The working memory system provides the short-term storage facility necessary to perform complex cognitive tasks. The prefrontal cortex (PFC) plays a key role; neurons in the PFC respond to stimuli and continue to maintain elevated persistent activity after the stimulus is removed. Working memory degradation is a common symptom in neurological disease.
I propose that working memory performance is driven by interactions of excitatory and inhibitory neurons and modulated by calcium dynamics. In contrast to previous models that require a carefully constructed or adapting network architecture, this model retains novel stimuli using a fixed network of neurons connected with probabilities only depending on the cell type. Persistent activity is irregular, with the coefficient of variation of the interspike intervals exceeding 0.5. Patterns are robustly maintained even in the presence of distracting stimuli, yet the network switches to new, strongly presented "urgent" patterns. The field potential of the excitatory cells exhibits a gamma rhythm. Statistical properties of the network and the role of key parameters are considered.
I consider the role of subcellular calcium and how the chemical dynamics are affected by electrical activity and dendritic geometry. The role of chemical and electrical feedback is examined, and I conclude with a description of the novel features of a custom simulation tool I wrote to perform this study.