Structure: the Geometry of Quality Space
A calibration fixes constants; any theory can be fitted to a single quality. A structure — the pattern of relations among many qualities, or among the components of — can be fitted only if it has the shape the theory committed to before the data were seen. UHM commits in advance to complex projective geometry for quality, seven axes and seven Fano lines for , a window and two bands for a living state. Each commitment forbids something measurable. This page lists the commitments with their statuses, the methods that test them, and the results that would count for and against. It reports no data.
What a structural test can and cannot show
- Geometry and the operational gate. Rays of an orthonormal eigenbasis are orthogonal, so their pairwise Fubini–Study distances alone do not reproduce a general similarity geometry. A chosen quality-ray map requires its own calibration. Keeping eigenrays fixed while changing eigenvalues can change purity and canonical ; phase Gap likewise does not determine . Being in does not by itself meet the integration/differentiation conditions. Cross-person or model alignment is evidence about the declared structure, not a consciousness certificate.
- Phenomenal bridge [I/H]. A preregistered structural match can support a specified map between experience reports and states; it does not alone prove identity. Revised T-214 supplies no universal no-go against an internal bridge: an arbitrary internal map is not a weakly point-surjective Lawvere evaluator. An empirical mismatch tests the chosen bridge and reconstruction assumptions, not that withdrawn theorem.
- It needs calibration for its metric form. Without the map from perceived dissimilarity to (calibration, K3) every metric test below is ordinal.
The invariants
Each row gives what UHM asserts, the status of the assertion, and where it is proved or stated. "Math" is the status of the statement about ; "reading" is the status of its reading as a statement about experience.
| # | Invariant | Math | Reading | Source |
|---|---|---|---|---|
| S1 | Choose a quality-ray map into with FS distance; on this separated metric space enriched Yoneda is isometric and isomorphic points coincide | [D] for the chosen space; [T] for the conditional metric results | Identification with experience [I/H] | enriched Yoneda |
| S2 | For a nondegenerate native spectrum, six eigenvalues and a 42-dimensional unitary orbit describe distinct data; degeneracies reduce orbit dimension | [T] (T-300) | Reading these as intensity and quality is [I/H] | qualia structure |
| S3 | The chosen native representation has seven axes, 21 complex pairs and 48 real coordinates; its traceless operator representation splits as | [T] for the representation (T-301); no phenomenal-completeness theorem | Names of channels [I]; extensions and histories require additional data | taxonomy, decoder |
| S4 | Supported cycle phases are diagonal-rephasing invariants; on full support the seven Fano triangles span seven of 15 independent cycles | [T] (T-301) | Reading any of these as quality is [I/H] | decoder |
| S5 | Seven line readouts group the 21 edges; the isotropic Fano channel scales every off-diagonal pair equally and preserves supported phases | [T] for incidence and this channel | Preferential empirical block transparency needs a separate hypothesis | Fano channel |
| S6 | Intra-line Gap below inter-line Gap; testable as triple holonomy on 7 lines against 28 non-lines | [H] (F-Gap-1) | — | F-Gap-1 |
| S7 | Under , , a line product is ; a single edge flip is detected but not located within that line | [T] under the vertex-sign model (T-306) | Arbitrary PSD coherences need not obey this sign model | what a line checks |
| S8 | Purity/reflection cuts are equivalent to ; and differentiation need separate checks | [T under chosen cuts]; full four-way independence withdrawn | is operational [D], phenomenal reading [I/H] | revised window |
| S9 | The chosen native purity/reflection/integration cuts imply , ; the bands alone do not establish L2 | [T] for these necessary inequalities (T-321, T-323) | Physical or phenomenal interpretation requires the declared bridge | F-Band |
| S10 | For constant and the specified Fano generator , moduli scale as and supported cycle angles stay fixed; other generators need their own analysis | [T] for this dynamics and support | Fading of experience remains [H] under a calibrated readout | how qualia die |
| S11 | At fixed populations/moduli, phase affects admissibility through PSD; selected equal-twist families reported a line angle near rad at | Family-specific report; no universal phase/integration opposition follows (T-307) | Interpretation [I/H] | scope and counterexamples |
| S12 | Emotional space is 30-dimensional; valence is | [T] (T-147); valence [C] (C.1) | names of emotions [I] | emotional taxonomy |
| S13 | Number of independent slow components in fMRI lies in | [H] (F-ISF) | — | F-ISF |
Dimension of quality space. Declare the register and normalized readout before choosing . A native E-axis has rank one; a seven-dimensional conditional clock block is one optional construction in a specified model, not a factor recovered uniquely from . Similarity data may constrain a chosen ray-model dimension, but cannot by themselves establish that the fitted rays are eigenvectors or an ontological experiential register (realization contract).
What the geometry forbids
A commitment is worth as much as what it excludes. From the invariants above, UHM forbids:
- Metric violations. For a calibrated FS model, require for all ordered triples, with uncertainty and tolerance fixed by the protocol. The historical -radian tolerance is a design choice, not a theorem. Requiring the absolute triangle slack to be small would wrongly force near-collinearity. Distances lie in .
- Unrealisable matrices. A finite dissimilarity matrix is realised by rays of if and only if, for some phases , the Hermitian matrix , , is positive semidefinite of rank at most (item 7 of the enriched Yoneda theorem [T]). In particular at most qualities can be pairwise at the maximal distance .
- Full-space covering radius. For and , a -cover of all needs at least probes. Thus 93 probes cannot have global covering radius below rad on . This is not a universal lower bound on statistical estimation precision or on resolution of a restricted stimulus family; the finite-probe theorem gives an error bound for its stated covering design.
- A conscious state outside the bands. A system that independent evidence in calls conscious, whose reconstructed lies outside the window or outside F-Band, falsifies the reconstruction; a state meeting all four criteria with or outside its band falsifies the algebra (F-Band; zero counterexamples over states per band so far).
- Drift while fading [H]. The Fano-only law fixes supported cycle angles while reducing their carriers. Predicting stable psychological qualities additionally requires a declared register/readout and its dynamics. It is that compound empirical hypothesis, not a consequence of coherence damping alone, which a drift test can challenge.
Methods
Similarity geometry: MDS and representational similarity analysis
Collect full pairwise dissimilarity matrices within one modality. Predeclare the stimulus-to-ray map and normalized representatives, then compare calibrated dissimilarities with their FS distances. Arbitrarily selected eigenvectors at degeneracy are not observations; the rays of an orthonormal eigenbasis are mutually orthogonal and cannot alone encode a nontrivial similarity matrix. If the readout yields mixed register states, predeclare a full-state metric such as Bures instead of silently treating them as rays. The calibration protocol supplies measurement error, cross-validation and chosen thresholds; its numerical tolerances are not consequences of FS geometry.
The Tsuchiya paradigm: enriched categories and unsupervised alignment
Tsuchiya, Phillips and Saigo (2022) model a declared quality space using dissimilarities; Gromov–Wasserstein methods compare specified metric-measure structures without shared labels. UHM can test the additional FS-ray hypothesis against alternatives. Enriched isomorphic points coincide within one separated metric space; an isometry between two spaces is not pointwise identity or proof of shared experience. Under the appropriate metric-measure hypotheses, zero Gromov–Wasserstein distance identifies structures up to measure-preserving isometry of their supports. Sampling, calibration and the phenomenal bridge remain separate (conditional metric result; research precedent).
Realisability and the rank of quality space — proposed here [Pr]
For a calibrated matrix, search over phases for the smallest rank of a positive semidefinite realising it within measurement error. Two predictions follow [H]: (a) exists — the matrix is realisable at all; (b) within one modality stabilises as stimuli are added, rather than growing with the stimulus count. A competing geometry (real Euclidean, hyperbolic) is compared by cross-validated fit at equal parameter count; a systematically better fit of a non-projective geometry counts against S1.
Topological analysis
Persistent homology of the similarity structure reports the topology of the sampled set of qualities — the hue circle, for instance, is one loop. It does not see the ambient space: a circle embeds in as well as in the plane. Topology therefore tests S1 only through what the ambient geometry constrains — bounded diameter, the maximal-distance count, realisability — and is used here to choose stimulus sets that cover a manifold without holes in the probe cover (item 3 above).
Structure of itself
- Seven axes and the Fano blocks in self-report: confirmatory factor analysis of the 28-item audit, testing whether the 7-triad block structure fits better than random tripartitions (Console V0-VAL, a self-report analogue of F-Gap-2).
- Seven axes in neural data: the number of independent slow components in fMRI (F-ISF [H]), and the seven-component stress tensor (experiment IV.1).
- Fano holonomy: compare supported on seven lines with the 28 non-line triangles using independently calibrated signed phases and declared uncertainty. The historical 180-state construction reported rad; this descriptive result is not physical validation. The Fano channel preserves every supported cycle angle, including non-line cycles, so that channel alone cannot create a line/non-line phase separation. Such separation is a separate hypothesis about the source and full dynamics (current empirical status).
- Window and bands: , , , , across sleep–wake and anaesthesia with a frozen (calibration, K1–K2). On the uniform diagonal the window is exactly , , (Step 5 [T]) — three readouts that must move together.
- Fading [Pr]: collect repeated similarity judgements on a fixed stimulus set, with predeclared calibration and an independently justified dynamical/readout model. Native coherence damping alone predicts neither register eigenvalues nor a psychological similarity order. Stable similarity under decreasing intensity tests the additional bridge; drift challenges that empirical model on its admitted domain, not the conditional phase-preservation identity.
Confirmation and refutation
| Invariant | Would confirm | Would refute |
|---|---|---|
| S1 geometry | Calibrated matrices realisable at a stable ; projective geometry fits better than Euclidean at equal parameters; cross-subject alignment near-isometric | Metric violations beyond tolerance; more pairwise-maximal qualities than ; unrealisable matrices after calibration |
| S2 intensity/quality | Isospectral states differ in quality, not intensity; context changes quality at fixed | Identical full invariants, distinguishable experience (refutation criterion) |
| S3 seven axes | 7-triad structure beats random tripartitions; | Block structure no better than random; systematically outside |
| S4–S6 Fano | A preregistered source/dynamics hypothesis predicts a reproducible line/non-line difference | Failure of that empirical hypothesis; not cycle invariance, which predicts no separation from Fano damping alone |
| S8–S9 window and bands | Wake inside, N3 below, both exits as predicted; conscious states inside F-Band | Any of the falsifiers of calibration; a conscious state outside F-Band |
| S10 fading | Under the declared dynamical/readout bridge, intensity falls while similarity order holds | Drift contradicts that bridge on its admitted domain; not phase preservation under the specified Fano-only law |
A single refutation acts at the status of the claim it hits: against a [T] row it means an error in the proof or in the bridge to data (the experimental protocol's structural level), against an [H] row it removes the hypothesis, against an [I] reading it removes the support that reading had.
Next: Engineering → · Back: Calibration