Computational Models of Encoding and Replay of Sequences in the Hippocampus
No Thumbnail Available
Place cells in the rodent hippocampus display various electrophysiological phenomena such as non-local replay sequences and phase precession that are widely studied for their role in navigation and spatial memory consolidation. Distinct patterns of activation are observed during locomotion and during resting periods that give rise to phase precession and replays respectively. To investigate the mechanisms through which these two different phenomena could arise within the same network, Romani and Tsodyks developed a computational firing rate model of place cells in the rodent CA3 hippocampal subfield in their 2015 study. In this model, simple internal dynamics combined with a pre-defined and fixed connectivity structure resulted in both non-local sequences and phase precession. For this thesis I first replicated and then elaborated on some of these findings by creating a flexible toolbox, where a synaptic learning rule is implemented to obtain the connectivity structure through an exploration trial instead. I found that during resting periods, the network exhibited very structured and consistent non-local sequences throughout the spontaneous activity, showing that replay-like events can indeed arise from a simple dynamical model featuring a learned connectivity matrix.
Faculteit der Sociale Wetenschappen