Drosophila melanogaster · male central nervous system · connectome v1.0
Descending neurons are the cables that carry commands from the fly’s brain to its nerve cord. Pick one to see which muscles it can reach, directly or through one or two relay neurons, and whether its net push is excitatory or inhibitory.
Schematic outline, not to scale · coloured lines are descending neurons
This site maps how a fruit fly’s brain is wired to its muscles, from the neurons that carry commands down the neck to the motor neurons that move each body part. Here are the ideas it can show you, each with a place to see it for yourself.
Every command from the brain passes through about 1,300 descending neurons in the neck. Most hand it to relay neurons in the nerve cord, which pass it to the motor neurons that drive muscles. The giant fiber is a rare shortcut: it contacts the jump muscle’s motor neuron directly.
Connections excite (orange) or inhibit (blue), and two inhibitions in a row add up to excitation. One neuron usually fans out to many body parts, while many neurons converge on the same shared relays.
The giant fiber listens to looming detectors (LC4, LPLC2) and drives the jump: an escape reflex from “something is coming” to “take off”. pIP10, a male-specific neuron, recruits known song neurons and the small wing muscles used to sing.
A connection makes an action possible, not certain. The site checks itself against experiments and flags weak or uncertain results. Its simulation recovers song and part of the jump, but misses steering and landing.
What it can’t tell you: how a movement actually unfolds in a living fly. The model has no rhythm, sensory feedback, electrical synapses or neuromodulators, and transmitters are predicted rather than measured. Treat every result as a hypothesis to test.
data.js didn’t load. Run python build.py in the brain-to-body folder, then reload this page.Reach adds up across the neurons, so the body map and wiring maps show them firing together. The activity simulation covers single neurons only.
This browser couldn’t start the 3D view, so the body parts are listed here instead.
Drag to turn the fly; ⌘ or Ctrl + scroll to zoom. Hover a body part for numbers, click it to map the wiring the selected neuron uses to reach it.
Each row is one route a signal can take, read left to right: from the neuron you picked, through any relay neurons, to the motor neurons that move this body part. Click a row to have that route explained in words.
Shade shows the effective share of a motor pool’s synaptic input that traces back to the selected neurons, on a log scale. Hue shows whether those paths add up to excitation or inhibition.
Shade shows the share of a body part’s motor neurons that fire at least once while the selected neurons are driven, in a simulation built on this wiring. Press Play to watch the activity arrive in 50 ms steps. A model, not a recording: see “Simulated activity” below.
Pick an action. The simulation switches on the descending neurons that experiments link to it, or, where none is known, the ones that best recruit the muscles it needs, and records every neuron that fires in the next half-second. It shows which neurons the wiring can recruit, not the movement itself: the model has no rhythm, no body and no sensory feedback.
Loading the actions…