This paper presents a theoretical study of orientation tuning in the primary visual cortex, focusing on how intrinsic cortical connections contribute to the selectivity of neuronal responses to visual stimuli. The authors develop a simple neural network model that incorporates both orientation-selective input from the lateral geniculate nucleus (LGN) and orientation-specific cortical interactions. They show that orientation selectivity can arise from within the cortex through a symmetry-breaking mechanism, even when the LGN inputs are only weakly anisotropic. The model predicts that the width of the orientation tuning is relatively independent of the contrast and angular anisotropy of the visual stimulus. It also predicts that the transient population response to a change in stimulus orientation exhibits a slow "virtual rotation," and that neuronal cross-correlations exhibit long time tails, whose sign depends on the preferred orientations of the cells and the stimulus orientation. The model is based on a network with an architecture of a cortical hypercolumn, consisting of excitatory and inhibitory neurons that respond selectively to a small oriented visual stimulus. The neurons are parameterized by an angle θ, which denotes their preferred orientation. The interactions between neurons are modeled using functions that depend on the difference between their preferred orientations. The model also incorporates a stochastic dynamics framework, where neurons switch stochastically between a quiescent state and an active state. The activity profile of the network is represented by a continuous function m(θ, t), which denotes the mean activity level of neurons with preferred orientations near θ at time t. The authors analyze the mean-field equations of the network and show that the orientation tuning can be dominated by cortical circuitry when the amplitude of the cortical angular modulation, J₂, is sufficiently strong. They also show that the tuning width is largely independent of stimulus properties such as contrast and anisotropy. The model predicts that the transient response to a change in stimulus orientation exhibits a slow "virtual rotation," and that neuronal cross-correlations exhibit long time tails, whose sign depends on the preferred orientations of the cells and the stimulus orientation. The results are supported by experimental findings and have implications for understanding the role of cortical interactions in orientation selectivity. The study also highlights the importance of neuronal cross-correlations in understanding the cooperativity among cortical neurons.This paper presents a theoretical study of orientation tuning in the primary visual cortex, focusing on how intrinsic cortical connections contribute to the selectivity of neuronal responses to visual stimuli. The authors develop a simple neural network model that incorporates both orientation-selective input from the lateral geniculate nucleus (LGN) and orientation-specific cortical interactions. They show that orientation selectivity can arise from within the cortex through a symmetry-breaking mechanism, even when the LGN inputs are only weakly anisotropic. The model predicts that the width of the orientation tuning is relatively independent of the contrast and angular anisotropy of the visual stimulus. It also predicts that the transient population response to a change in stimulus orientation exhibits a slow "virtual rotation," and that neuronal cross-correlations exhibit long time tails, whose sign depends on the preferred orientations of the cells and the stimulus orientation. The model is based on a network with an architecture of a cortical hypercolumn, consisting of excitatory and inhibitory neurons that respond selectively to a small oriented visual stimulus. The neurons are parameterized by an angle θ, which denotes their preferred orientation. The interactions between neurons are modeled using functions that depend on the difference between their preferred orientations. The model also incorporates a stochastic dynamics framework, where neurons switch stochastically between a quiescent state and an active state. The activity profile of the network is represented by a continuous function m(θ, t), which denotes the mean activity level of neurons with preferred orientations near θ at time t. The authors analyze the mean-field equations of the network and show that the orientation tuning can be dominated by cortical circuitry when the amplitude of the cortical angular modulation, J₂, is sufficiently strong. They also show that the tuning width is largely independent of stimulus properties such as contrast and anisotropy. The model predicts that the transient response to a change in stimulus orientation exhibits a slow "virtual rotation," and that neuronal cross-correlations exhibit long time tails, whose sign depends on the preferred orientations of the cells and the stimulus orientation. The results are supported by experimental findings and have implications for understanding the role of cortical interactions in orientation selectivity. The study also highlights the importance of neuronal cross-correlations in understanding the cooperativity among cortical neurons.