FLYBOARD

The full 166,700-neuron MaleCNS matrix is stepped by a documented leaky integrator - v decays by exp(-dt/0.1), synaptic input is W times the spike vector plus background, noise and button current, and a neuron spikes and resets at v >= 1 - so voltages are genuinely simulated and the glow is that field. L1. Not L2: the CSR weight array is taken from the graph once in the constructor and never assigned to again anywhere in the package; a repo-wide grep for plasticity, STDP, Hebbian or learning-rate terms returns nothing. The ORGASM button is the obvious trap and the project itself defuses it - it injects current into the real dopaminergic population for a hold window and the README labels it a FAKE REWARD BURST with a 1%-of-population cap. Injecting current into dopamine cells is stimulation, not a learning rule.