The virtual trial — who the model says responds, and who does not

Across a synthetic population of eighty emerged cerebra, stimulant responders are gain-dominated and non-responders are wiring-dominated and partly cap-rescuable. An honest refutation shows the cap-unrescued residual is a dose-cap and stiffness limit, not a severe-wiring tail, arguing for amplitude matched to the deficit. Every fraction is an in-silico coupling state. efficacy=0.

The closing chapter of Part II runs the model across a heterogeneous synthetic population — eighty emerged cerebra, each with a sampled fault mix (wiring, gain, threshold) and stiffness spread — under the stimulant, the cap, and both, and reads off the distribution of outcomes. Stimulant responders are gain-dominated; non-responders are wiring-dominated and partly rescued by the cap; and a bundled sub-claim is refuted and promoted to a finding: the cap-unrescued residual is not a pure severe-wiring tail but a dose-cap / stiffness limit. Every fraction is an in-silico coupling-state outcome, not a clinical response rate — efficacy = 0.

The population and what is measured

The population is eighty emerged cerebra (seventy-six affected), each assigned a sampled mix of wiring, gain and threshold severity plus a stiffness dispersion, with the patient geometry held bit-faithful to the autism endpoints (no wiring → health synchrony, full wiring → the wiring-deficit synchrony). Each is run under a dose-capped stimulant titrated to target, the cap at the window amplitude (duty-cycled), and both. The question is not “does it work” — efficacy is zero — but how the coupling-state outcomes distribute against the fault mix and the cap amplitude.

Stimulant responders are gain-dominated

The stimulant-responder fraction falls monotonically across rising-wiring strata — 0.76, 0.42, 0.00 for low, mid and high wiring — and correlates negatively with the wiring share of the fault (r = −0.60). This is the population image of the single-patient result: gain fixes the gain axis, not the wiring. A patient whose fault is mostly gain responds to the stimulant; a patient whose fault is mostly wiring does not.

Over-synchronisation is an amplitude phenomenon

The adverse (over-sync) fraction rises monotonically with cap amplitude — 0.14, 0.57, 0.99, 1.0, 1.0 across inj 0.08/0.15/0.30/0.60/0.90 — and is minimised at the window. Amplitude sets the instantaneous adverse fraction; duty-cycling sets the cumulative exposure (the molecular load, per §20). Together these say the same thing the operating principle says: the window amplitude is not a preference but the minimum that routes without over-syncing the milder cases.

The honest refutation: the residual is a dose-cap, not a wiring tail

The core claim holds: stimulant-non-responders carry a higher wiring share (0.42) than responders (0.21), and the cap rescues 32% of them by pacing the missing long-range routing — removably, not as repair. But a bundled exploratory sub-claim — that the cap-unrescued residual is the pure severe-wiring tail — is refuted. Because the cap is axis-appropriate for wiring, it preferentially rescues the more wiring-dominated non-responders (cap-rescued mean wiring share 0.45 > residual 0.40), so the residual is actually less wiring-dominated. The residual is instead dominated by patients whose unresolved deficit is gain/threshold beyond the stimulant’s dose cap and/or high intrinsic stiffness dispersion — which neither the dose-capped stimulant nor the wiring-targeted cap addresses — plus a small high-wiring group that a fixed window amplitude over-synced.

Two actionable points fall out, and both feed back into the operating principle. First, the two operators are cleanly axis-specific — cap to wiring, stimulant to gain — and what is left over is a dose-cap / stiffness limit, not a wiring tail. Second, the few over-syncs, together with the non-zero adverse fraction even at the window, argue for a cap amplitude matched to the individual wiring deficit (closed-loop, titrated), not held fixed — exactly the matched-amplitude requirement of §21. Refutations are findings; this one sharpens the verdict rather than denting it. Every fraction and coupling value here is an in-silico coupling state, not a clinical response rate or a dose; efficacy = 0 throughout. This is a mechanism-level result about the fault structure of autism as represented in the VP framework — not medical advice, a diagnosis, a treatment protocol, or a cure.