Google Research, HHMI Janelia Research Campus and scientific collaborators have completed a connectome—an exhaustive wiring map—of the adult male fruit fly’s brain and central nervous system.
The dataset covers more than 166,000 neurons and is described as the first complete map for a male fruit fly. It follows a prior complete female fly brain map and gives neuroscientists a second, comparable reference model for an organism central to biological research.
What changed
Connectomics aims to reconstruct the connections among individual brain cells. For this project, researchers sliced the fly brain and body into thin sections, imaged them, and used computational systems and AI to assemble millions of two-dimensional images into three-dimensional neural structures.
Human experts, supported by computing and AI, then classified the resulting cells into 11,691 neuron types. The map spans both the brain and ventral nerve cord—the insect structure analogous to a spinal cord—so it can trace circuits from sensory inputs to motor output rather than treating the brain as an isolated organ.
That scope matters. The resource includes visual-to-motor pathways, including an example route from R1-R6 visual neurons to the DNg13 motor neuron through multiple intermediary cells. It also identifies male-specific LoVP92 neurons associated with the fly’s “love spot,” a region implicated in courtship behavior.
Why the male map is more than a duplicate
Most neurons appear to be shared across male and female fruit flies, but the comparison exposes meaningful exceptions. Some neurons are sex-specific; others are present in both sexes but connect with different neighboring cells. Google describes this latter group as dimorphic.
That distinction enables researchers to ask more precise questions about behavior. A neuron may be structurally present in both animals, for example, while its connections differ in ways that change how sensory information is processed or how actions are triggered.
The newly released map is already supporting companion studies of visual systems, taste and social behavior. Those projects are an early test of the dataset’s value: a wiring diagram is useful not simply because it is large, but because it can make experiments more targeted and biological hypotheses more falsifiable.
The business and technology implication
For AI and research-infrastructure leaders, the project is a concrete example of where machine learning can expand scientific throughput without eliminating expert work. The task required both large-scale image processing and human-led interpretation of neurons and their types.
This model—automating the assembly and analysis of massive experimental datasets, then putting domain experts in the loop for validation and classification—is increasingly relevant beyond neuroscience. Fields that produce high-volume imaging or spatial data, including biology, materials science and medical research, face similar bottlenecks.
It also shows the strategic value of durable research collaborations. The project combines institutional biology expertise from Janelia with Google Research’s computational capabilities and contributions from the broader scientific community. Such work is long-horizon infrastructure building, not a typical product release, but its outputs can become a shared foundation for many downstream studies.
What to watch next
The immediate question is how broadly researchers can use the male and female maps to connect anatomical differences to observed behavior. More companion studies and reproducible tools for interrogating the data will be stronger indicators of scientific impact than the raw neuron count alone.
Google also points to ongoing work on full vertebrate brain maps. Those efforts will be substantially harder: vertebrate nervous systems are vastly larger and more complex than a fly’s. Still, the male fruit fly connectome demonstrates a key operational lesson for ambitious AI-for-science programs: progress depends on pairing better models and computing with carefully designed data pipelines, expert review and research assets that others can build on.
