To perform most behaviors, animals must send commands from higher-order processing centers in the brain to premotor circuits that reside in ganglia distinct from the brain, such as the mammalian spinal cord or insect ventral nerve cord. How these circuits are functionally organized to generate the great diversity of animal behavior remains unclear. An important first step in unraveling the organization of premotor circuits is to identify their constituent cell types and create tools to monitor and manipulate these with high specificity to assess their functions. This is possible in the tractable ventral nerve cord of the fly. To generate such a toolkit, we used a combinatorial genetic technique (split-GAL4) to create 195 sparse transgenic driver lines targeting 196 individual cell types in the ventral nerve cord. These included wing and haltere motoneurons, modulatory neurons, and interneurons. Using a combination of behavioral, developmental, and anatomical analyses, we systematically characterized the cell types targeted in our collection. In addition, we identified correspondences between the cells in this collection and a recent connectomic data set of the ventral nerve cord. Taken together, the resources and results presented here form a powerful toolkit for future investigations of neuronal circuits and connectivity of premotor circuits while linking them to behavioral outputs.

Dopaminergic neurons (DANs) drive learning across the animal kingdom, but the upstream circuits that regulate their activity and thereby learning remain poorly understood. We provide a synaptic-resolution connectome of the circuitry upstream of all DANs in a learning center, the mushroom body of Drosophila larva. We discover afferent sensory pathways and a large population of neurons that provide feedback from mushroom body output neurons and link distinct memory systems (aversive and appetitive). We combine this with functional studies of DANs and their presynaptic partners and with comprehensive circuit modeling. We find that DANs compare convergent feedback from aversive and appetitive systems, which enables the computation of integrated predictions that may improve future learning. Computational modeling reveals that the discovered feedback motifs increase model flexibility and performance on learning tasks. Our study provides the most detailed view to date of biological circuit motifs that support associative learning.