Stabilized Supralinear Network Model of Responses to Surround Stimuli in Primary Visual Cortex

Abstract

In the mammalian primary visual cortex (V1), there are complex interactions between responses to stimuli present in the cell’s classical receptive field or “center” and in the surrounding region or “surround”. The circuit mechanisms underlying these behaviors are likely to represent more general cortical mechanisms for integrating information. Here we develop a circuit model that accounts for three important features of surround suppression (suppression of response to a center stimulus by addition of a surround stimulus): 1) The surround stimulus suppresses the inhibitory as well as excitatory currents that the cell receives; 2) The strongest suppression arises when the surround orientation matches that of the center stimulus, even when the center stimulus orientation differs from the cell’s preferred orientation; and 3) A surround stimulus of a given orientation most strongly suppresses that orientation’s component of the response to a plaid center stimulus (“feature-specific suppression”). We show that a stabilized supralinear network (SSN) with biologically plausible connectivity and synaptic efficacies that depend on cortical distance and orientation difference between units can consistently reproduce phenomena (1) and (3), and, qualitatively, phenomenon (2). We explain the mechanism behind each result. We argue that phenomena (2) and (3) are independent: the model with some aspects of the connectivity removed still produces phenomenon (3) but not (2). The model also reproduces the rapid time scale of activity decay observed in mouse V1 when thalamic input to V1 is silenced. Finally, we show that these results hold both in networks with rate-based and conductance-based spiking units.

Significance Statement A visual neuron responds to stimuli present in its classical receptive field or “center”. These responses are modulated in a complex manner by stimuli outside the receptive field i.e in “the surround”. Understanding the underlying circuit behind center-surround interactions is crucial to understanding fundamental brain computations. Here, we focus on a set of key center-surround phenomena in the primary visual cortex. We demonstrate how complex aspects of cortical computation can be carried out by the local cortical circuit. We demonstrate that the stabilized supralinear network, a mechanism that accounts for a multitude of cortical response properties, can also account for these phenomena, given appropriate connectivity. We also demonstrate that this mechanism can be achieved in a biologically realistic spiking network.