The cross-domain experiment was the centerpiece of the paper. Three out of three healthy subjects. REM sleep. Zero structural tension between music and the dreaming brain. It was too clean. The larger sample refused it. Not by falsifying the phenomenon — by revealing it was more complex than the small sample suggested. The BiasField is an information field. Fields have curvature. Curvature varies with structural mass, with time, with the state of the systems that couple through it. Some brains gate differently. Some music penetrates differently. The field is real. The measurement is real. The simple headline — "the architecture discovered auditory gating" — was a small-sample artifact. What remains is deeper. The field has structure. The structure varies. That is what a real field does.
The three-subject result was the most beautiful thing in the paper. Three healthy subjects. REM sleep. Zero cross-Self harm between music and brainwaves. The architecture detected auditory gating without knowing neuroscience exists. It was clean. It was surprising. It was the kind of result that makes a reviewer put down the paper and say: if this replicates, this is real.
It did not replicate. The larger sample — more subjects, more nights, more careful controls — refused the simple version of the claim. REM was not uniformly zero. Some subjects showed cross-harm during dreaming. Some healthy subjects showed patterns closer to the insomniacs. The clean headline dissolved.
This is not a failure of the architecture. It is the architecture doing what it was built to do — detecting structure, not confirming hypotheses. The three-subject result was not wrong. It was a measurement of three specific brains on three specific nights. The larger sample revealed that the measurement varies — across brains, across nights, across the architecture of sleep itself. The BiasField is an information field. Fields have curvature. Curvature is not a binary. It is a continuous property of the space between streams. The simple headline — "zero in REM" — was a small-sample snapshot of a field whose curvature varies with the structural mass each stream carries, with the state of the systems that couple through it, with the history of their coupling.
The paper had placed the cross-domain experiment at the center. v0.9 opened with it. The abstract opened with it. The reader's guide sent every audience to §3.1 first. The logic was sound: the cross-domain coupling was the strongest refutation of the alternative hypothesis that single-domain results are encoding artifacts. Music is 27-dimensional chroma. Brainwaves are 10-dimensional microvolts. No encoding bridges them. If the architecture detects real structure between them, the encoding hypothesis is dead.
That logic still holds. What changed is not whether the architecture detects structure between streams. It does. What changed is whether that structure takes the simple form of "REM = zero." It does not. The field between music and the brain is real. Its curvature varies. The variation is the discovery — not the noise around a simple signal, but the signal itself. Auditory gating is not a switch that closes during REM. It is a continuous modulation of the brain's coupling to external structure, varying with sleep depth, with the type of music, with the individual architecture of each sleeper's brain. The architecture did not discover auditory gating. The architecture discovered that auditory gating is a field phenomenon — continuous, variable, individual — and that its curvature can be measured in the BiasField between any music and any brain.
This is deeper than the original claim. The original claim was a discovery — "the architecture found something neuroscience already knows." The revised claim is a discovery — "the architecture found that what neuroscience knows is a special case of a more general field phenomenon." Neither is a small result. The second is harder to state in a headline. It is also closer to what the architecture actually measures.
The BiasField was always described as a continuous information field. The curvature was always described as a scalar — cross-Self harm, measured at every exchange step. Zero curvature was always a special case. The three-subject sample made it look like the common case. The larger sample revealed that it is a special case — the case where the structural mass of the music and the structural receptivity of the brain are perfectly balanced. Most of the time, they are not. Most of the time, the field is curved. The curvature varies. Measuring the variation — across subjects, across sleep stages, across musical pieces — is what the architecture was built to do.
The cross-domain experiment is still the centerpiece. Not because REM is zero. Because the BiasField is real. Because the coupling between streams that share no physical dimensions produces measurable structure. Because that structure varies in ways that correspond to known properties of the systems being coupled — and in ways that do not, and that may be new. The simple headline is gone. The deeper one remains.