Control experiment
Identical environmental input, two graphs. The left fly is the published FlyWire wiring. The right fly uses a degree-preserving shuffle of the same synapse counts. If behaviour matches, biological specificity is not doing the work.
Real FlyWire wiring
Time in current zone: —
idle at 0.00, 0.00
Shuffled edges
Time in current zone: —
idle at 0.00, 0.00
Lab stimuli
Independent of Bitcoin. Drive identified sensory neurons directly.
Model parameters
Computational, not biological measurements. Not fitted to BTC.
Activity trace retained each step after a putative firing event.
Input must exceed this before a neuron contributes new activity.
Scales the tanh of supra-threshold drive.
Converts log(1 + synapse count) into a unitless weight.
Inhibitory (GABA/glutamate) synapses are down-weighted relative to acetylcholine so a rate model does not veto the whole antennal lobe. Connectivity itself is unchanged.
Small model noise from a seeded RNG. Not a trading randomiser.
Does wiring matter?
Both flies receive the same odour, wind and looming. Only the connectome graph differs. The live experiment always uses the real graph; shuffled exists only here.
Real NEUTRAL vs shuffled NEUTRAL
Why did the fly do that?
- Strongest environmental stimulus
- — (0%)
- Primary active cell
- —
- FlyWire ID
- —
- Strongest pathway mass
- —
- Dominant descending output
- —
- Current behaviour
- idle
- Observed indication
- NEUTRAL · 0%
Real connectome subgraph
0 neurons · 0 edges · 0 active
Left: sensory · centre: interneurons · right: descending. Brightness is simulated activity on real FlyWire cells. Cyan traces are currently conducting edges. Click a neuron to inspect its published identity.