used by the brain. In the first work, we focus on the problem of spike sorting, which consists in the classification of the waveforms detected by the extracellular recording. We investigated the use of Gaussian mixture models to perform feature extraction and clusterization of the waveforms. After using real and simulated data to evaluate the best strategies of our algorithm, we compared our spike sorting to two other known methods. Our algorithm, which combines two main feature extraction techniques with a new method of dimensionality reduction, showed better, or similar, results compared to the two other spike sorters. In the second work, we analyzed the current metrics used to find neurons presenting some spatial correlate, or information, such as place cells, headdirection cells, and speed cells. Using real and simulated data, we evaluated how the spatial information estimated by each metric correlated to the empirical information obtained using a NaiveBayes classifier. Our results showed that the two main current metrics fail to detect the information content of neurons with high basal firing rate, and thus bias the universe of spatial correlates to lowfiring cells. Finally, in the third work, we explored the differences in the proposed mechanisms underlying place cell firing: rate coding and temporal coding. We analyzed the dynamics of coupling between place cells and theta oscillations, comparing it to the firing rate dynamics. Our results revealed an asymmetry between the two mechanisms, with the theta coupling place cells preceding its major changes in firing rate, which corroborate the hypothesis that temporal and rate coding are independent mechanisms. Beyond that, we found that temporal coding represents more extensively the future positions of the animal than the past ones, suggesting its involvement in trajectory planning. In summary, the results of this thesis contribute to both the understanding of neuronal encoding and the development of new techniques in the field of hippocampal electrophysiology.