Fig. 7: Recurrent interaction modeling can explain the existence of resilient and vulnerable neurons.
From: Cortical recurrence supports resilience to sensory variance in the primary visual cortex
a Ring topology of the network, with the preferred orientation of each neuron. Inset: Recurrent connectivity profile for each neuron, computed as a difference (black) of excitatory (red) and inhibitory (blue) profiles, controlled by a measure of concentration κexc and κinh, respectively. b VTF without recurrent connectivity (i.e., only inputs convolved with receptive fields) and with varying RF’s Half-Width at Half-Height (HWHH). The CV identity curve is shown in black. Inset shows examples of receptive fields RF. c VTF with recurrent connectivity, under two configurations retrieved by searching for VTF parameters close to those of neuron A, B (Fig. 2b) and C (Supplementary Fig. 1). d Delay to half maximum firing rate of the model (τ) for each connectivity profile, shown as a contour plot of κexc and κinh. [5%;95%] range of the parameters corresponding to the VTFs in (c) are displayed below the scale bars. e VTF parameters obtained from the model for each connectivity profile, shown as a contour plot of κexc and κinh. [5%;95%] range of the parameters of the VTFs shown in (c) are displayed below the scale bars.
