EEG Power Law as Cortical Turbulence Spectrum — The Dual Cascade Hypothesis

Every EEG recording shows the same background: power drops steadily as frequency rises, following a rough power law (PSD ~ 1/f^α). For decades this was treated as colored noise — physiological background to be removed before analyzing the "real" oscillations. The Cooray 2026 paper changes that framing: if cortical activity is governed by wave turbulence, the 1/f^α slope is not noise. It is a Kolmogorov-Zakharov (KZ) cascade spectrum — the neural analog of the Kolmogorov energy cascade in fluid turbulence. The slope of that spectrum encodes which turbulence regime the cortex is operating in.

What Turbulence Theory Predicts

Classical fluid turbulence (Kolmogorov 1941, K41) produces a characteristic PSD slope of f^(-5/3) ≈ f^(-1.67) in the inertial range. This is strong turbulence, where the nonlinear coupling term dominates the linear oscillation term (ε >> 1, the coupling parameter).

Weak wave turbulence (Kolmogorov-Zakharov theory, KZ) applies when ε << 1 — waves interact weakly, and resonant interactions mediate energy transfer. The KZ spectrum's exponent depends on the form of the wave dispersion relation and whether 3-wave or 4-wave interactions dominate:

Interaction type	KZ spectrum prediction	Character
3-wave resonance	PSD ~ f^(-8/3) ≈ f^(-2.67)	Steeper than K41
4-wave resonance	Varies by dispersion; typically f^(-2) to f^(-3)	Even steeper
K41 strong turbulence	f^(-5/3) ≈ f^(-1.67)	Shallowest power law

What EEG Actually Shows

Standard EEG aperiodic analysis (FOOOF/SpecParam on scalp recordings) finds spectral slopes between α ≈ 1.5 and 3, with the typical resting state at α ≈ 2–2.5 in the 1–40 Hz range. This sits comfortably inside the weak turbulence prediction band (f^(-2) to f^(-8/3)), and significantly steeper than the K41 prediction (f^(-1.67)).

A March 2026 stereotactic intracranial EEG study (MDPI Bioengineering; 37 brain regions, 106 subjects) revealed a more complex structure: two separate power-law regimes in the same signals, at different frequency bands:

Low-frequency band (~0.5–4 Hz): shallower power-law slope (α closer to 1–1.5)
High-frequency band (~33–80 Hz): steeper power-law slope (α ≈ 2–3), differing between cortical and subcortical areas

This dual power-law structure is the empirical finding that wave turbulence theory predicts but had not been identified in neural data in this framing.

The Dual Cascade Interpretation

Classical wave turbulence theory produces a dual cascade: an inverse cascade (energy moving toward large scales / low frequencies) and a direct cascade (energy moving toward small scales / high frequencies). The driven scale — where energy is injected — sits in between, with different exponents on each side.

In EEG:

The alpha peak (~10 Hz) may be the "pump" frequency — the scale at which thalamocortical resonance drives the system and injects energy into the neural field
Below the alpha peak: inverse cascade flowing toward low frequencies → shallower slope
Above the alpha peak: direct cascade flowing toward high frequencies → steeper slope

This maps directly onto the 2026 intracranial EEG finding: shallower exponent in the delta/theta band (below alpha), steeper exponent in the gamma band (above alpha). The alpha peak sits at the crossover.

The Discriminating Test

If this framing is correct, the two frequency regimes in intracranial EEG carry the signatures of distinct cascade directions, not two separate biological mechanisms. The discriminating predictions:

1. Slope dependency on alpha frequency: If the alpha peak is the pump, shifting the alpha frequency (e.g., via working memory load, which suppresses alpha, or with anesthesia, which slows it) should shift the crossover point between the two power-law regimes. Both slopes should pivot around the moving pump.

2. Cortical vs. subcortical exponent difference: Wave turbulence theory predicts that different regions of the dissipation zone (analogous to different distances from the energy injection scale) show different slopes. The 2026 intracranial EEG finding that cortical and subcortical exponents differ in the high-frequency band is consistent: subcortical structures drive the pump; cortical areas are further in the cascade.

3. Cognitive state transitions change the exponent: The direct cascade exponent should steepen during focused attention (more energy dissipation at high frequencies, higher gamma) and shallow during drowsiness (less high-frequency activity). This is consistent with the established finding that the aperiodic EEG exponent steepens with cognitive engagement and flattens during sleep.

4. Anesthesia as turbulence collapse: General anesthesia suppresses cortical dynamics toward a near-linear regime. Wave turbulence predicts that reducing the nonlinear coupling (ε → 0) should make the PSD flatter and more monochromatic — the cascade collapses toward a single-frequency driven oscillation. The burst-suppression EEG pattern under anesthesia is broadly consistent with this.

The Paradox Revisited

concept neural turbulence reynolds documents the Re_neural paradox: substituting Robinson corticothalamic parameters into the Cooray weak turbulence equation gives Re_neural ≈ 1.2 (subcritical), yet the cortex shows empirical power laws consistent with turbulence.

The EEG spectrum adds a second layer to this paradox:

Empirical EEG slopes (α ≈ 2–2.5 in the 1–40 Hz range) are steeper than K41 (f^(-1.67)) and consistent with weak turbulence (KZ: f^(-8/3) ≈ -2.67 for 3-wave)
But the cortex operates near ε ~ 1 (branching ratio σ ≈ 1), where weak turbulence theory breaks down
So the EEG spectrum looks like weak turbulence output, yet the coupling strength says it should be strong turbulence

Three possible resolutions:

Neural K41 has a different exponent than fluid K41: The cortical dispersion relation modifies the strong turbulence cascade exponent away from -5/3. If neural field waves are dispersive (unlike pressure waves in simple fluids), the K41 analog could give steeper slopes.
Two cascades at two regimes: The dual-frequency structure suggests the two bands may genuinely be in different turbulence regimes — the low-frequency band in something like K41 and the high-frequency band in weak turbulence — with the alpha pump sitting at the transition.
The Reynolds number is wrong: The ν_eff computation from Robinson parameters may not capture the correct viscosity analog for strong turbulence, and the actual effective Re_neural (once correctly computed for strong turbulence) is large.

Clinical Implication: The Aperiodic Exponent as a Turbulence Regime Probe

The aperiodic EEG exponent (1/f^α slope) has been linked to:

E/I balance (excitation-inhibition ratio): higher α → more inhibition; lower α → more excitation
Cognitive aging: slope flattens with age and pathology
ADHD, schizophrenia, epilepsy: distinctive aperiodic signatures
Anesthesia depth: slope flattens under general anesthesia

Wave turbulence reframes these findings: the slope change is a turbulence regime shift, not merely a biological parameter change. When excitation-inhibition balance shifts, ε changes — the coupling between waves changes — and the cascade exponent changes correspondingly. The "E/I ratio" in clinical neuroscience is a different label for the wave turbulence coupling parameter.

This reframing has a concrete implication: slope measurements in different frequency bands should disambiguate E/I shifts (which affect ε globally) from cascade direction changes (which affect the two frequency bands differently). A drug that blocks gamma-aminobutyric acid (GABA) inhibition would shift ε upward, flattening both cascade slopes together. A drug that selectively disrupts thalamocortical alpha resonance would shift the crossover frequency while preserving the cascade exponents.

Status of the Field

Prediction	Evidence status
EEG shows power-law behavior	Established (Beggs & Plenz 2003 and many subsequent)
Two power-law regimes at different frequency bands	Emerging (2026 intracranial EEG; other studies confirm frequency-dependent exponent)
Alpha peak as energy pump frequency	Proposed; supported by alpha-dominant thalamocortical resonance; not formally tested in wave turbulence framing
Slope tracks cognitive state	Established (aperiodic exponent literature 2020–2026)
Slope reinterpretation as cascade exponent	Theoretical; derives from Cooray 2026; no experimental test designed for this specific prediction
K41 vs. KZ discrimination test	Open gap — no study has run this

Key Sources

Cooray, Gerald K. (2026). "Wave Turbulence and Cortical Dynamics." arXiv 2507.23525; Frontiers in Computational Neuroscience DOI 10.3389/fncom.2026.1682176. PMC13047063. The foundational derivation of the wave turbulence kinetic equation for neural fields and dual cascade.
MDPI Bioengineering (March 2026). "Scale-Free Neurodynamics as Functional Fingerprint of Brain Regions." Analysis of 37 brain regions in 106 subjects, revealing dual power-law regimes in 0.5–4 Hz and 33–80 Hz bands.
Donoghue, Thomas et al. (2020). "Parameterizing neural power spectra into periodic and aperiodic components." Nature Neuroscience. DOI 10.1038/s41593-020-00744-x. The FOOOF/SpecParam methodology for extracting the aperiodic exponent.
Donoghue, Thomas et al. (2024). "A neurophysiological basis for aperiodic EEG and the background spectral trend." Nature Communications. DOI 10.1038/s41467-024-45922-8. Mechanistic model grounding the aperiodic exponent in neural dynamics.
Beggs, John M. and Plenz, Dietmar (2003). "Neuronal Avalanches in Neocortical Circuits." Journal of Neuroscience. The critical branching ratio σ ≈ 1 measurement establishing the cortex at near-criticality.

Abhishek