The retina is unequaled as a model system for linking genes to cells and synapses to brain mechanisms underlying purposive behaviors. >30 types of retinal ganglion cells (RGCs) convey specialized output signals to the visual centers of the brain. Each type is distinctive in form, response properties, brain connections, and functional roles. A key step in understanding the mechanistic basis of these parallel visual representations is to work out the ‘wiring diagram’ linking each type to bipolar and amacrine cells. Almost nothing is known about such wiring for the great majority of RGC types. Serial electron microscopic (SEM) reconstruction provides a unprecedented opportunity to map these synaptic circuits in all their richness, and my lab has been engaged for several years in an ambitious project to do just that. Here we propose a series of projects grounded in this approach that should fundamentally augment our ability to explain how the stimulus selectivity of individual RGC types arises from their connectivity patterns. Our first goal is to provide the first definitive accounting of the connectivity between all the known types of bipolars cells and virtually every major type of RGC. Bipolar cells relay photoreceptor signals to the inner retina, but there are about 15 distinct bipolar types differing in response polarity, cone contributions, and temporal kinetics. Using an SEM volume in which we have already reconstructed the majority of RGC and bipolar cells, we will generate a comprehensive, unbiased bipolar-to-RGC connectome. We will then combine SEM and functional studies to probe two sets of RGC circuits inferred from the literature and our preliminary data. In the first of these subprojects, we will probe specialized bipolar and amacrine networks that permit ipRGCs to encode luminance and which appear to provide a route by which ipRGCs can exert inhibitory control over the primary rod pathway. Because ipRGCs provide a stable representation of luminance, we surmise that this network represents a key system for shutting the rod system down under bright light conditions, a form of network light adaptation. In the second subproject, we will probe two aspects of the retinal circuits underlying image stabilization, traceable to a different RGC type: the ON direction-selective (ON-DS) cells. We will test the idea that the distinctive speed tuning of these cells is attributable to one or more specific amacrine-cell types that veto the response to rapid motion. We will test the hypothesis that speed tuning varies topographically in these cells to match the geometry of optic flow produced by self-motion. We will also assess the asymmetric connectivity from starburst amacrine cells to ON-DS cells to test the hypothesis that such asymmetry (like the DS it produces) also varies topographically and is fully congruent with that of ON-OFF DS cells.