A small mind.An open-ended composition.
Sound begins with your touch.
Jonny Campbell · The note · The current version
The local fly-brain model generates a 300 ms episode. We stretch its 30 activity frames into a musical phrase. New seeded episodes are prepared in the background; each begins with the network at rest.
Undertone follows total activity. Glass follows downstream activity and MN9. Thread follows a second group of downstream neurons. These groups are arbitrary musical partitions, not biological sound-processing circuits.
JavaScript synthesises every sound: slow sine-wave drones, struck partials, and soft plucks, through filtering, delay and procedural reverb. Pitches and musical rules are composed by us; measured neural activity selects and shapes events.
Energy and disconnection request a new neural episode, usually in a moment. Space, colour and level affect playback immediately or at the next note. Disconnecting leaves sensory activity, and therefore the undertone, but removes downstream voices once the new episode arrives. Existing notes finish their decays.
The particle ring is an artistic visualisation driven by the audio and activity, not brain anatomy. If an episode is late, the last score repeats and is labelled as such. No learning, body simulation or continuous neural state is implied.
Connectome: FlyWire whole-brain connectome (FAFB), public release v630, FlyWire Consortium, CC BY-NC 4.0. Connectivity table and excitatory/inhibitory signs as prepared by Shiu et al. (2024), github.com/philshiu/Drosophila_brain_model (MIT). Modified: re-encoded and compressed for in-browser simulation (data licence). Model after Shiu et al., Nature 634, 210–219 (2024). Please cite Dorkenwald et al. 2024 and Schlegel et al. 2024; neurotransmitter signs from Eckstein, Bates et al. 2024.