Cortical representation of touch in silico
Display more detailsDisplay less details
Science and technology
Key wordsbiophysical reconstructed model; somatosensory (barrel) cortex; computational neuroscience
This repository contains 10 realizations of calculations of connectivity structures for a model somatosensory (barrel) cortical column (thalamus, L4 and L2/3, see corresponding code and article). There are 5 realizations for a 1-column (1 barrel) networks, and 5 for a 3 column (3 barrels) network. The code to generate this model can be found on github (see url below).The model is a biologically inspired, computationally efficient network model of somatosensory cortical columns. It was created by (1) reconstructing the barrel cortex in soma resolution using multi-channel mosaic scanning confocal microscopy, (2) defining a mathematical model (Izhikevich model) of cortical neurons whose action potential threshold adapts to the rate of ongoing network activity impinging onto the postsynaptic neuron and (3) connecting each neuron in the network using statistical rules of pair-wise connectivity based on experimental observations. The input consists of whisker data (here: angle and curvature, but other metrics such as deviation from baseline are also possible). Based on this, thalamic spike trains are generated (thalamic neurons are considered simple filter-and-fire Poisson neurons), that then form the input to the cortical model.The following files are included in each subfolder:* cellinfo: information about all cells: (Number of Cells-by 6, note that L4 cells come first, followed by L23 cells) matrix, see github for further explanation)* CMDMs: all connectivity data* CMDMs_(…)_ConData* CMDMs_(…)_ParaMat* CMDMs_(…)_ParaMat_reduced: a reduced version of the previous set, to speed up simulations* CMDMs_(…)_WhiskerModulationModel: the model for the direct modulation by motor cortex NB Note that the measured cell densities are in mice, but most axon/dendritic distribution patterns, connectivity and synaptic efficacy in the literature are from rats. To make these fit, the measured cell densities were scaled to fit rat data. So the resulting cell locations are appropriate for rat. Note that once the connectivity data are generated, these cell locations are not used in the network simulations (as neurons are simulated as point neurons).