Multi-column multi-layer computational model of neocortex

Abstract

We present a multi-layer multi-column computational model of neocortex that is built based on the activity and connections of known neuronal cell types and includes activity-dependent short term plasticity. This model, a network of spiking neurons, is validated by showing that it exhibits activity close to biology in terms of several characteristics: (1) proper laminar flow of activity; (2) columnar organization with focality of inputs; (3) low-threshold-spiking (LTS) and fast-spiking (FS) neurons function as observed in normal cortical circuits; and (4) different stages of epileptiform activity can be obtained with either increasing the level of inhibitory blockade, or simulation of NMDA receptor enhancement. The aim of this research is to provide insight into the fundamental properties of vertical and horizontal inhibition in neocortex and their influence on epileptiform activity. The developed model was used to test novel ideas about modulation of inhibitory neuronal types in a developmentally malformed cortex. The novelty of the proposed research includes: (1) design and implementation of a multi-layer multi-column model of the cortex with multiple neuronal types and short-time plasticity, (2) modification of the Izhikevich neuron model in order to model biological maximum firing rate property, (3) generating local field potential (LFP) and EEG signals without modeling multiple neuronal compartments, (4) modeling several known conditions to validate that the cortex model matches the biology in several aspects,(5) modeling different abnormalities in malformed cortex to test existing and to generate novel hypotheses.