Modeling Nonlinear Dynamics in Slices of Primary Visual Cortex
Although the primary route of information through the layered structure of the primary visual cortex (V1) is generally known, the quantitative details of V1¡¦s internal transmission are only recently being unpacked. One high-throughput method for quantifying layer-specific transmission strengths is laser-based glutamate uncaging (LGU) combined with voltage-sensitive dye (VSD) imaging: glutamate is uncaged with great spatial specificity, activates the neurons in its immediate neighborhood, and VSD reveals their targets. This thesis presents biologically-inspired computational models of a V1 coronal slice whose parameters are informed by LGU-VSD data and whose outputs in turn replicate the visual features of outputs from LGU-VSD experiments. The modeling exercise¡¦s biggest success, due to the correspondence between mathematical model features and biological features of V1, is that it provides mechanistic explanations for experimental observations. The most impressive mechanistic insight explained findings that L4,,³L23 activity spread was the inter-layer projection both most suppressed by the blockage of excitatory NMDAR (ion channels that are receptive to N-methyl-D-aspartate) currents and most enhanced by the blockage of inhibitory gamma-aminobutyric acid (GABA) currents. I replicated these observations by modeling this projection as more broad and shallow than the other projections. Without modeling, the suitability of this mechanistic explanation could not have been directly tested, nor an alternative explanation ruled out. Moreover, single-cell receptive field mapping data collected subsequently to this modeling finding¡¦s publication revealed that L4,,³L23 connections were, as compared to other projections, plentiful but weak relative to the inhibition acting on the target layer.