PROJECT SUMMARY The ability to decipher meaning from degraded sounds is critical for everyday hearing. A key strategy to extract buried signals is to integrate stored acoustic representations with the incoming sound stream. Such top- down/bottom-up interaction appears to be a fundamental feature in sensory systems. Indeed, disruptions of such processes may be involved in disorders such as autism and dyslexia. Unfortunately, little is known about the mechanisms by which top-down information modulates sensory function. Here, we propose a circuit-level analysis of a massive descending pathway from the auditory cortex to the inferior colliculus (IC) which is thought to be important for top-down modulation. Although this pathway has been shown to be critical for many processes important for auditory perception, how it supports such diversity of function is not known. During the last funding period, we uncovered an unexpected degree of heterogeneity in this pathway. For example, we observed that neurons in the two cortical layers from which this pathway is derived, layer 5 (L5) and layer 6 (L6), have distinct physiological, morphological and connectivity patterns. L5 neurons receive direct thalamic input, burst when stimulated and send projections containing giant terminals to the lateral cortex (LC) of the IC. L6 neurons receive sparse local input, but sample from a broad area of the cortex and end in small terminals on the distal rim of the LC. These data suggest that L5 neurons send a rapid and secure signal to the LC, while L6 neurons modulate LC function based on broad multisensory integration. We also found that the corticocollicular projections interdigitate with periodic neurochemical clusters (“patches” and “matrix”) in the LC and that the presence of this patch/matrix system governs virtually all input and output to and from this structure. These findings raise several questions that will be addressed in the current proposal. First, how does the L5/L6 system interface with the neurochemical and connectional mosaic comprising the LC? To address this question, we will use machine learning tools to parse the neurochemical cell classes in the LC, and then use a novel form of Cre-dependent trans-synaptic tracing to map parallel pathways originating in L5 or L6 and ending at one of the many targets of the LC. Second, we will examine the impact of L5 and L6 terminals on LC neurons using optical and electrophysiological approaches to determine the relative impacts and geometry of these inputs. Third, we will determine what messages are sent by L5 and L6 to the LC by imaging their axons in response to auditory and non-auditory stimulation using a novel microprism-based two-photon imaging approach in awake mice. We can thus determine if L5 and L6 send distinct messages to the LC, and whether the messages vary across the patch/matrix topography. Therefore, by leveraging the patch/matrix organization to parse the dense thicket of inputs and outputs to ...