PROJECT SUMMARY/ABSTRACT Higher brain functions include the ability to learn expectations about the world, update those expectations appropriately when given new sensory information, and use those continually updating expectations to guide behavior. How neural circuits implement these flexible information-processing dynamics is not known. We propose a novel research project that, consistent with the goals of the BRAIN initiative, uses innovative, methodologically integrated approaches to understand how activity pattens in a specific circuit in the primate brain support flexible updating used for behaviorally relevant decisions. The circuit includes two main components with known properties relevant to our proposed studies. The first component is the dorsolateral prefrontal cortex (dlPFC), which includes neurons that encode ongoing processing of expectations and sensory evidence in working memory. The second component is the locus coeruleus (LC)-norepinephrine (NE) neuromodulatory system, which can affect working-memory representations in the dlPFC. However, it is not known whether and how LC-NE modulations of working-memory representations in dlPFC contribute to flexible decision-making. Building on our previous work on flexible decision-making and effects of the LC-NE system on neural information processing, we propose and test the hypothesis that LC-mediated NE release in dlPFC governs how dlPFC neural populations flexibly combine learned expectations held in working memory with incoming sensory information to form decisions that guide behavior. We test this hypothesis by training monkeys on a novel task that allows us to quantify how learned expectations and new sensory information are combined in a flexible, context-dependent manner to make saccadic decisions. We then elucidate the underlying circuit mechanisms, via three Aims that each leverage an innovative set of approaches. Aim 1 is to measure how LC and dlPFC activity at a single-neuron resolution relates to flexible decision-making. Aim 2 uses multiple techniques, including electrical microstimulation for temporal specificity and chemogenetics for pathway specificity, to test for causal roles of temporally specific firing patterns of LC->dlPFC projections on flexible decision-making. Aim 3 uses computational modeling to relate LC-dlPFC circuit properties (including NE-mediated changes in neuronal gain) to computational principles that support flexible decision-making. Each Aim alone provides new insights into correlative, causal, and computational contributions of the LC-dlPFC circuit to flexible decision-making. Taken together, these studies provide a novel, unified view of how the LC- PFC circuit performs critical computations that flexibly combine expectations with evidence to inform decisions.