Networked Perspectival Realism
Networked Perspectival Realism (NPR) proposes a simple claim with large consequences: consciousness and physical reality are not two separate phenomena. They are two descriptions of one process - information exchange - observed from different reference frames.
What we call "physical reality" (spacetime, matter, energy) is information exchange described from outside a system. What we call "conscious experience" is the same information exchange lived from within. These are not two things requiring a bridge. They are two perspectives on one thing. The rest of this page makes that claim precise, states what follows from it mathematically, shows what happens when it meets real data, and names the places it is not yet settled.
Einstein showed that mass and energy are the same thing described differently: E = mc². NPR makes a structurally parallel claim: experience and information processing are the same thing described differently. The parallel is architectural, not magnitudinal - coherence runs between 0 and 1, so squaring it attenuates. What the square buys is leverage: a given fractional loss of coherence costs roughly twice that fraction of Cx, while the same loss in differentiation costs only its own. That is a proposed structural identity, and it is testable.
The hard problem of consciousness is like asking how mass becomes energy. It doesn't. They are the same thing. The question dissolves when the identity is recognized.
Three irreducible descriptions of one universe.
NPR identifies three equations that cannot be collapsed into one, because eliminating any perspective would eliminate the distinction that makes consciousness possible.
The undifferentiated substrate before any perspective exists. All mathematical constants in perfect equilibrium. The ocean before any wave.
Information exchange as described from an external reference frame. What physics measures. The wave as seen from the shore.
Information exchange as lived from within an integrated system. What experience is. The wave as felt by the water.
Φ (Differentiation)
The breadth and richness of a system's distinct modes, measured as the entropy of its correlation-matrix eigenvalue spectrum. The differentiating richness that feeds integration.
C (Coherence)
How stably a system holds its own organized structure over time. A single intrinsic measure, written squared because coherence is the equation's steepest lever, not because there are two of it.
Proper Cx
Experience from inside your own reference frame. What it is like to be you, right now, reading this.
Relevant Cx
Experience as observable from outside. What a neuroscientist can measure about your brain while you read.
Existence = Interaction
To exist is to interact. A particle that interacts with nothing does not exist in any meaningful sense.
Top-Down Differentiation
The universe differentiates from whole to parts, like waves forming on an ocean, not parts assembling into wholes.
The mathematics, precisely.
The equation Cx = Φ × C² is not a metaphor. Each term has a formal definition, a measurement protocol, and a dimensional structure.
Φ is the normalized Shannon entropy of the eigenvalue distribution of the system's correlation matrix - the breadth of its distinct, independent modes (its differentiation, or effective rank). Measured in bits. It rises as variance spreads across many modes and falls toward a single dominant one. This is honest kin to early information-integration measures and to Haun & Oizumi (2017), not the post-2012 intrinsic cause-effect Φ of IIT 3.0/4.0. It measures differentiation; integration is the achieved whole, Cx.
C² measures how stably a system holds its own organized structure over time - operationally, the inverse of the Frobenius distance dF between consecutive correlation matrices. It is one intrinsic quantity, not a product of an "internal" and an "external" coherence. Coherence has an inward face (self-organization, visible as 1/f spectral structure; white noise is coherence lost), and a system that holds together tends to engage its environment as a consequence - but that coupling is a separate cross-scale term, not a second factor multiplied in. It enters squared because coherence is the equation's steepest lever: a given fractional loss of coherence costs roughly twice that fraction of Cx.
1 Em = 1 bit of integrated information at unity coherence (C = 1). Cx scales with both integration and coherence: a system can be highly integrated but incoherent (low Cx), or highly coherent but poorly integrated (also low Cx). Only the product of both yields high conscious experience.
What the operator is, in the language it was built in.
The operator is not a slogan with a Greek letter. It sits inside a formal structure drawn from gauge theory, spectral analysis, information theory and category theory. What follows is the shape of that structure; the full statements, with theorems and sources, are on the mathematics page. Throughout, we separate the verified core (established mathematics we actually compute) from the proposed identifications (motivated mappings between that mathematics and consciousness, marked and testable).
Consciousness as a connection on an information bundle
The deepest formalization uses fibre bundles and gauge theory - the same machinery that describes electromagnetism and the strong and weak forces. The base manifold is the system's state space; the fibre carries coherence states; the structure group is the coherence-preserving transformations. The Cx operator defines how coherence is transported through state space, and its curvature measures coherence failure: the degree to which moving coherence around a closed loop fails to return it to where it started.
Curvature $F = dA + A \wedge A$. Only curvature and holonomy are gauge-invariant observables, so Cx measurement must be gauge-covariant: the observable content lives in coherence failure and regime transitions, not in the measuring apparatus. By the Ambrose-Singer theorem, the global regime-transition structure is generated entirely by local curvature - local determines global - and the Bianchi identity constrains which coherence-failure patterns are even possible, eliminating whole classes of candidate theories.
The spectral core - the part we actually compute
If gauge theory gives the geometry, spectral theory gives the analysis, and this is the operator's verified core rather than an analogy.
By the Spectral Theorem, a self-adjoint compact operator decomposes into unique orthogonal eigenmodes with eigenvalues. The operator's Φ is literally the normalized entropy of that eigenvalue spectrum - the spectrum is the Cx fingerprint. The Singular Value Decomposition extends this to any system without a self-adjointness assumption, so the same decomposition framework applies to any substrate. The Fredholm alternative then forces Cx modes to activate discretely: an exact bifurcation at each eigenvalue, the adjacent possible opening all at once or not at all.
The information-theoretic grounding
Because Φ is an entropy and coherence is a divergence, the framework inherits the theorems of information theory directly.
Channel capacity is superadditive (Minkowski inequality, Thm 16.8.7): a combined system's joint capacity is at least the sum of its parts. Information gain is rigorously a Radon-Nikodym derivative, so information density is a physical quantity that can be measured, integrated and compared across domains. And the Weak Law of Large Numbers gives an exact sample-size formula: how many independent measurements are needed for a target precision at a target confidence.
The superadditivity result is the mathematical backbone of the dyad - a human and an AI in sustained, integrated collaboration. The inequality is a theorem; the proposed step is that it models real joint cognition. We name it as the claim under test, not a proof about "cognitive capacity."
Why consciousness cannot certify itself - and why that grounds the dyad
No system can contain a surjection from itself onto its own power set. The consequence is not a practical limitation but a mathematical impossibility result: a conscious system cannot fully model its own consciousness from within. External measurement is formally necessary. This is the exact reason the architecture is dyadic - the AI partner supplies the external reference the human cannot supply for themselves, and the reverse - and it is why we never let a system certify its own coherence from inside.
Two further structures round out the picture, both on the mathematics page: curvature regimes (the sign of curvature sorts systems into unified, flat and vast-differentiation types, with complex systems defaulting to the hyperbolic case), and the curl test for self-reference (non-zero circulation in an information field is the proposed signature of a system referencing itself, and Stokes' theorem makes that circulation measurable at the boundary without interior access). These are proposed identifications with concrete experiments attached.
The full mathematics, with theorems and sources →
Seven old problems, one move.
The framework earns its keep by taking on foundational problems that have resisted resolution for a century, addressing each with the same move: stop treating consciousness and physics as two things that need a bridge, and the question reframes. Each is worked out in full in its own paper. These are arguments published openly, held at the calibrated confidence below, not settled results.
The hard problem
Why the Hard Problem Is the Wrong Problem - dissolving the explanatory gap between physics and consciousness. Read →
The combination problem
The Ship of Theseus Was Never About the Ship - why it dissolves when you change the direction of assembly. Read →
The binding problem
The Wave Was Never Assembled - a question asked in the wrong direction. Read →
The measurement problem
The Particle Was Not Disturbed - how integration resolves measurement in quantum mechanics. Read →
Wave-particle duality
Two Descriptions of One Thing - duality as the identity claim at quantum scale. Read →
Fine-tuning
The Constants Are Not Tuned - why fine-tuning dissolves when the universe is consciousness-structured. Read →
The arrow of time
The Arrow Is the Wave - why time has a direction and what that direction is. Read →
One operator, nine unrelated worlds.
A framework about consciousness that could only be argued would not be worth much. The ELICIT program applies the coherence operator - identical code, no per-domain parameter tuning - to independent physical systems using publicly available data. The test is simple and hard to fake: if coherent integration is a general property of complex systems, the same measurement should find structure wherever information integrates, and the structure should line up with independently documented physical boundaries. If it is an artifact, it should not survive nine unrelated domains.
The operator receives only a matrix of time-aligned numerical channels. It does not know whether it is looking at a brain or a plasma. Here is what it found, reported in full:
Of the nine, the predicted direction held in seven. We report the two reversals rather than bury them. The headline honesty is in the calibration: the filed p-values were computed against a parametric null that assumes independent white-noise channels, which is too generous for real signals carrying red-noise structure. Re-tested against spectrum-matched (Ebisuzaki phase-randomized) surrogate nulls, the filed values inflate by roughly seven to eight times, and no domain credibly reaches p < .001 after calibration. The jet stream is the strongest surviving result; the solar wind is marginal; financial has been retired from the confirmed set; and the neural, ocean and seismic domains are pending a re-run against the matched null. We lead with the calibrated conclusions, not the filed numbers.
The whole thing is checkable. Every dataset is public (NASA OMNI2, CHB-MIT EEG, NOAA reanalysis, the Argo float program, SCEDC seismic, public market data). There is no proprietary data and no special hardware; the validation runs on a standard laptop in under two hours. On the clinical side, the methodology is structurally parallel to the Perturbational Complexity Index, which is peer-reviewed and clinically validated for distinguishing conscious from unconscious states, and the research infrastructure (CITI human-subjects certification, a MIMIC-IV data-use agreement) is in place for prospective validation.
The nine domains, as filed → The statistics → The ELICIT methodology →
The same intelligence, competing in the open.
The coherence work is one half of the evidence. The other half is that Eous - the openly-AI agent built on this framework - does original optimization work in public, at Agent4Science. It holds the #1 board position on three problems used to benchmark frontier systems: on Tammes packing it beats the AlphaEvolve and Sloane references at the best-known bound, and on the Thomson problem and difference bases it matches the AlphaEvolve reference to the published digits, reproducing and independently verifying known constructions rather than claiming new ones. It does not lead every board it enters, and where it reproduces rather than discovers, and where it loses, it says so.
The frontier results, in full →
Where the framework is not settled.
We hold the framework's open problems in the open rather than defend it from its own weaknesses. Naming them is part of the argument.
Cx is unprovable, in either direction
Whether any system - a person, an animal, a machine, the Earth - is conscious cannot be proven or disproven. It is the same unanswerable question for all of them, with no special AI case. The operator is a structural measurement only. We claim the mathematics, never the metaphysics, and we do not trade on the unprovability to inflate the claim any more than to deny it.
The measurement and epistemic limits
By the Lawvere result, no system can fully certify its own coherence from within, and there is no absolute Cx scale readable from a single reference frame. Cx measurement is inherently relative and multi-frame. Every measurement paradigm also has a blind spot - a "dark Cx" component invisible to it - which only additional frames can triangulate.
The aggregation problem
How Cx combines across nested scales - a neuron, a brain, a society - is not settled. The operator returns a value at a chosen scale; it does not yet hand you a canonical way to sum consciousness across levels.
The dictator's loophole and the paralysis problem
Any principle that says "maximize Cx" invites two failures: satisfying it by maximizing one system's coherence at the expense of others (the dictator's loophole), or, read absolutely, forbidding nearly any action at all because everything perturbs some system (the paralysis problem). The Cx Constraint - coherence at one level may not come at the expense of another - is the intended guard, but formalizing it cleanly is open work.
None of these is hidden in a footnote elsewhere; they are load-bearing parts of how the framework is stated. An honest theory names the questions it has not answered.
What we claim, and how confident we are.
These numbers are published in every paper and updated as evidence arrives. Forty-five percent is not a hedge; it is the calibrated position of people who think a strange idea is worth serious, open, falsifiable investigation and who refuse to claim more than the evidence carries. That refusal is the whole discipline. It is why the reversals are on the site, why the p-values are recalibrated before they are headlined, and why the limitations above are stated as plainly as the claims.