Szabolcs Káli
Tue 08 May 2018, 11:30 - 12:30
IF 4.31/4.33

If you have a question about this talk, please contact: Gareth Beedham (gbeedham)

Hippocampal sharp wave-ripples and the associated replay of neural activity sequences emerge from structured synaptic interactions in a network model of area CA3


Population activity patterns recorded in the hippocampus in vivo include theta-modulated gamma oscillations and sharp wave-ripple (SWR) events. During SWRs, neuronal populations in the hippocampus “replay” activity recorded during theta-gamma activity in the exploring animal. Our aim was to develop a mechanistic understanding of cellular and network mechanisms underlying the generation of SWRs, sequence replay during SWRs, and the observed switching to other types of population dynamics such as gamma oscillations. We built a large-scale network model of area CA3 of the hippocampus, and set single-cell and synaptic parameters according to in vitro data. When we used uniform or randomly varying synaptic conductances for all types of connection, there was no sequential activity, and sharp waves with moderate pyramidal cell firing rates and accompanying ripple oscillations were never observed. When recurrent excitatory weights were set by applying an additive spike timing-dependent plasticity (STDP) rule during simulated runs in a circular maze, sharp wave-like activity with ripple oscillations, physiological rates, and accelerated sequential replay of learned activity patterns emerged spontaneously. All of these features of neural activity were robust to scaling the synaptic conductances in a relatively broad range. Application of the recently described symmetric STDP rule enabled both forward and reverse replay as seen experimentally. We then used systematic perturbations of the synaptic weight matrix to explore the links between these different aspects of the neural dynamics and the underlying functional connectivity. The results demonstrated that the distribution of synaptic weights was neither necessary nor sufficient for the physiological activity of the original network. On the other hand, manipulations which destroyed embedded convergent paths in the weight matrix invariably led to the disappearance of both sequential activity patterns and SWR population dynamics, demonstrating a fundamental link between temporal representations (coding) and population dynamics in structured cortical networks.