Project Summary Each movement of the eyes is the outcome of a competition between the stimulus at the fovea, the target of the movement, and other potential targets. The frontal eye field (FEF) is thought to maintain a map of priority for saccadic eye movements, combining information about the salience of stimuli with their behavioral relevance to guide the flow of eye movements. The objective of this work is to understand how priority maps form in FEF. The overall hypothesis is that FEF neurons form these maps by learning to anticipate saccades and as a result, integrate a wide range of sensory, motor, and cognitive signals into a singular representation that reflects expectations about the timing and probability of saccades. Recently, we developed a novel simulation of associative learning during natural oculomotor behavior that permits measurement of the visuomotor properties of neurons that arise from a given learning goal. The first aim is to use this simulation to determine if FEF neurons in silico develop visuomotor properties like FEF neurons in vivo when they learn to anticipate saccades. We will simulate several FEF networks, each designed to learn distinct goals related to oculomotor behavior and characterize the properties that develop in each. Our preliminary data suggests that when neurons anticipate movement goals, their visuomotor properties capture many important characteristics of FEF neurons. Specifically, the modeled neurons develop dual visual- and movement-related responses, their visual sensitivity shifts across space around the time of saccades, and they respond more vigorously when visual stimulus in their receptive field is the target of a saccade. Furthermore, both the visual and movement responses are relatively early, consistent with a short-latency subpopulation of FEF neurons. The second aim is to assess a prediction of the model, that expectations about saccades are encoded in FEF visual responses. To do this, we will record single-neuron activity in the FEF of monkeys while they complete blocks of a delayed saccade Go/NoGo task. The model predicts that manipulations of the probability or time at which a saccade follows a visual stimulus will modify subsequent visual responses. The third aim is to determine how reward affects the visual sensitivity of FEF neurons. We will alter the reward contingences in the Go/NoGo task to test if FEF neurons encode reward- related information or if their apparent sensitivity to reward is at the service of encoding movement-related information. The outcomes of the second and third aims will be applied to improve the model as needed and generate new predictions. Collectively, this work will establish how the visual responses of FEF neurons are shaped by experience about saccades and reward and provide a rigorous basis for understanding the formation of priority maps in FEF. This basic knowledge is a prerequisite to identifying the source of deficits in saccadic behavior in disorder...