Abstract Action selection computations are at the core of goal-directed behavior. Models of action selection suggest that appropriate actions are selected through competition between potential choice options. Choice competition is thought to occur across a multi-regional network that span frontal cortex, basal ganglia, and their downstream circuits where neuronal population encoding potential choice options mutually inhibit each other to provide opponent control of choice activity. Yet, how these brain regions interact to mediate this process and the loci of choice competition remain unresolved. We recently identified a frontal cortico-basal-ganglia-superior colliculus network responsible for action selection of directional licking during decision-making. Distinct neuronal populations in this multi-regional network encode opposing choice options for lick direction and exhibit push-pull dynamics prior to a licking movement, reflecting choice competition. Remarkably, activating or suppressing the superior colliculus is sufficient to bidirectionally control the push-pull choice competition dynamics within the network, implicating the superior colliculus as a key network node that can mediate choice competition. These data suggest a working model in which circuits within frontal cortex and basal ganglia encode competition choice options and they influence downstream superior colliculus in a topographically confined fashion to drive opponent control of choice activity for specific actions. Leveraging a suite of recently developed technologies, this proposal aims to precisely define a mesoscale cortico-basal-ganglia-colliculus network for action selection (Aim 1 and 2) and directly probe the interactions of frontal cortex, basal ganglia, and superior colliculus to resolve the loci of choice competition (Aim 3). The outcome will test longstanding theories of action selection and elucidate their neural circuit implementations.