Project Summary Different cognitive behaviors appear to engage distinct activity patters across brain-wide circuits. This distributed nature poses a big challenge to understanding which specific activity patterns are causal to different behaviors, for a few reasons. First, it is technically challenging to perturb neural activity at large scales. Second, the same brain regions are often involved in disparate cognitive processes, but they appear to interact and communicate differently depending on behavioral demands. Third, cognitive behaviors do not exist in a vacuum. For example, you may walk around as you deliberate about your future college, but that action is not required for the decision. Thus, to truly understand the neural mechanisms of cognition, we need to use circuit perturbations to disentangle distributed neural activity and interaction patterns that are causal to a behavior from those that are simply incidental to it. Perturbation methods currently available to neuroscientists cannot accomplish this because they tend to target one or few regions at a time, and not account for inadvertent changes in the activity of other interconnected brain regions. To address these challenges, we propose to develop a new set of methodologies for simultaneous, distributed perturbation of multiple cortical regions using patterned light. Specifically, we will design a custom apparatus for head-fixed mice that uses a digital micromirror device to deliver spatially stochastic light patters at cortex-wide scales. This will allow us to borrow concepts from systems identification, used in electrical engineering and sensory-receptive-field mapping, to infer how large-scale patterns of cortical activity underlie decision-making behaviors in a data-driven fashion. Specifically, we will first develop an open- source hardware and software suite to enable these experiments, which will be disseminated to the community at large. We will then perform proof-of-principle experiments in which we will combine spatially stochastic optogenetics with reflectance imaging or extracellular electrophysiological recordings using silicon probes, to estimate distributed cortico-cortical interactions in mice running spontaneously. Finally, we will employ these approaches in mice performing two ground-truth decision-making tasks in virtual reality, for which we have strong expectations for the patters of behavioral deficits caused by the perturbation of different cortical areas. These initial experiments will therefore establish the feasibility and showcase the versatility of our approach. They will also pave the way for future work using these new methods to probe how distributed cortical interactions support complex cognitive tasks. I expect the methods we develop will be readily applicable to multiple other behaviors, neural systems, and model organisms to reveal the elusive causal link between neural interactions and behavioral function.