Visual processing has been extensively studied in anesthetized animals and awake non-moving, visually fixating monkeys. However, little is known about how self-generated visual feedback that arises naturally during eye and body movements is processed. We address this question in behaving mice actively navigating a virtual environment, passively viewing visual stimuli or moving in absence of any visual stimulus in complete darkness. To this end mice are head-fixed on a spherical treadmill, which is coupled to a simple virtual environment that provides visual flow in response to the mouse walking on the treadmill. This approach allows us to compare visual processing to identical visual stimuli in self-generated and passively viewing conditions, and allows us to quantify the contribution of visual and motor-related input to visual cortex. In addition we can probe neural responses to brief perturbations of the coupling between locomotion and visual feedback. We measure the activity of neurons and afferent axons in V1 using methods of two photon imaging of genetically encoded calcium indicators. In previous work, we were able to show that the responses of most cells cannot be explained based solely on a classical, passive, feed-forward model of visual processing. On average the activity of V1 is better explained by locomotion than by visual input alone. In addition we could show that a subset of cells responds selectively to mismatch between locomotion and visual flow. We could now show that a subset of these mismatch selective cells has spatial receptive fields, and we were able to identify a cortical source of motor related input to V1. We interpret these data to suggest that mouse V1 receives strong feedback input, potentially in the form of a corollary discharge which in turn could be used to amplify neural responses to unanticipated visual input.