Meeting ID: 880 887 9442 Passcode: XC9V1u Yuri Dabaghian Learning and memory are fundamentally collective phenomena, brought into existence by highly organized spiking activity of large ensembles of cells. Yet, linking the characteristics of the individual neurons and synapses to the properties of large-scale cognitive representations remains a challenge: we lack conceptual approaches for connecting the neuronal inputs and outputs to the integrated results at the ensemble level. For example, numerous experiments point out that weakening of the synapses correlates with weakening of memory and learning abilities—but how exactly does it happen? If, e.g., the synaptic strengths decrease on average by 5%, then will the time required to learn a particular navigation task increase by 1%, by 5% or by 50%? How would the changes in learning capacity depend on the original cognitive state? Can an increase in learning time, caused by a synaptic depletion, be compensated by increasing the population of active neurons or by elevating their spiking rates? Answering these questions requires a theoretical framework that connects the individual cell outputs and the large-scale cognitive phenomena that emerge at the ensemble level.
We propose a modeling approach that allows bridging the “semantic gap” between electrophysiological parameters of neuronal activity and the characteristics of spatial learning, using techniques from algebraic topology. Specifically, we study influence of synaptic transmission probability and the effects of synaptic plasticity on the hippocampal network’s ability to produce a topological cognitive map of the ambient space. We simulate deterioration of spatial learning capacity as a function of synaptic depletion in the hippocampal network to get a better insight into the spatial learning deficits (as observed, e.g., in Alzheimer’s disease) and understanding why development of these deficits may correlate with changes in the number of spiking neurons and/or of their firing rates, variations in the “brain wave” frequency spectra, etc. The results shed light on the principles of spatial learning in plastic networks and may help our understanding of neurodegenerative conditions.