An outstanding question in Neuroscience is to understand how the brain integrates sensory and non-sensory information. Strikingly, this integration takes place even in the primary sensory cortices, where sensory processing is strongly modulated by non-sensory context. For example, in primary visual cortex (V1), responses depend not only on visual stimulation, but also on locomotion (Niell and Stryker 2010, Neuron 65). Specifically, locomotion controls the spatial integration of V1 neurons, reducing the suppression they receive from the surround of their receptive fields (Ayaz, et al. 2013, Curr Biol 23). I will investigate the neural mechanisms underlying this phenomenon. My working hypothesis is that different classes of cortical interneurons contribute in a very specific manner to surround suppression and to modulation by locomotion. I will test this through a combination of experiment and computational modeling. First, I will measure how the different neural classes are activated for different stimulus sizes during rest and running. I will use 2-photon imaging to record the calcium activity of neurons in the superficial layers of V1 in mouse lines where I can simultaneously identify the classes of the interneurons thanks to a red fluorescent protein. Second, I will summarize these measures with a neural network model in which the different classes of neurons will be described by different variables, to match the experimental data. The circuits and computations uncovered by this project are likely to generalize to other areas of the cerebral cortex, and therefore to constitute fundamental organizational principles for cortical integration of internal and external inputs.