PROJECT SUMMARY Understanding the causal interactions in large neural ensembles is key for developing techniques to alter cognitive behavior through targeted manipulation of the brain. This is a challenging goal because commonly used methods for recording neural responses in the human brain do not provide information about physical connections of neurons and allow only extremely sparse sampling of neurons in a circuit (typically <1%). Here, we develop an innovative path forward using a multi-disciplinary approach that combines recent theoretical and experimental advances by the two PIs (Kiani and Mazzucato). In Aim 1, we introduce a novel theoretical framework to infer a map of causal functional connectivity (CFC) based on sparse sampling from neurons in a circuit. Our framework successfully recovers the structure of functional interactions, identifies hub neurons in the circuit, and has multi-scale properties that make it applicable on a variety of data, ranging from spiking of individual neurons to aggregated spiking of clusters of neighboring neurons to local field potentials. In Aim 2, we test if the CFC inferred from a population of simultaneously recorded prefrontal neurons successfully predicts how microstimulation perturbs neural activity in the circuit. Specifically, we show the existence of hub neural clusters, identified through CFC, whose microstimulation has large and predictable impacts on the population response dynamics. Finally, in Aim 3, we explore if the CFC and perturbation effects at rest predict how microstimulation alters behavior during a perceptual decision-making task. We hypothesize that resting CFC combined with the population activity prior to microstimulation successfully predicts the effect of microstimulation both on the circuit activity and the behavior. The approach, data and analyses proposed in each of these aims are novel and the combination will provide a practical solution for a long-standing problem in systems neuroscience.