Project Summary Auditory-guided behavior is ubiquitous in everyday life, whenever auditory information is used to guide the decisions we make and the actions we take. One such behavior is auditory categorization, a process that reflects the ability to transform bottom-up sensory stimuli into discrete perceptual categories and use these perceptual categories to drive a subsequent action. Although this process is well-documented at the behavioral and cognitive levels, surprisingly little is known about the explicit neural circuit mechanisms that underlie categorical computation and how the result of this computation drives behavioral outcomes. We believe that the transformation of auditory information into an appropriate behavioral response is necessarily a brain-wide endeavor. The deep layers of the auditory cortex give rise to several massive projection systems that exert influence over many downstream brain areas. Of these, extratelencephalic (ET) neurons within layer 5b have long been regarded as canonical “broadcast” neurons, pooling inputs from a variety of sources and transmitting signals throughout the brain. These neurons are unique in that they provide the only direct connection between the neocortex and various behaviorally relevant subcortical structures, placing them in a privileged position where they can readily influence auditory-guided behavior. To understand the role that ET neurons play in auditory-guided behavior necessitates in vivo, cell-type specific recordings, in awake behaving animals. To this end, we have designed a novel auditory categorization task that can be readily learned by head-fixed mice. Our preliminary data, both anatomical and physiological, posits that ET neurons become selective to discrete perceptual categories across learning, and this selectivity is mediated by top-down input from higher-order cortex. The goal of this proposal is to leverage cutting-edge techniques to test three specific hypotheses: (1) ET neurons are necessary for auditory categorization, and this necessity is both learning-dependent and specific to distinct axon collaterals (Aim 1), (2) ET response properties change across learning to reflect discrete perceptual categories (Aim 2), and (3) ET learned categorical selectivity is shaped via top-down inputs from higher-order cortex that act as a flexible, task-dependent filter (Aim 3). Combined, this research will take an important first step towards understanding the role of descending circuits in auditory-guided behavior will unveil the greater auditory pathway that lies beyond its classical terminus in primary auditory cortex.