Poster D50, Monday, March 27, 8:00 – 10:00 am, Pacific Concourse
sFROST: a Spiking Model of Working Memory Maintenance
Zachary Hutchinson1, Sebastien Helie2, Shawn W. Ell1; 1University of Maine, 2Purdue University
Neurocomputational modeling is a powerful tool for grounding explanations of cognitive processes within neuroanatomical and neurophysiological constraints. The FROST (FROntal-Striatal-Thalamic) model of working memory maintenance proposes that to-be-remembered items are encoded in the lateral prefrontal cortex and maintenance depends upon reverberant activity in a network of cortical and subcortical structures (i.e., basal ganglia, thalamus). The original version of FROST predicted continuous membrane potentials. In the present work, we develop a spiking version of FROST (sFROST) that offers an increase in neurobiological realism. Benchmark simulations demonstrate that sFROST successfully accounts for single-cell recording data and human behavioral data. The impact of scaling the model (i.e., increasing number and connectivity of neurons) and adding a more detailed basal ganglia network will be investigated.
Topic Area: EXECUTIVE PROCESSES: Working memory