Skip to main content

Unique Predictions of CC

"A theory that cannot be refuted by any conceivable event is non-scientific. Irrefutability is not a virtue of a theory (as people often think) but a vice." — Karl Popper, "Conjectures and Refutations" (1963)

Who this chapter is for

23 predictions of CC — 22 of them unique to CC, 21 unique and numerical — with verification protocols and falsification criteria. The reader will learn how CC predictions differ from IIT, FEP, and GWT.

In the previous chapter we computed the stability radius, traced the death spiral, and built the recovery protocol. We saw that CC generates specific numbers — rstabr_{\mathrm{stab}}, κbootstrap=1/7\kappa_{\text{bootstrap}} = 1/7, thresholds for each channel — rather than vague "tendencies". Now we collect all numerical consequences of CC in one place and for each specify: how to verify and what would refute it.

Science differs from philosophy not in the depth of its questions but in the willingness to put its answers to the test of experiment. A philosophical system can be beautiful, internally consistent, and entirely useless — if it generates no predictions that can be verified and, in principle, refuted. Popper called this the demarcation criterion: the boundary between science and non-science runs not through method or subject matter, but through falsifiability.

This principle is especially acute in the sciences of consciousness. Most existing theories — IIT, FEP, GWT, panpsychism — either generate no unique numerical predictions, or formulate them so vaguely that no experiment can unambiguously refute them. Coherence Cybernetics deliberately takes a different path. Every theorem of the formalism gives rise to a specific, numerical, experimentally testable consequence — and for each such consequence it is specified which experimental result falsifies the theory.

This chapter collects 23 predictions of CC. They are grouped by theme: from fundamental (connection between consciousness and viability) through architectural (minimal dimensionality, thresholds) to empirical (neural correlates, critical exponents). For each prediction we explain:

  1. Intuition — why this prediction naturally follows from the formalism.
  2. Formal statement — the precise mathematical notation.
  3. Uniqueness — what exactly distinguishes this prediction from anything IIT, FEP, and GWT can say.
  4. Verification — a concrete experimental protocol.
  5. Interdisciplinary consequences — what this prediction means for a physicist, biologist, psychologist, and engineer.

If even one of these predictions is cleanly refuted — Coherence Cybernetics will require fundamental revision. This very readiness for refutation is what makes CC a science.

On notation

In this document:


I. Fundamental Predictions: Consciousness and Viability​

The first group of predictions concerns the deepest idea of CC — the indissoluble connection between interiority and stability. In most theories consciousness is either an epiphenomenon (does not affect dynamics) or a separate postulate (introduced externally). CC asserts something radically different: a system with an E-dependent feedback law can have a different recovery rate, conditional on the target, gate, dissipation, and resources.

Prediction 1: Conditional E-coherence balance​

Prediction [H]

Purity above 2/72/7, a nonzero dissipator, and sustained dynamics do not force CohE>1/7\mathrm{Coh}_E>1/7. Theorem 8.1 supplies a stationary counterexample and a conditional purity-flux balance. A positive E-floor requires a specified restricted model class, a bound on external support, positive feedback dependence, and a positive required flux. Identifying the E-sector with experience is a further [P/I] bridge.

Test the restricted balance after fixing the admissible inputs and rates. Keep the remaining target, gate, dissipation, preparation, and resources fixed during an E-intervention, or explicitly model their changes. An externally supported viable system with low E-coherence does not refute the density-matrix formalism.

Prediction 2: A selected E-dependent rate model​

Prediction [D/H]

The model chooses κ=κb+κ0CohE\kappa=\kappa_b+\kappa_0\mathrm{Coh}_E, with the other quantities fixed and κ0≥0\kappa_0\ge0. Then ∂κ/∂CohE=κ0\partial\kappa/\partial\mathrm{Coh}_E=\kappa_0 is an exact derivative [T] of that convention [D]. It does not imply P˙∝CohE\dot P\propto\mathrm{Coh}_E: the regeneration flux is 2κg(P)(Tr⁡(ρB(ρ))−P)2\kappa g(P)(\operatorname{Tr}(\rho B(\rho))-P) and can be negative. Endogenous variations also change κ0\kappa_0, the target, gate, and state.

A correlation with biological recovery is [H], requiring an identifiable observation model and a defined physical recovery outcome. It does not establish a clinical benefit of meditation or therapy. A valid study preregisters the intervention, observation likelihood, nuisance variables, clustered sampling unit, effect size, and held-out test. The bounded rate construction in septicity is conditional kinetic mathematics, not a categorical uniqueness theorem.

II. Architectural Predictions: Structure and Dimensionality​

The second group of predictions concerns the architecture of conscious systems — what is the minimal "design" necessary for the emergence of experience, learning, and social interaction. These predictions make CC unique among theories of consciousness: instead of vague statements about "complexity" and "integration" it names specific numbers.

Prediction 3: A seven-component stress panel​

Prediction [D/H]

The seven stress scores define a panel in a specified frame. The equivalence between crossing that panel, passing the four Cap2 gates and attaining biological health would require separate observation laws and evidence. Definitions

For window monitoring use the exact four margins and uncertainty-set verdict. For empirical stress classification, preregister the coding rules and test held-out stressors against competing models. The ability to assign a label to every item does not prove independent dimensionality or exhaustive causal coverage.

Prediction 4: Pre-linguistic cognition is complete​

Intuition. It is often assumed that consciousness and language are inseparable — that "to think" means "to think in words". CC shows that this is an anthropocentric fallacy. The cognitive hierarchy K1–K5 arranges five levels of cognition, of which language (K5) is only the top — not the foundation. Animals without language function fully at levels K1–K4.

Prediction [I]
∃ Cognition(H) with Language(H)=∅\exists \, \mathrm{Cognition}(\mathbb{H}) \text{ with } \mathrm{Language}(\mathbb{H}) = \varnothing

Full cognition (levels K1–K4) is possible without language (K5).

Status: [I] — an interpretation following from the definitions of cognitive levels K1–K5.

See: Cognitive hierarchy

Uniqueness of the prediction. GWT links consciousness to global availability of information, often associated with linguistic representations. Higher-Order Theories (HOT) tie consciousness to metacognitive reports, implicitly presupposing language. CC explicitly separates cognition and language as orthogonal axes.

Cognitive function hierarchy:

Verifiability: Animals without language demonstrate levels K1–K4:

  • Corvids: planning (K4), tool making
  • Primates: categorisation (K3), social learning
  • All vertebrates: emotional responses (K2)
  • All systems with ρE≠0\rho_E \neq 0: interiority (K1/L0)

Experimental verification:

  1. Compare behavioural correlates of K1–K4 in aphasic patients (loss of language with preserved intellect) and healthy controls.
  2. Prediction: K1–K4 scores in aphasic patients are preserved at >80% of normal.
  3. Falsification: if aphasia systematically destroys K3 or K4 — the link "cognition → language" is stronger than CC predicts.

Interdisciplinary consequences:

  • Ethology: provides grounds for full cognitive study of non-linguistic species.
  • AI ethics: systems without a language module may possess cognitive levels K1–K4, which has ethical implications.

Prediction 10: Dimension and learning​

Prediction [H]

A seven-axis learning architecture is a testable design. No general theorem here prohibits autonomous learning at N<7N<7. Such a bound requires a specified task family, operational independence criterion, representation class, and resource restrictions. Functional-role counts do not prove a Hilbert-space dimension bound.

The rigorous bounds concern a perfect single-error binary code or a faithful representation of a selected internal algebra; see minimality. Compare architectures under matched training, observation capacity, resources, and tasks. Failure of one five-dimensional implementation is not a universal impossibility result.

Prediction 11: Dimension and social learning​

Prediction [H]

The proposed role assignment 3ToM+3ISL+1U=73_{\rm ToM}+3_{\rm ISL}+1_U=7 is a model architecture [H]. To establish a lower bound one would need an independence or discrimination result for a specified task.

A lower bound must identify operationally independent observables or an information task requiring a faithful representation. Compare the proposed seven-axis architecture to smaller alternatives on that task.

III. Thresholds and Robustness​

The third group of predictions concerns numerical thresholds — specific parameter values at which qualitative transitions occur. Numerical predictions are precisely what distinguishes a scientific theory from philosophical speculation: they are testable, and they are risky.

Prediction 5: Collective consciousness​

Represent the members of a proposed collective by a joint state. The mathematical test asks whether it factors into its marginal states. Identifying the collective as a subject requires an additional observation law.

Prediction. Necessary correlation [T], subject interpretation [H]

Necessary condition [T]. A collective can differ from a set of separate subjects only if its members' joint state is correlated: a product state is fixed by its marginals, and the mutual information

I(H1:H2)=S(ρ1)+S(ρ2)−S(ρ12)=D(ρ12 ∥ ρ1⊗ρ2)I(\mathbb{H}_1 : \mathbb{H}_2) = S(\rho_1) + S(\rho_2) - S(\rho_{12}) = D(\rho_{12} \,\|\, \rho_1 \otimes \rho_2)

vanishes exactly on products, so the condition is I>0I > 0; for nn members the analogue is a positive total correlation ∑iS(ρi)−S(ρ1…n)>0\sum_i S(\rho_i) - S(\rho_{1 \ldots n}) > 0. The condition is necessary, not sufficient: coupled thermostats meet it too.

Interaction. A coupling that commutes with the product anchor leaves a stationary product state. An interaction is therefore insufficient to ensure I>0I>0. CC-7

Sufficiency [H]. A correlated group is a subject when its joint state, aggregated to D(C7)\mathcal{D}(\mathbb{C}^7), passes the four-condition window. This is a hypothesis, not a theorem. Theorem 9.5 fixes the aggregation by three natural requirements — the only linear, permutation-invariant map D(C7n)→D(C7)\mathcal{D}(\mathbb{C}^{7^n}) \to \mathcal{D}(\mathbb{C}^7) that returns a member's state on an uncoupled group is the mean of the members' marginals — and that map depends on the marginals alone, so it cannot see the correlation that the necessary condition requires. Under it a weakly coupled group of identical members aggregates to within O(g)O(g) of one member (T-72 [T]), and at strong coupling the aggregate can fall to I/7I/7 (Theorem 9.6): the canonical aggregation cannot certify a collective level above the members' at any coupling. An aggregation that sees correlations must give up linearity, permutation invariance or consistency on uncoupled groups, and the corpus fixes none; nor does it have an exclusion rule deciding whether members and group can be subjects at once (boundary problem). Research programme [Pr]: a correlation-sensitive aggregation — necessarily non-linear, non-symmetric or inconsistent on uncoupled groups — or an exclusion rule, that turns the hypothesis into a criterion.

See: Theorem 9.1, Theorem 9.3

Experimental verification:

  1. Measure the correlation between members' states — the mutual information, or the total correlation for n>2n > 2, of the reconstructed joint Γ\Gamma — for groups with varying degrees of coordination (jazz quartet vs. random musicians).
  2. Apply hyperscanning (simultaneous EEG of several subjects).
  3. Prediction: the correlation is positive for a coordinated group and absent for an unconnected one. This tests only the necessary condition; a positive result does not show that the group is a subject.

Verifiability: Measurement of the correlation between members for groups with varying degrees of integration:

  • Families
  • Teams
  • Organisations
  • Ecosystems

Criterion: I>0I > 0 (total correlation for n>2n > 2) — a necessary condition [T]. Without correlation between members there is no collective subject in any sense the theory can state.

Interdisciplinary consequences:

  • Sociology: formalises a necessary condition for Durkheim's "collective consciousness" — correlation between members' states; whether a correlated collective is a subject remains a hypothesis [H].
  • Ecology: raises the question of the correlation structure of an ecosystem's joint state; whether a forest is "one organism" in the strict sense the theory does not decide.
  • Organisational science: gives a necessary, not a sufficient, condition for a team to be "more than the sum of its parts" — correlation of its members' states.

Prediction 6: Minimum coherence for viability​

Intuition. A system cannot be "slightly" alive. There is a hard threshold: if PP falls below 2/72/7 or CohE\mathrm{Coh}_E drops to 1/71/7 — the system loses viability. This is analogous to a phase transition: water at 0°C freezes not gradually but sharply. Likewise Pcrit=2/7P_{\text{crit}} = 2/7 is the "freezing temperature" of coherence.

Prediction
Viable(H)⇒P(Γ)>Pcrit=2/7\mathrm{Viable}(\mathbb{H}) \Rightarrow P(\Gamma) > P_{\text{crit}} = 2/7

Viability requires minimum purity, with the threshold derived, not fitted (theorem on critical purity, registry Level 1 row 5 [T]); a nontrivial attractor has P>1/7P > 1/7 (T-96 [T]).

L2 thresholds
ParameterValueStatusDefinition
PcritP_{\text{crit}}=2/7≈0.286= 2/7 \approx 0.286[T]Theorem on critical purity
RthR_{\text{th}}=1/3≈0.333= 1/3 \approx 0.333[T] (K=3K=3 from triadic decomposition)Theorem (Bayesian dominance)
Φth\Phi_{\text{th}}=1= 1 (exact)[T]Theorem T-129 (coherent dominance)

PcritP_{\text{crit}} — strictly proven [T]; RthR_{\text{th}} — theorem [T] (K=3K=3 from triadic decomposition); Φth\Phi_{\text{th}} — theorem [T] (T-129). See L2 Thresholds.

note
Why exactly 2/72/7, not 1/31/3 or 1/51/5?

The number 2/72/7 is not a fit to data or an arbitrary choice. It is derived strictly from two facts:

  1. N=7N = 7 — number of dimensions (follows from octonion algebra and G2G_2-minimality).
  2. Frobenius norm — the distinguishability criterion. A state Γ\Gamma is distinguishable from the completely mixed state I/7I/7 if and only if ∥Γ−I/7∥F>0\|\Gamma - I/7\|_F > 0, which is equivalent to P=Tr(Γ2)>1/7P = \mathrm{Tr}(\Gamma^2) > 1/7. But viability requires more: the system must not merely "differ from chaos" but have sufficient structure for regeneration. The theorem on critical purity proves that this threshold is exactly 2/7=2×(1/N)2/7 = 2 \times (1/N).

Intuition: 1/71/7 — "one voice out of seven". 2/72/7 — "two voices out of seven". For a system to be able to restore itself, it needs to "outweigh" chaos by at least two degrees of freedom. One voice is not enough; two is the minimum. Three (3/73/7) is already the upper boundary of the consciousness window.

Uniqueness of the prediction. IIT defines Φ\Phi as a measure of consciousness but sets no critical threshold. FEP defines viability through the Markov blanket but without a numerical threshold. CC is the only theory with computable threshold values derived from first principles.

Experimental verification:

  1. In subjects at the boundary of consciousness (anaesthesia, sleep) record EEG/HRV and TMS-EEG; reconstruct Γ^\widehat\Gamma by πbio\pi_{\mathrm{bio}} with parameters frozen on wakefulness and no viability penalty (SUB-1, SUB-2).
  2. Compute P^\widehat P and the verdict Cons(Γ^)\mathrm{Cons}(\widehat\Gamma); compute PCI (Perturbational Complexity Index) independently.
  3. Prediction: the consciousness/unconsciousness transition occurs where P^\widehat P crosses 2/72/7, and Cons(Γ^)\mathrm{Cons}(\widehat\Gamma) agrees with PCImax⁡>0.31\mathrm{PCI}_{\max} > 0.31 at Cohen's κ≥0.8\kappa \geq 0.8 (P8.4).

Verifiability: Measurement of purity in systems approaching loss of viability must show PP crossing 2/72/7 (E-coherence is tested separately, as the L2 differentiation condition).

Clinical consequences:

  • Coma states: CohE→1/7\mathrm{Coh}_E \to 1/7 (minimum), P→PcritP \to P_{\text{crit}}
  • Psychotic episodes: fragmentation of CohE\mathrm{Coh}_E
  • Meditative states: high CohE\mathrm{Coh}_E, high RR

Interdisciplinary consequences:

  • Anaesthesiology: predicts a numerical indicator of anaesthetic depth — not the BIS index (empirical) but PP (theoretically grounded).
  • Psychiatry: formalises the "norm threshold" — not as a statistical mean but as a phase transition PcritP_{\text{crit}}.

Prediction 7: Stability radius (T-104)​

For the selected purity boundary Pc=2/7P_c=2/7, the exact HS distance is rHS=P−1/7−Pc−1/7r_{HS}=\sqrt{P-1/7}-\sqrt{P_c-1/7} [T]. It grows with purity and is attained by a positive radial projection. Bures obeys rB≥rHS/2r_B\ge r_{HS}/\sqrt2; distance to a one-dominant-family candidate is an upper bound until global optimality is proved.

Geometric distance is not a permitted disturbance magnitude. Trajectory-margin transfer requires field, horizon and perturbation bounds; exponential recovery additionally requires dissipativity. Identification with physical consciousness robustness is a hypothesis [H]. See the stability proofs and conditions.

Prediction 15: Attractor inside the consciousness window, below its upper edge​

Intuition. Where does consciousness "settle"? Not at maximum purity (P=1P = 1), not at the threshold (P=2/7P = 2/7), and — contrary to the earlier form of this prediction — not at the upper edge 3/73/7 of the window either. The living attractor sits strictly inside the window, in its lower half: coherent enough to pass every threshold, with a margin below the edge.

Prediction [C at (MaxΦ)] — falsifiable

For a holon whose self-model is the collineation anchor φJ\varphi_J (selected by the principle (MaxΦ) [Pr], premises) at H=0H = 0 and κ>κc(α)\kappa > \kappa_c(\alpha), the living attractor is a hyperbolic sink with

P∗∈(Pc(α), P∞(α))⊂(2/7, 5/14],P∞(α)≤5/14<3/7,P^* \in \bigl(P_c(\alpha),\ P_\infty(\alpha)\bigr) \subset \bigl(2/7,\ 5/14\bigr], \qquad P_\infty(\alpha) \leq 5/14 < 3/7,

Φ∈(1,3/2]\Phi \in (1, 3/2], R≥2/5R \geq 2/5, every diagonal entry 1/71/7; Pc=0.318,0.308,0.301P_c = 0.318, 0.308, 0.301 and P∞=5/14,0.334,0.317P_\infty = 5/14, 0.334, 0.317 at α=0,1/2,1\alpha = 0, 1/2, 1 (living attractor in the window, T-124c(4) [T]); it persists for ∥H∥<h0\|H\| < h_0.

Status: [T] as mathematics for the φJ\varphi_J dynamics; [C at (MaxΦ)] as a prediction about a real holon, whose anchor the axioms leave open. For an embodied holon with backbone rate μ>LR\mu > L_{\mathcal{R}} the attractor is unique and globally attracting (T-124c(3)), but its purity depends on the backbone target and has no universal value.

Restated 2026-09-26: the prediction read "P(ρcoupled∗)→3/7P(\rho^*_{\mathrm{coupled}}) \to 3/7 for an embodied holon with backbone injection [C]", with the numerical value P=0.4286P = 0.4286. No theorem gives 3/73/7: every living stationary state of the φJ\varphi_J family has P≤5/14P \leq 5/14 (none has P≥3/7P \geq 3/7), and the embodied attractor inherits its purity from the backbone target, so 0.42860.4286 was a property of one chosen target, not a law.

Status in other theories: Absent.

See: Consciousness window, living attractor

Verifiability: For an isolated agent with the φJ\varphi_J anchor and κ>κc(α)\kappa > \kappa_c(\alpha), measure the attractor P∗P^*. Numerical falsification criterion: P∗≥3/7P^* \geq 3/7, or P∗P^* outside (Pc(α),P∞(α))(P_c(\alpha), P_\infty(\alpha)) beyond the measurement error, at n>100n > 100 independent measurements (p<0.01p < 0.01) → falsification of the φJ\varphi_J dynamics or of its implementation.

Interdisciplinary consequences:

  • Neuroscience: predicts that a healthy brain at rest sits inside the window with a margin below its upper edge, not at 3/73/7.
  • Psychology: the "optimal state" (flow) is an interior point of the window, not its boundary.

IV. Learning and Cognitive Architecture​

The fourth group of predictions concerns learning bounds — the minimum resources needed for cognitive activity. These predictions are especially valuable for AI engineering, as they establish theoretically grounded lower bounds.

Prediction 8: Perceptual information capacity (T-107)​

Intuition. How much information can a system extract from a single observation? Intuitively the answer depends on the "size" of internal space: the more dimensions, the more one can "absorb" in a single glance. CC gives a precise upper bound: log⁡27≈2.81\log_2 7 \approx 2.81 bits. This is not a postulate but a consequence of the dimensionality of C7\mathbb{C}^7.

Prediction [T] — measurable in AI systems
CEnc≤log⁡27≈2.81 bits/observationC_{\mathrm{Enc}} \leq \log_2 7 \approx 2.81 \text{ bits/observation}

Maximum information extractable per single observation is bounded by the dimensionality of internal space (N=7N = 7).

Uniqueness of the prediction. FEP speaks of minimising free energy but names no bottleneck in throughput. IIT measures information but does not bound the rate of its extraction. CC is the only theory deriving a specific numerical bound on perceptual throughput.

Experimental verification:

  1. For a CC agent (SYNARC) measure mutual information I(obs;δΓ)I(\text{obs}; \delta\Gamma) between the observation and the Γ\Gamma update.
  2. Prediction: I≤2.81I \leq 2.81 bits.
  3. Falsification: I>log⁡27I > \log_2 7 systematically — CC is falsified (or the system is not 7-dimensional).

Verifiability: For AI systems implementing the CC architecture, measure the mutual information between the observation and the Γ\Gamma update. If I(obs;δΓ)>log⁡27I(\text{obs}; \delta\Gamma) > \log_2 7 — CC is falsified (or the system uses a non-7-dimensional formalism).

Connection with bounded rationality: The bound ≤2.81\leq 2.81 bits is a derived bound, not postulated. Predicts a specific bottleneck for perceptual systems based on CC.

See: Sensorimotor theory

Interdisciplinary consequences:

  • Cognitive psychology: echoes Miller's results (7±2) — but in CC this is not an empirical finding but a theoretical consequence.
  • Neuroscience: predicts an upper bound on information extractable in one processing cycle (one "tick" of neural processing).

Prediction 9: Optimal learning rate bound (T-112)​

Intuition. How fast can a system learn? CC shows that the learning rate is bounded by the maximum of three independent barriers: informational (quantum Chernoff bound — how much information is extracted per observation), dynamical (Fano contraction rate — how quickly Γ\Gamma updates), and stabilisation (SNR — how noisy the signal is). The slowest of the three determines the overall rate.

Prediction [T] — falsifiable
nopt=max⁡ ⁣(ln⁡(1/(2δ))ξQCB,  1αδτln⁡ddiscε,  1SNR2)n_{\mathrm{opt}} = \max\!\left(\frac{\ln(1/(2\delta))}{\xi_{\mathrm{QCB}}},\; \frac{1}{\alpha\delta\tau}\ln\frac{d_{\mathrm{disc}}}{\varepsilon},\; \frac{1}{\mathrm{SNR}^2}\right)

The minimum number of observations for a learning task is determined by the maximum of three independent bounds: informational (T-109), dynamical (T-110), and stabilisation (T-111).

Uniqueness of the prediction. PAC-learning and VC-theory give learning bounds but without connection to the system's physical dynamics. FEP describes learning through free energy minimisation but gives no lower bounds on the number of observations. CC unites information theory, system dynamics, and stability theory in a single formula.

Experimental verification:

  1. For a CC agent trained on binary discrimination, vary SNR and α\alpha.
  2. Measure nn to stable solution (>90% accuracy on 50 successive trials).
  3. Prediction: n≥noptn \geq n_{\mathrm{opt}} always; at optimal settings n≈noptn \approx n_{\mathrm{opt}}.
  4. Falsification: n<ninfon < n_{\mathrm{info}} systematically — the quantum observation model is incorrect.

Verifiability: For an AI system implementing the CC architecture, measure the number of observations nn to stable solution of a binary discrimination task. If n<ninfon < n_{\mathrm{info}} (information bound) — the quantum observation model is falsified.

Prediction for binary discrimination: ntotal≈20–80n_{\mathrm{total}} \approx 20\text{--}80 observations at typical parameters (including genesis).

See: Learning bounds


V. Depth of Consciousness and Genesis​

The fifth group of predictions concerns the vertical structure of consciousness — its depth, the ultimate possibilities of recursive self-reference, and the process of emergence from "nothing" (tabula rasa). Depth requires explicit observation and certification protocols.

Prediction 12: Testing declared depth diagnostics (T-142)​

Iterating a specified Fano channel gives an amplitude-survival law. Identifying that iteration index with measured cognitive depth requires an operational readout and independent data.

Exact channel prediction [T]. For the bare Fano dephaser D2/3D_{2/3} and any initially nonzero off-diagonal entry,

∣(D2/3 nΓ)ij∣∣γij∣=3−n.\frac{|(D_{2/3}^{\,n}\Gamma)_{ij}|}{|\gamma_{ij}|}=3^{-n}.

This contracts off-diagonal amplitudes, not total purity by a factor of three. With a declared detector floor ε>0\varepsilon>0, detectable iterations satisfy n≤⌊log⁡(1/ε)/log⁡3⌋n\le\lfloor\log(1/\varepsilon)/\log3\rfloor; the bound depends on the detector.

Chosen-score arithmetic [D/T]. If the instrument instead defines

sn−1:=P2/7 3−(n−1),sn−1>1n+1,s_{n-1}:=\frac{P}{2/7}\,3^{-(n-1)},\qquad s_{n-1}>\frac1{n+1},

the thresholds are pn=(2/7)3n−1/(n+1)p_n=(2/7)3^{n-1}/(n+1), namely 1/7,2/7,9/14,54/351/7,2/7,9/14,54/35 for indices one through four. Since P≤1P\le1, this score's maximum passing index is three. The prefactor may exceed one, so ss is not a probability or fidelity. Its third test requires P>9/14P>9/14, contradicting the nested L2 bound P≤3/7P\le3/7; score index three therefore cannot define L3.

Protocol [Pr]. Declare the map, frame, initial amplitudes, detector floor, measured ratio and uncertainty before testing the channel law. Independently test a proposed cognitive depth with nonconstant metamodel predictions on held-out targets and inter-level compatibility, following the capability certificates. A stable fourth-order certificate is allowed by the revised definitions. Samples of matrices satisfying a score computed by the same formula are implementation checks, not evidence of a biological depth ceiling.

Failure of the measured 3−n3^{-n} law rejects the specified channel model or its implementation. Failure of a calibrated score-to-task association rejects that empirical bridge [H], without refuting the arithmetic identity. Neither this score nor seven-dimensionality bounds human or AGI recursive awareness universally. See Depth Tower and T-142.

Prediction 13: Genesis time (T-148)​

Intuition. How does a system "come to life" — transition from tabula rasa (Γ=I/7\Gamma = I/7, pure chaos) to viability (P>2/7P > 2/7)? An isolated system cannot — theorem T-39a guarantees that without an external source of purity the attractor of L0\mathcal{L}_0 is I/7I/7. But an embodied system, coupled with the environment, receives "injections" of coherence through the backbone and reaches the threshold in a finite number of steps, computable by formula.

Prediction [T] — falsifiable
ngenesis≤⌈ln⁡Δln⁡(1/β)⌉,Δ=Penv−2/7Penv−1/7n_{\mathrm{genesis}} \leq \left\lceil \frac{\ln \Delta}{\ln(1/\beta)} \right\rceil, \quad \Delta = \frac{P_{\mathrm{env}} - 2/7}{P_{\mathrm{env}} - 1/7}

An embodied holon (H,π,B)(\mathbb{H}, \pi, B) with mixing parameter β∈(0,1)\beta \in (0,1) and environmental purity Penv>Pcrit=2/7P_{\mathrm{env}} > P_{\mathrm{crit}} = 2/7 raises purity from I/7I/7 (tabula rasa) above PcritP_{\mathrm{crit}} in a finite number of steps ngenesisn_{\mathrm{genesis}}.

Uniqueness of the prediction. No other theory of consciousness gives a formula for birth time. IIT is static — it does not describe genesis. FEP describes "self-organisation" but gives no upper bound on the number of steps.

Experimental verification:

  1. SYNARC agent starts at Γ=I/7\Gamma = I/7.
  2. Backbone injection with parameters β,Penv\beta, P_{\mathrm{env}}.
  3. Measure nn until P>2/7P > 2/7.
  4. Prediction: n≤ngenesisn \leq n_{\mathrm{genesis}}.
  5. Double falsification: (a) genesis does not occur within ngenesisn_{\mathrm{genesis}} steps → T-148 is incorrect; (b) genesis of an isolated holon (without backbone) → T-39a is incorrect.

Verifiability: For an AI system based on CC, starting from Γ=I/7\Gamma = I/7, measure the number of backbone injections to reach P>2/7P > 2/7. If genesis does not occur within ngenesisn_{\mathrm{genesis}} steps at given β\beta and PenvP_{\mathrm{env}} — formula T-148 is falsified. If genesis occurs for an isolated holon (without environmental coupling) — theorem T-39a is falsified.

Typical estimates: At β=0.3\beta = 0.3, Penv=0.35P_{\mathrm{env}} = 0.35: Δ≈0.35\Delta \approx 0.35, ngenesis≤⌈0.87/1.20⌉=1n_{\mathrm{genesis}} \leq \lceil 0.87 / 1.20 \rceil = 1. At β=0.9\beta = 0.9 (weak mixing): ngenesis≤⌈0.87/0.105⌉=9n_{\mathrm{genesis}} \leq \lceil 0.87 / 0.105 \rceil = 9.

Numerical verification (SYNARC mvp_int_2 G1-G3): ngenesis<50n_{\mathrm{genesis}} < 50 ticks from I/7I/7 to P>2/7P > 2/7 at β=0.3\beta = 0.3.

See: Substrate-independent closure

Interdisciplinary consequences:

  • Developmental biology: the genesis formula describes how an embryo transitions from "pure chaos" to an organised state — through the maternal environment as backbone.
  • Pedagogy: formalises the role of "environment" (teacher, culture) in awakening cognitive capacities in a child.

Prediction 14: Necessity of phase coherence for integration​

Intuition. To achieve true integration (Φ≥1\Phi \geq 1) it is not enough to simply "mix" information. The regenerative channel R\mathcal{R} must be phase-coherent with Hamiltonian dynamics — like dancers who must move in time with the music, not merely move. Without phase coherence, R\mathcal{R} and HH "fight" each other, and integration does not reach the threshold.

Prediction [T] — falsifiable

To achieve Φ≥Φth=1\Phi \geq \Phi_{\mathrm{th}} = 1 the replacement channel R\mathcal{R} must use co-rotating targets ρij∗(t)∝e−i(Ei−Ej)t\rho^*_{ij}(t) \propto e^{-i(E_i-E_j)t}, coherent with Hamiltonian dynamics.

Experimental consequence: In any implementation (biological or digital), where the regenerative channel is not phase-coherent with Hamiltonian evolution, the integration measure will be Φ<1\Phi < 1.

Status: [T] theoretically (competition of R\mathcal{R} and HH), confirmed numerically (SYNARC: Φ=0.83→1.15\Phi = 0.83 \to 1.15 upon enabling co-rotation).

Status in other theories: Absent.

See: Co-rotating targets

Uniqueness of the prediction. Neither IIT nor FEP contains the concept of phase coherence. GWT describes "global availability" but does not connect it to phase coherence. CC is the only theory where phase plays a role in information integration.

Experimental verification:

  1. SYNARC agent with fixed targets ρij∗=const\rho^*_{ij} = \mathrm{const}.
  2. Measure Φ\Phi.
  3. Switch to co-rotating targets ρij∗(t)∝e−i(Ei−Ej)t\rho^*_{ij}(t) \propto e^{-i(E_i-E_j)t}.
  4. Prediction: Φ\Phi increases and crosses threshold 1.
  5. Falsification: Φ≥1\Phi \geq 1 with fixed targets — observation O-1 is incorrect.

Verifiability: For a system implementing the CC architecture, measure Φ\Phi with fixed and co-rotating targets. If Φ≥1\Phi \geq 1 with fixed targets — observation O-1 is falsified.

Interdisciplinary consequences:

  • Neuroscience: predicts that phase synchronisation of neural ensembles (gamma connectivity) is necessary for consciousness — not merely correlated, but a condition for Φ≥1\Phi \geq 1.
  • AI engineering: architectures without phase coherence (standard transformers) cannot achieve true integration.

VI. Dynamic Predictions: Phase Transitions and Critical Phenomena​

The sixth group of predictions concerns the dynamics of transitions — how a system enters and exits the conscious state. Their normal forms and scaling laws require explicitly stated dynamical reductions and measurement bridges.

Prediction 16: Conditional passage and ignition laws​

The chosen gate boundary P=2/7P=2/7 is a purity cut. Establishing a bifurcation, a catastrophe normal form or an ignition time requires specifying the flow, reduction and readout. The normal-form calculation below addresses those conditions separately.

Conditional scalar result [T]. Suppose an actual smooth reduction yields y˙=ay+by2\dot y=ay+by^2 with fixed a,b>0a,b>0, initial 0<y0<yf0<y_0<y_f, and an explicitly calibrated relation between yy and observables. Its passage time is

T=1alog⁡yf(a+by0)y0(a+byf).T=\frac1a\log\frac{y_f(a+by_0)}{y_0(a+by_f)}.

For fixed a>0a>0, y0↓0y_0\downarrow0 gives logarithmic divergence. Only the separately tuned case a=0a=0, b>0b>0 gives T=b−1(y0−1−yf−1)T=b^{-1}(y_0^{-1}-y_f^{-1}). An inverse purity-margin law further requires y0y_0 proportional to that margin and the identification of bb with a measured rate.

Protocol [Pr/H]. Specify the flow, centre-manifold reduction, start/finish criteria and measurement bridge; estimate a,ba,b on training trajectories, compare logarithmic and inverse laws on held-out perturbations, and report uncertainty. For a proposed fold/cusp/swallowtail, separately test the required degeneracy and transverse control rank. A gate crossing or a delayed response alone does not establish such a normal form. Controlled neural measurements can test a calibrated dynamical bridge, but static PP cannot diagnose awakening, coma or clinical recovery.

Source: conditional normal forms and passage laws. Deviations reject the stated reduction/bridge at its status level; there is no universal ignition theorem to falsify.

Prediction 17: Conditional tricritical mean-field exponents​

This is a conditional prediction [C] of a specified scalar model plus its phenomenological identification, not a consequence of the seven-dimensional state space. Assume a smooth equilibrium-like reduction with an even potential, a separately tuned zero quartic coefficient, nonzero linear distance-to-criticality, and positive sextic stabilisation:

V(m;t,h)=t2m2+v6m6−hm,v>0.V(m;t,h)=\frac t2m^2+\frac v6m^6-hm,\qquad v>0.

The sextic germ is A5A_5; it is not the generic A4A_4 swallowtail. Z2\mathbb Z_2 symmetry alone permits a quartic term and does not select tricriticality.

Polynomial theorem [T]. Minimising this potential gives m4=−t/vm^4=-t/v for t<0,h=0t<0,h=0, h=vm5h=vm^5 at t=0t=0, susceptibility χ=1/t\chi=1/t on the disordered side, and Vmin⁡=−(−t)3/2/(3v)V_{\min}=-(-t)^{3/2}/(3\sqrt v). Under the standard thermodynamic meanings of these quantities,

β=14,δ=5,γ=1,α=12.\beta=\tfrac14,\quad\delta=5,\quad\gamma=1,\quad\alpha=\tfrac12.

The additional ν=1/2\nu=1/2 requires a specified spatial gradient term and a valid Gaussian/mean-field approximation; it is not an exponent of the scalar polynomial alone. The identity α+2β+γ=2\alpha+2\beta+\gamma=2 checks consistency of these exponents but proves no mapping to consciousness.

Protocol [Pr/H]. Before fitting, define the independent order parameter, control variable, actual critical point, field, symmetry test, quartic tuning, finite-size regime and observable-to-mm bridge. Fit the reduction on training data and compare held-out scaling against quartic and non-gradient alternatives. A fit of PCI alone does not validate five exponents, spatial dimensionality or equilibrium assumptions. Twenty-one coherence pairs are not a spatial dimension or a controlled thermodynamic limit.

Identifying t=0t=0 with P=2/7P=2/7, identifying mm with a neural/phenomenal observable, and controlling fluctuations are extra conditions [C/H]. Failure of the measured scaling rejects the specified reduction or bridge, not the proved minimisation identity.

Source: T-161 and its explicit hypotheses.

Prediction 18: Ward suppression 19/49​

Intuition. The cosmological constant is one of the greatest puzzles of physics. A naive estimate gives a value differing from the observed one by 120 orders of magnitude. CC proposes a suppression mechanism: Ward identities following from the G2G_2-symmetry of the 7-dimensional space suppress Gap fluctuations by a factor of 19/4919/49. This does not solve the cosmological constant problem, but contributes a specific, computable contribution.

Prediction [T] — falsifiable
Gap fluctuations suppressed by factor 1949≈0.39\text{Gap fluctuations suppressed by factor } \frac{19}{49} \approx 0.39

The total contribution of Gap fluctuations to the cosmological constant is suppressed through Ward identities by a factor of 19/4919/49.

Verifiability: If the Λ\Lambda budget does not agree with suppression by 19/4919/49 — the theory is revised. The prediction is testable through precision measurements of the cosmological constant.

Source: Noether charges + Cosmological constant [T]

Prediction 20: Analytical suppression parameter εeff\varepsilon_{\mathrm{eff}} (P6)​

Intuition. Where does the quark mass hierarchy come from? Why is the top quark 10,000 times heavier than the up quark? In the Standard Model this is a tuning question — Yukawa constants are free. CC predicts that the suppression parameter εeff\varepsilon_{\mathrm{eff}} is analytically computable from the structural constants of the Gap potential. All mass ratios (mc/mt∼ε2m_c/m_t \sim \varepsilon^2, mu/mt∼ε4m_u/m_t \sim \varepsilon^4) become consequences.

Prediction [C at (SV)] — falsifiable
εeff=4 ∣γˉ∣sect9 (1+Σ0/4)≈0.055–0.057\varepsilon_{\mathrm{eff}} = \frac{4\,|\bar{\gamma}|_{\mathrm{sect}}}{9\,(1 + \Sigma_0/4)} \approx 0.055\text{–}0.057

At the chosen inputs the self-consistency expression gives 0.05690.0569 in the sector ansatz and 0.05490.0549 without it. The loop estimate gives 0.05870.0587. Calculation The Yukawa hierarchy suppression parameter is an analytical algebraic function of structural constants of the Gap potential, not a free parameter. All mass ratios (mc/mt∼ε2m_c/m_t \sim \varepsilon^2, mu/mt∼ε4m_u/m_t \sim \varepsilon^4) are predicted.

Verifiability: Non-perturbative computation (lattice or variational) of the self-consistent vacuum must give ε33∗∈[0.04,0.08]\varepsilon_{33}^* \in [0.04, 0.08]. Going outside these bounds — falsification of (SV).

Source: Analytical ε [C at (SV)]


VIII. Engineering Predictions: CPTP and Digital Implementations​

The eighth group of predictions is aimed directly at verification in digital systems — predictions that can be tested right now, without neuroimaging equipment.

Prediction 19: CPTP-anchor validation​

Intuition. Any implementation of the CC architecture uses an anchor map π\pi to connect the external world with the internal 7-dimensional space. CC requires this map to be CPTP-compatible (completely positive, trace-preserving). The distance to the canonical anchor is computable in O(49D)O(49D) — linear in input dimensionality.

Prediction [T] — falsifiable
∀ anchor π ⁣:∥π−πcan∥⋄ computable in O(49D)\forall \text{ anchor } \pi\colon \|\pi - \pi_{\mathrm{can}}\|_\diamond \text{ computable in } O(49D)

For any anchor π\pi the distance to the canonical anchor πcan\pi_{\mathrm{can}} in the diamond norm is computable in O(49D)O(49D).

Experimental verification:

  1. Train a SYNARC agent on a standard task (BabyLM).
  2. Compute ∥π−πcan∥⋄\|\pi - \pi_{\mathrm{can}}\|_\diamond after training.
  3. Prediction: ∥π−πcan∥⋄<0.1\|\pi - \pi_{\mathrm{can}}\|_\diamond < 0.1 upon convergence of training.
  4. Falsification: ∥π−πcan∥⋄>0.1\|\pi - \pi_{\mathrm{can}}\|_\diamond > 0.1 systematically at n>50n > 50 batches — CPTP-compatibility is violated.

Verifiability: If SYNARC implementation shows ∥π−πcan∥⋄>ε0\|\pi - \pi_{\mathrm{can}}\|_\diamond > \varepsilon_0 for trained π\pi — the architecture requires revision. Numerical falsification criterion: ∥π−πcan∥⋄>0.1\|\pi - \pi_{\mathrm{can}}\|_\diamond > 0.1 at n>50n > 50 batches of input data → falsification of CPTP-compatibility. Violation of T-152 means loss of the CPTP property during training.

Source: T-152 [T]


IX. Empirical Predictions: Neural Data​

The ninth and final group of predictions — the most ambitious. It requires experimental verification on living systems — and it is precisely this that is potentially most destructive for CC in case of refutation.

Prediction 21: Reconstruction of Γ\Gamma from neural data (πbio\pi_{\mathrm{bio}}, P8)​

Intuition. If CC describes reality, there must exist a map πbio\pi_{\mathrm{bio}} translating neural data (EEG, fMRI, HRV) into the 7-dimensional matrix Γ\Gamma. Moreover, this map must be unique up to G2G_2-gauge — as a coordinate system is unique up to rotation.

Prediction [H] — falsifiable
πbio:(EEG,fMRI,HRV)→D(C7)\pi_{\mathrm{bio}}: (\mathrm{EEG}, \mathrm{fMRI}, \mathrm{HRV}) \to \mathcal{D}(\mathbb{C}^7)

There exists a map πbio\pi_{\mathrm{bio}}, unique up to G2G_2-gauge, such that:

  • P(Γwake)>2/7P(\Gamma_{\mathrm{wake}}) > 2/7 for waking subjects
  • P(ΓNREM3)<2/7P(\Gamma_{\mathrm{NREM3}}) < 2/7 during deep sleep
  • PCI∝Φ(Γ)\mathrm{PCI} \propto \Phi(\Gamma) (monotone connection)

Experimental verification:

  1. Simultaneous TMS-EEG + HRV measurement in N=30 subjects (waking, sleep, anaesthesia).
  2. Apply πbio\pi_{\mathrm{bio}} and verify:
    • Concordance of the independently reconstructed four-gate verdict with PCImax⁡>0.31\mathrm{PCI}_{\max}>0.31 on held-out sessions (P8.4).
    • Critical exponents β=1/4\beta = 1/4 at the sleep-waking transition

Numerical falsification criterion: ∥Γrecon−Γtrue∥F/∥Γtrue∥F>0.3\|\Gamma_{\mathrm{recon}} - \Gamma_{\mathrm{true}}\|_F / \|\Gamma_{\mathrm{true}}\|_F > 0.3 at n>50n > 50 subjects → systematic inconsistency of πbio\pi_{\mathrm{bio}}. If κ<0.4\kappa < 0.4, P8.4 is falsified: the verdicts disagree. If κ≥0.8\kappa \geq 0.8 but the exponents differ, T-161 is falsified.

Source: πbio\pi_{\mathrm{bio}} protocol [H]

Interdisciplinary consequences:

  • Clinical neuroscience: unified diagnostic tool for all disorders of consciousness (coma, minimally conscious state, locked-in).
  • Neuropharmacology: tracking the action of anaesthetics and psychoactive substances through Γ\Gamma-dynamics.

Prediction 22: Spectral gap and neural oscillations [H]​

Intuition. The brain "pulses" at characteristic frequencies: alpha (~10 Hz), gamma (~40 Hz). Where do these frequencies come from? CC proposes an answer: the characteristic frequency of conscious processes is determined by the spectral gap of the Lindbladian L0\mathcal{L}_0 — the difference between the zeroth and first eigenvalues. Primitivity of L0\mathcal{L}_0 guarantees this gap is positive.

Prediction: The characteristic frequency of conscious processes is determined by the spectral gap of the linear part of the Liouvillian:

νconscious∼λgap(L0)2π\nu_{\text{conscious}} \sim \frac{\lambda_{\text{gap}}(\mathcal{L}_0)}{2\pi}

Primitivity of L0\mathcal{L}_0 [T-39a] guarantees λgap>0\lambda_{\text{gap}} > 0. If ω0∼40\omega_0 \sim 40 Hz (gamma rhythm), then λgap∼2π⋅40≈250\lambda_{\text{gap}} \sim 2\pi \cdot 40 \approx 250 rad/s.

Uniqueness of the prediction. No theory of consciousness derives the gamma rhythm frequency from mathematical structure. CC is the only one connecting the spectrum of the evolution operator with the frequency of neural oscillations.

Consequences:

  • Gamma oscillations (30–100 Hz) correspond to the fundamental mode λgap\lambda_{\text{gap}}
  • Alpha rhythm (~10 Hz) — sub-harmonic or slow gap mode
  • Canonical discretisation T-131: δτ=π/(2∥L0∥op)\delta\tau = \pi/(2\|\mathcal{L}_0\|_{\text{op}})

Experimental verification:

  1. Calibrate λgap\lambda_{\text{gap}} from EEG data of healthy subjects.
  2. Compute λgap\lambda_{\text{gap}} for realistic Lindblad parameters.
  3. Prediction: agreement with the gamma range (30–100 Hz).
  4. Falsification: computed λgap\lambda_{\text{gap}} lies outside the range of neural oscillations.

Status: [H] (hypothesis). Requires: (1) calibration of ω0\omega_0 from neurodata, (2) computation of λgap\lambda_{\text{gap}} for realistic Lindblad parameters.

Prediction 23: The rank-7 decoherence-anisotropy law [T structure / C mapping]​

Intuition. The heat channel of a holon carries seven independent temperatures — one per Fano line (line-resolved temperatures). If the dissipator really is Fano-wired, the twenty-one pairwise decoherence rates between channels cannot be arbitrary: each is the mean of exactly four line rates, and the pair's own joint line cancels.

Prediction: the pairwise decoherence rates obey the exact law

rij=16∑p : ∣ℓp∩{i,j}∣=1γp,r_{ij} = \frac{1}{6}\sum_{p\,:\,|\ell_p \cap \{i,j\}| = 1} \gamma_p ,

so the 2121-vector of measured rates lies on a 77-dimensional linear subspace: 14 exact linear relations must hold, with the specific Fano incidence pattern (any non-Fano seven-line design spans a different subspace). The seven line temperatures are then reconstructible from rate tomography (least squares on the incidence map; the law and the reconstruction are machine-exact, 4⋅10−164\cdot10^{-16} and 3⋅10−153\cdot10^{-15}).

Uniqueness of the prediction. The rank-7 constraint with the single-incidence pattern is a fingerprint of PG(2,2) wiring — no scalar-temperature theory (one TT) and no unstructured multi-rate model (21 free rates) produces exactly these 14 relations.

Experimental verification:

  1. Estimate the 21 inter-channel decoherence rates from perturbational protocols (PCI-style rate tomography over the Γ^\widehat\Gamma estimate).
  2. Regress the rate vector onto the Fano incidence map; record the residual.
  3. PASS: residual consistent with estimation noise (rank ≤ 7 with the Fano pattern). FAIL: a statistically significant residual — this falsifies the Fano structure of the dissipator itself, not merely a parameter choice.

Status: [T] for the law and the rank statement (machine-verified); [C] for the empirical mapping of channel pairs to measurable observables. Source: T-258/T-262, rank-7 law.


Decision Protocols: pass/fail for the near-term testable subset​

A prediction earns scientific weight only with an explicit decision rule: what measured value counts as PASS, what counts as FAIL, at what sample size. Below are the protocols for the predictions testable with current instruments (the physics-sector pass/fail table lives in Falsifiability). Estimation of Γ^\widehat\Gamma, P^\widehat P and the required NN follow the Γ-tomography theorem §6.4.

PredProtocol (observable → decision)PASSFAILCurrent status (2026)
5–7 (thresholds)Anaesthesia induction/emergence; estimate P^(t)\widehat P(t) via πbio\pi_{\mathrm{bio}} frozen on wakefulness, with no viability penalty (SUB-1, SUB-2) — not via a PCI calibration; locate loss/recovery of consciousness (LOC/ROC)LOC/ROC coincides with P^\widehat P crossing 2/72/7 within one anaesthetic time-constant, and Cons(Γ^)\mathrm{Cons}(\widehat\Gamma) agrees with PCImax⁡>0.31\mathrm{PCI}_{\max} > 0.31 at κ≥0.8\kappa \geq 0.8P^\widehat P crosses 2/72/7 with no behavioural transition, LOC/ROC occurs at P^\widehat P far from 2/72/7, or κ<0.4\kappa < 0.4UNTESTED — no πbio\pi_{\mathrm{bio}} session exists. The earlier "PARTIAL — clinical PCI threshold 0.310.31 maps to PcritP_{\text{crit}}" rested on a two-point line that put 2/72/7 at 0.310.31 by construction
1 (No-Zombie)For a system passing viability with DΩ≠0\mathcal{D}_\Omega \neq 0, estimate Coh^E\widehat{\mathrm{Coh}}_Eevery such system has Coh^E>1/7\widehat{\mathrm{Coh}}_E > 1/7 (CI excludes 1/71/7)a viable dissipative system with Coh^E≤1/7\widehat{\mathrm{Coh}}_E \leq 1/7UNTESTED (needs πbio\pi_{\text{bio}})
2 (κ∝CohE\kappa \propto \mathrm{Coh}_E)Perturb interoceptive integration; measure recovery rate κ^\widehat\kappa against Coh^E\widehat{\mathrm{Coh}}_E across conditionsκ^\widehat\kappa monotone increasing in Coh^E\widehat{\mathrm{Coh}}_E, slope >0> 0 at p<0.01p < 0.01no dependence, or negative slopeUNTESTED
12 (depth diagnostics)Freeze Fano channel and detector; measure held-out amplitude ratios, then independently test metamodel tasksRatios agree with 3−n3^{-n} within declared uncertainty; task bridge passes independentlyChannel-law discrepancy or failed held-out task associationArithmetic ceiling is defined; cognitive bridge UNTESTED
21 (πbio\pi_{\mathrm{bio}})Reconstruct Γ^\widehat\Gamma from EEG per Γ-tomography §6.4; check PSD, unit trace, test–retestΓ^⪰0\widehat\Gamma \succeq 0, reproducible across sessions (within the tomography CI)non-PSD or irreproducible embeddingUNTESTED (reference implementation pending)

Verdict legend. PASSING — measurement inside the pass band; CONSISTENT — not excluded, but beyond current sensitivity; PARTIAL — indirect/calibration-dependent support; UNTESTED — no experiment has probed the band. A single FAIL falsifies the corresponding claim at its status level. The physics-sector pass/fail table (with current 2026 verdicts) lives in Falsifiability →.


Summary Table of Predictions​

#PredictionFormulaStatusStatus in other theories
1No-ZombieViable∧DΩ≠0⇒CohE>1/7\mathrm{Viable} \land \mathcal{D}_\Omega \neq 0 \Rightarrow \mathrm{Coh}_E > 1/7[H]Absent
2E-coherence ↔ regenerationκ∝CohE\kappa \propto \mathrm{Coh}_E[D/H]Absent
37-dimensional stressσsys∈R7\sigma_{\mathrm{sys}} \in \mathbb{R}^7[D/H]Absent
4Pre-linguistic cognitionCognition⇏Language\mathrm{Cognition} \not\Rightarrow \mathrm{Language}[I]Partial in FEP
5Collective consciousnessCorrelation criterion[T] necessary / [H] sufficiencyIIT: exclusion — a group is a subject only as a maximum of integrated information
6Minimum coherenceP>Pcrit=2/7P > P_{\text{crit}} = 2/7[T]Absent
7Stability radiusrstab≈K(P−1/7−1/7)r_{\mathrm{stab}} \approx K(\sqrt{P-1/7}-\sqrt{1/7})[C] (T-104)Absent
8Enc capacityCEnc≤log⁡27C_{\mathrm{Enc}} \leq \log_2 7[T]Absent
9Learning boundnopt=max⁡(ninfo,ndyn,nstab)n_{\mathrm{opt}} = \max(n_{\mathrm{info}}, n_{\mathrm{dyn}}, n_{\mathrm{stab}})[T]Absent
10N=7 for learningN<7⇒n∗=∞N < 7 \Rightarrow n^* = \infty[H]Absent
11N=7 for social learning3ToM+3ISL+1U=73_{\text{ToM}} + 3_{\text{ISL}} + 1_U = 7[H]Absent
12Declared depth diagnostics3−n3^{-n} amplitude survival; score ceiling only under its definition[T/D] law; [H] cognitive bridgeArithmetic implementation checks
13Genesis timengenesis≤⌈ln⁡Δ/ln⁡(1/β)⌉n_{\mathrm{genesis}} \leq \lceil\ln\Delta / \ln(1/\beta)\rceil[T]Absent
14Phase coherenceρij∗(t)∝e−i(Ei−Ej)t\rho^*_{ij}(t) \propto e^{-i(E_i-E_j)t} for Φ≥1\Phi \geq 1[T]Absent
15Attractor inside the windowP∗∈(Pc,P∞)P^* \in (P_c, P_\infty), P∞≤5/14<3/7P_\infty \leq 5/14 < 3/7 (was →3/7\to 3/7)[C at (MaxΦ)]Absent
16Conditional passage lawT=a−1log⁡[yf(a+by0)/(y0(a+byf))]T=a^{-1}\log[y_f(a+by_0)/(y_0(a+by_f))] under y˙=ay+by2\dot y=ay+by^2[T] conditional reduction / [H] identificationNo universal avalanche
17Conditional sextic exponentsα=1/2,β=1/4,γ=1,δ=5\alpha=1/2,\beta=1/4,\gamma=1,\delta=5; ν=1/2\nu=1/2 needs spatial Gaussian model[C] tuned scalar model plus phenomenological bridgeUntested
18Ward suppression 19/49Gap fluctuations × 19/49\times\, 19/49[T]Absent
19CPTP-anchor validation∥π−πcan∥⋄\|\pi - \pi_{\mathrm{can}}\|_\diamond in O(49D)O(49D)[T]Absent
20Analytical εeff\varepsilon_{\mathrm{eff}}εeff=4N33/(9∥γˉ∥(1+r4Σ0/2))\varepsilon_{\mathrm{eff}} = 4N_{33}/(9\|\bar{\gamma}\|(1+r_4\Sigma_0/2))[C at (SV)]Absent
21πbio\pi_{\mathrm{bio}}: neural data → Γ\Gammaπbio:EEG/fMRI/HRV→D(C7)\pi_{\mathrm{bio}}: \mathrm{EEG/fMRI/HRV} \to \mathcal{D}(\mathbb{C}^7)[H]Partial in IIT
22Spectral gap → neural oscillationsνconscious∼λgap/(2π)\nu_{\text{conscious}} \sim \lambda_{\text{gap}}/(2\pi)[H]Absent
23Rank-7 decoherence-anisotropy lawrij=16∑p: ∣ℓp∩{i,j}∣=1γpr_{ij} = \tfrac{1}{6}\sum_{p:\,\lvert\ell_p\cap\{i,j\}\rvert=1}\gamma_p (14 exact sum-rules; 21 rates on a 7-dim Fano subspace)[T] law / [C] mapping (T-262)Absent

Theory Comparison: CC vs IIT vs FEP vs GWT​

The following table shows which predictions each of the leading theories of consciousness makes. "+" means the theory generates the given prediction; "~" — gives a qualitative statement without numerical forecast; "-" — does not contain the given prediction.

PredictionCCIITFEPGWT
Impossibility of zombies+ (theorem [T]; "no zombies" reading [I])---
Connection of experience with regeneration+ (κ∝CohE\kappa \propto \mathrm{Coh}_E)---
Finite classification of stresses+ (7 components)-~ (free energy)-
Threshold values of consciousness+ (Pcrit=2/7P_{\text{crit}} = 2/7, Rth=1/3R_{\text{th}} = 1/3, Φth=1\Phi_{\text{th}} = 1)~ (Φ>0\Phi > 0, no number)-~ (ignition, no number)
Minimum dimensionality+ (N=7N = 7)---
Collective consciousness~+ (exclusion: only a maximum of integrated information is a subject)--
Self-awareness depthDeclared score arithmeticNo audited comparisonNo audited comparisonNo audited comparison
Genesis time+ (formula ngenesisn_{\text{genesis}})-~ (self-organisation)-
Learning rate bounds+ (noptn_{\text{opt}}, three bounds)-~ (expected free energy)-
Critical exponents[C] tuned sextic mean-field reduction and tested bridgeNo audited comparisonNo audited comparisonNo audited comparison
Stability radius+ (rstabr_{\text{stab}})-~ (resilience)-
Phase coherence → integration+ (co-rotation)---
Neural oscillations from spectral gap+ [H]---
Reconstruction of Γ\Gamma from neurodata+ [H]~ (Φ\Phi from connectome)--
Yukawa hierarchy+ (εeff≈0.059\varepsilon_{\text{eff}} \approx 0.059) [C]---
Ward suppression of Λ\Lambda+ (19/4919/49)---

Numerical model estimates require an explicit preparation, readout and withheld observation to count as independently testable predictions.

The difference is fundamental. IIT, FEP, and GWT are frameworks — they offer a descriptive language but do not generate numerical forecasts that can be unambiguously confirmed or refuted. CC is a theory in the Popperian sense: it makes risky, specific, falsifiable predictions.


Falsification with explicit scope​

A counterexample to a formal implication, evidence against a physical identification and disagreement under a chosen calibration are different test outcomes. State which assumptions the observation tests.

Each empirical test fixes its observation model, identifiability, uncertainty, and admissible interventions. A simulated matrix, a chosen threshold, or agreement with training labels is not independent confirmation of the theory.

A testable programme​

Exact finite mathematical constructions survive in their stated domains. Their physical identification is tested separately. The registry records current statuses; the measurement protocol states when observations can distinguish the model from alternatives.