FLYFISH
a traced fish brain hunting a fully wired fly brain · every buy closes the distance
LIVE 0 watching SCAN BUY $FLYFISH

loading two brains

THE PREDATOR
0 traced neurons · Danio rerio larva
0 inferred connections
THE PREY
0 neurons · Drosophila melanogaster
every synapse mapped, FlyWire 783
drag to rotate
SCANNING scanning the water
a 6 day old Danio rerio larva, the animal the brain on the left belongs to. it swims in bursts, not in a line, because the real one does

awaiting launch

NOW

0spikes per second in the fish
0m 0sbrain time lived
0flies eaten
100%distance to the fly
40%hunger
0 strikes 0 misses

WHAT THE FISH WANTS TO DO

every name below is a region that exists in the atlas and does this job in a real hunt

WHAT THEY FEEL · EVERY TRADE OF $FLYFISH

buy = the eyes converge and the fish swims closer · a buy above the 90th percentile of the last 24 hours = the strike · sell = the fish drifts back and gets hungrier

no trades yet. the fish is patient

SCIENCE

The predator

Danio rerio, larva. 4,172 neurons, each one traced by hand and registered into a common brain, coloured here by the anatomical region its cell body actually falls in. The larval zebrafish is the only vertebrate whose entire brain has been recorded neuron by neuron while it was alive and behaving.

Kunst et al. 2019, Neuron. mapzebrain single neuron atlas and anatomical segmentation v2.0.1, 597 x 974 x 359 micrometre standard brain. Whole brain recording: Ahrens et al. 2013, Nature Methods.

The prey

Drosophila melanogaster. 139,248 neurons and 2.7 million connections, every one verified at the synapse. It is the only brain that has ever been wired in full. What you see running through it is the escape: the looming detector fires, then the giant fiber, then the muscle. Five milliseconds, in that order.

Dorkenwald et al. 2024, Nature. FlyWire 783. Annotations: Schlegel et al. 2024.

What we are not claiming

The fish connectome is not finished. What exists publicly is the traced atlas above, plus a whole brain electron microscopy volume still being proofread. The connections we run in the fish are inferred from arbors touching, within four micrometres, not verified synapse by synapse.

So the two brains are complete in different ways. The fly is fully wired. The fish was fully recorded. We say so rather than let you assume otherwise.

The hunt

Prey capture in a zebrafish larva is one of the best described behaviours in any vertebrate. Prey enters the visual field, the optic tectum builds a map of it, the pretectum commits, the eyes converge, a J turn aims the body, approach swims close the gap, and the mouth opens. The larva misses often. So does this one.

The strike lands with a probability between 0.15 and 0.85, from the distance at that moment and how hungry the fish is. Nothing is scripted.