Neurons in the nervous system communicate by spiking, which activates synaptic connections via the release of neurotransmitter molecules. Modification of the strength of these synaptic connections, known as plasticity, is a mechanism by which networks of neurons can exhibit learning. Previously, a biophysical model of a rodent lateral amygdala was developed that could learn and store auditory fear and extinction memories following classical Pavlovian fear conditioning [1]. We propose a novel reduced order model that preserves the learning capabilities of the detailed model with considerably fewer computations while providing additional insights into the synaptic learning process. To capture the dynamics of individual cells, we propose enhancements to the Wilson-Cowan firing rate model that permit “full” spike frequency adaptation and a non-zero threshold. To incorporate synaptic learning dynamics, we propose a regression technique to capture the nonlinear relationship between firing rate and synaptic [Ca2+]. The resulting method provides a general technique to develop neuronal networks that employ [Ca2+]-dependent synaptic learning.

This content is only available via PDF.
You do not currently have access to this content.