Janelia Research Campus, Howard Hughes Medical Institute
We study vision in flies to understand where behavior comes from.
Why flies
Flies use vision for many behaviors, and every neuron and connection in the fly visual system has now been mapped.
Now
This T4 neuron detects motion in one direction. We are measuring the activity of the mapped neurons during behavior, and automating our experiments so that we can study thousands of cell types.
The lab
We study how vision works, from single neurons processing light signals to brain-wide circuits that transform visual information into behavior. The fly visual system contains multiple interacting subsystems that support diverse behaviors, and its structure is now fully mapped. We use it to test long-standing theories of how brains work.
Our work combines connectomics, quantitative behavior, neurophysiology and computational modeling. We build our own instruments and share them, along with our data and code.
Research
Motion vision is our proving ground. Using whole-cell recordings from T4 and T5 neurons, the first direction-selective cells in the fly visual system, we built models that explain how these cells turn non-selective inputs into direction-selective signals. We are now mapping which receptors sit at which synapses, to learn what sets the timing this computation depends on.
With the FlyEM team and many colleagues, we cataloged every cell type in the fly optic lobe from a complete wiring diagram, and traced how visual information travels from the eye into the central brain. We found that the structure of the eye shapes the responses of neurons deep in the brain. We use these maps to choose which neurons to study.
Vision is inherently active, so no understanding of vision can be complete without behavior. We have identified neurons that detect the fly's own movement and steer walking, and neurons that detect approaching objects. We are extending this work to visual learning and to social behavior in complex environments.
The fly brain has thousands of cell types, and the detailed methods we use to study one circuit at a time do not scale to all of them. We are rebuilding our experimental approach with custom instruments for higher-throughput rigs, robots that handle flies, automated two-photon imaging, and adaptive experiments in which data from one experiment shape the next. Our approach is human-centered and AI-accelerated.
Resources
We share an explorer for every cell type in the fly central nervous system, galleries of visual pathways, the designs for our LED displays and rigs, and the code behind our papers.
Every cell type in the complete male central nervous system, 11,751 types and 164,838 neurons, with morphology, connectivity and spatial coverage. It includes and supersedes the optic lobe explorer.
male-cns v1.0Cell-type pages for visual projection neurons and visual central brain neurons, with input barcodes and predicted receptive fields. Companion to Hoeller et al. 2026.
Documentation for the G3, G4 and G6 generations of LED arena displays for visual neuroscience: hardware, firmware and software.
Open hardware for fly experiments: spherical treadmill, tethering station, fly holders, sarcophagus, and more, as 3D-print, laser-cut and PCB files.
Recent papers
Join
We are a small, cross-disciplinary group of biologists, engineers and computational scientists. We build our own tools and share them. Our current focus is scaling up experiments by automating fly handling, behavioral phenotyping and two-photon imaging.
Open position · posting November 14, 2026
Develop and validate methods for automating fly behavior and two-photon neurophysiology experiments, and collect the large, high-quality datasets they make possible. You will build automated pipelines from fly preparation through stimulus delivery, acquisition and quality control, design adaptive protocols in which incoming data select the next experiment, and relate the results to connectome-based models of visual circuits.
We welcome candidates from any technical background who are passionate about experimental design and execution; neuroscience, bioengineering or another quantitative field is ideal. Hands-on experience with AI-assisted coding is required. Experience with Drosophila is not required.
Write to MichaelThe official posting goes live on November 14, 2026; the link will appear here.
We are always glad to hear from potential postdocs; Janelia recruits year-round and postdocs here do not write grants. Send a CV, a one-page statement of what you have done and what you would like to do here, and the names of two or three references. Open positions, when we have them, are also posted on the Janelia job listing.
PhD students join through the Janelia and Johns Hopkins programs: the Joint Graduate Program in Neuroscience or the Cross Disciplinary Graduate Program in Biomedical Sciences (XDBio). You apply to Johns Hopkins (deadline December 3), name the Janelia labs you are interested in, spend the first year in Baltimore, and rotate at Janelia the following summer before thesis work here. Johns Hopkins students interested in a rotation should write to us directly.
Summer undergraduate research at Janelia runs through the HHMI Cech Fellows Program: nine weeks, a stipend, and on-campus housing. Applications open in November for the following summer. Apply through HHMI; the program asks that you not contact labs about your application.
Early-career programmers can work with the lab through Janelia's Scientific Computing Associate (SCA) program, a two-year appointment embedded in a lab and in the Scientific Computing team, to build research software with an open-source portfolio.
Janelia's Visiting Scientist Program supports stays from a week a month to a full sabbatical and covers travel and housing. Contact us before submitting a proposal; proposals are accepted at any time.
The lab builds its own instruments, including LED display arenas and automated behavior rigs. Engineering and technician roles are posted on the Janelia job listing. If nothing is listed, you are welcome to email us.