What The Fly Hears

Johnston's organ → AMMC → wedge
responding neurons 9,357

Now playing

Courtship song

drag to rotate · scroll to zoom

what's real, below ↓

What's real here, and what's a joke

The anatomy and the neural responses on this page are genuine — a real connectome, real neuron positions, a real simulation. Any claim that a fly likes, prefers or enjoys a song is a joke. This model has no reward circuit and no capacity to learn, so it cannot represent enjoying anything. It shows which neurons a sound reaches. That's the whole result, and it's strange enough on its own.

● Measured

Every neuron's position is its real soma coordinate from the MaleCNS connectome.
165,122 neurons and 89.7 million synapses, simulated with leaky integrate-and-fire dynamics.
9,357 neurons responded to sound, measured against a silence control at matched random seeds.
Flies really do hear only ~100–500 Hz, through the antenna, over millimetres.
A 1 kHz tone really does nothing. It's above their hearing. That's the control that proves this isn't a music visualiser.

● Approximated

The colours are arbitrary. Frequency bands were assigned to ear-receptor subgroups alphabetically, not by measured tuning. Which neurons respond is real; what colour they glow is not.
Two parameters were hand-tuned so the network wouldn't run away. A connectome fixes wiring, not conductance.
Synapse count stands in for strength — the dataset doesn't contain real synaptic weights.
Only 5.7% of neurons respond to sound at all. Most of the brain stays dark, correctly.

● Satire

"The fly likes this song." It can't. No reward circuit, no dopamine, no learning.
"It prefers X to Y." A joke about which stimulus drove more neurons — not a preference.
"Its brain lights up like it's mating." No mating was simulated. Courtship song here is a sound the fly hears, nothing more.
Anything about taste. The fly has no opinion. It has an antenna.

Stimulus	Audible to a fly	Ear response
Fly sine song — 150 Hz, courtship	100%	+20.02 Hz
Melody — G3 · B3 · D4 · G4	100%	+3.78 Hz
Fly pulse song — 250 Hz, 35 ms apart	98%	+3.16 Hz
Drum loop — 120 bpm	15.6%	+0.23 Hz
White noise	8.4%	—
1 kHz tone — above fly hearing	0%	+0.08 Hz

Built on the MaleCNS connectome — a complete map of an adult male Drosophila central nervous system, published 3 September 2026 by HHMI Janelia, the MRC Laboratory of Molecular Biology, the University of Cambridge and Google Research. Licensed CC-BY. I did not build the connectome — it took years and a large team. I downloaded it and simulated it.

Simulation follows Shiu et al., Nature 2024. Sound is band-limited to the fly's audible range and delivered to Johnston's Organ, the fly's ear. Responses are precomputed; your browser only does the live mixdown.