Abstract The computations performed by circuits in the visual system are numerous and varied, emerging from many cell types with unique combinations of synaptic inputs, molecular composition, and physiological properties. How are distinct computations distributed across neurons and circuits in the visual system, and how are anatomical and molecular features differentially engaged to support them? Addressing these fundamental questions requires a model with deep, cell-by-cell descriptions of gene expression and morphology. This project will examine computation by assessing the varied representation of visual features across cell types in the Drosophila medulla, the first site of major synaptic divergence and signal expansion in the fly visual pathway. By extending established genetic tools, this study will record visual selectivity in most of the ~70 medulla cell types in a fraction of the time required for traditional approaches. This broad physiological assessment will then be leveraged to explore how differences in wiring and gene expression relate to different visual computations. First, a linear/non-linear sum-of-inputs model, weighted according to data from the fly connectome, will describe the extent to which a neuron’s activity can be explained by its feedforward input. Second, a large-scale regression of response features with cell type-specific transcriptome data will reveal how genetic suites and individual molecules contribute to visual feature selectivity. And finally, manipulations of modulatory circuit elements will uncover how feedback and laterally connected cell types contribute to visual computation. Collectively, these experiments will describe how different forms of synaptic input and distinct genetic programs are differentially recruited to build the computations necessary for proper visual processing. The results of this study have the potential to qualitatively improve critical technologies related to the quality of human life, including retinal prosthetics and machine vision.