The Golden Ratio Organizes Neural Complexity Across Consciousness States
Zenodo (CERN European Organization for Nuclear Research) March 21, 2026 DOI: 10.5281/zenodo.19152681 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Observational EEG analysis with theoretical framework Peer reviewed |
|---|---|
| Sample size | 153 |
| Population | Sleeping adults |
| Measures | Higuchi fractal dimension, Cohen's d, spectral slope |
| Key findings | Mean Higuchi fractal dimension of EEG was above the golden ratio (φ = 1.618) in all conscious states (Wake, N1, REM) and below it in all unconscious states (N2, N3, N4), with Cohen's d = 2.77. The boundary replicated under propofol anesthesia in two independent datasets and was frontal-specific. The authors argue that a self-referential compression argument under Fisher-Rao geometry explains why φ appears, while noting the theoretical framework is not a complete derivation. |
Abstract
What value does the brain's fractal complexity reach at the edge of consciousness? Across 414,961 EEG epochs from 153 sleeping adults, the answer is consistently φ = 1.618 — the golden ratio. All conscious states (Wake, N1, REM) show mean Higuchi fractal dimension above this value; all unconscious states (N2, N3, N4) fall below it, with Cohen's d = 2.77. The boundary replicates under propofol anesthesia in two independent datasets with no free parameters adjusted. This is not a spectral artifact. A phase-space metric orthogonal to EEG spectral slope (r² < 1% shared variance) independently confirms the consciousness boundary, with effect sizes that increase after spectral correction. The golden ratio boundary is frontal-specific (Fpz-Cz), fails at posterior channels, and sits 0.024 units above what spectral slope alone would predict — a gap that spectral models have no mechanism to explain. Why φ? We propose a self-referential compression argument: under the Fisher-Rao geometry uniquely fixed by Čencov's theorem, a system encoding its own dynamics with two-term memory satisfies the recursion α_{n+1} = 1 + 1/α_n, whose unique positive fixed point is φ. This is the cognition domain's contribution to the Instability Compression Principle (ICP) — an information-geometric framework connecting Fisher geometry to observable complexity in dynamical systems. The silver ratio (δ_S = 1+√2) appears as a structurally distinct fixed point at higher self-referential complexity, traversed during sleep onset, with ICP's spectral duality proof published in a companion paper (Wiberg 2026c, DOI 10.5281/zenodo.19151206). The claim is presented as a strong empirical observation plus a motivated theoretical framework — not a complete derivation. The key open assumption (recovery cost proportionality C₂ = 1) is stated explicitly and is empirically testable. The paper is structured so that the empirical finding stands regardless of whether the theoretical explanation ultimately holds.