Project Summary In mammalian neocortex, sensory stimulation evokes highly structured patterns of neural activity. How groups of neurons with specific patterns of neural activity encode information is a question remaining at the forefront of systems neuroscience. Recurrent connections between cells with similar response properties are proposed to amplify stimulus features and serve as building blocks of neural ensembles, but how competing ensembles encoding different stimulus features interact to form a single coherent percept is unknown. In the case of vision, natural visual scenes are often comprised of multiple stimulus features that represent the subject or the background of a visual scene, each of which evoke competing and overlapping ensembles. Therefore, a fundamental computational problem for the brain is the parsing of salient features from the background of a visual scene. Cortical circuits must be able to quickly determine features that correlate to the primary figure and amplify ensembles encoding representation of those features. The overall hypothesis of this proposal is that like-to-like inhibition, mediated by somatostatin-expressing interneurons, suppresses competing ensembles to amplify ensembles encoding unique stimulus features. I hypothesize these mechanisms underly canonical cortical computations, such as surround suppression, as well as visual perceptual phenomena, such as figure-ground segregation. To test this hypothesis, this study uses a combination of in vivo holographic optogenetics, 2-photon calcium imaging, and behavioral assays in an effort to understand the mechanisms underlying the neural basis of perception. The specific aims of this project independently examine the circuit basis of orientation selectivity of surround suppression and tests whether these circuits are a neural correlate of figure-ground perception in mouse primary visual cortex. Taken together, these experiments will test the more fundamental ideas of the nature of neural circuits involved in perception and how percepts are represented in the brain. This proposal will therefore provide a basis for understanding psychiatric and neurological diseases that have yet eluded mechanistic understanding and medical treatment.