Project Summary Here, we propose to thoroughly characterize the origins of pairwise correlations in cortex using a synergistic mix of experimental methodologies, behavior, and computation in mice and macaques. We will elucidate the mechanistic underpinnings of normalization and test our hypothesis that changes in cortical pairwise correlations and other signature arise from ongoing cortical computations. In Aim 1 we will record from populations of neurons in the middle temporal visual area of trained, behaving monkeys to test the hypothesis that pairwise spike correlations, gamma oscillation and transient responses at the onset of visual stimuli arise in part from the dynamics of the circuits that normalize neuronal responses. These tests require measurements with a precision that is not feasible in mice. Conversely, the experiments in Aim 2 and 3 address questions that are not feasible in monkeys. In Aim 2 we will exploit the accessibility of mouse visual cortex by using both two-photon laser scanning microscopy and multielectrode arrays to comprehensively measure the relationship between normalization and pairwise correlations in populations of V1 neurons and measure how spatial separation within cerebral cortex affects that relationship. Finally, in Aim 3 we will establish the contributions of specific cell classes to normalization and pairwise correlations in mouse V1. We record the activity of pyramidal neurons and the three most thoroughly characterized classes of cortical interneuron (VIP, SST and PV) during normalization. We will then separately manipulate the activity of these cells classes to revealing the role that changes to the ratio of excitation and inhibition play in driving normalization. In this way, we will establish the role these neurons play in changing pairwise correlations within the excitatory pool of neurons. Results from all three Aims will be tied together using a new family of dynamic, recurrent circuit models of normalization to formalize the hypothesis that normalization imposes pairwise correlations and other activity signatures, and will use experimental data to constrain and refine these models.