DCN seminar by Seth Egger

Intelligence requires computing a policy for action by integrating information represented internally with that encoded in sensory populations. To determine the circuit mechanism of this computation, we examine cortical processing during smooth pursuit eye movements in two areas critical to the behavior: area MT, where neurons encode visual motion information and the smooth pursuit eye movement region of the frontal eye fields (FEFsem), a frontal cortical area causally involved in pursuit behavior. First, we determine the drive that MT neurons provide to each FEFsem neuron by analyzing population responses across areas. We then determine how these MT inputs are transformed into an action policy embedded in the FEFsem population response by recurrent circuit connections. Beyond simply integrating MT inputs, the circuit implements interactions between internal representations critical to the computation of the action policy. Together, the results connect sensory drive, individual neuron responses, and circuit dynamics to the formation and manipulation of internal representations critical to achieving intelligent behavior.