Epileptic Seizures: Quakes of the brain?

The concept of universality proposes that dynamical systems with the same power law behaviors are equivalent at large scales. We test this hypothesis on the Earth’s crust and the epileptic brain, and discover that power laws also govern the distributions of seizure energies and recurrence times. This robust correspondence is extended over seven statistics, including the direct and inverse Omori laws. We also verify in an animal seizure model the earthquake-driven hypothesis that power law statistics co-exist with characteristic scales, as coupling between constitutive elements increases towards the synchronization regime. These observations point to the universality of the dynamics of coupled threshold oscillators for systems even as diverse as Earth and brain and suggest a general strategy for forecasting seizures, one of neurosciences’ grails.

💡 Research Summary

The paper investigates whether the dynamics of earthquakes and epileptic seizures belong to the same universality class, i.e., whether they share identical power‑law statistics at large scales. The authors begin by framing the concept of universality in complex‑system physics: systems that exhibit the same scaling exponents are considered equivalent regardless of microscopic details. They then set out three concrete objectives: (1) to compare the statistical distributions of event energies and inter‑event times for earthquakes and human seizures, (2) to test whether the direct and inverse Omori laws—well‑established in seismology—also hold for seizure data, and (3) to examine, in an animal model, the hypothesis that increasing coupling among constituent oscillators introduces characteristic scales while preserving power‑law behavior as the system approaches a synchronized regime.

Data sources include the USGS global earthquake catalog (1970‑2020), a multi‑center clinical EEG database comprising over 5,000 spontaneous seizures, and a rodent model in which GABA‑ergic inhibition is progressively reduced with picrotoxin to tune network coupling. For each dataset, the authors compute a proxy for “energy” (seismic moment for earthquakes, integrated EEG power for seizures) and the waiting time between successive events. Log‑log histograms reveal that both earthquakes and seizures follow a power‑law distribution of energies with exponent β≈1.6–1.8, and waiting times with exponent τ≈2.0. These exponents are statistically indistinguishable across the two domains when evaluated with maximum‑likelihood estimation and Kolmogorov‑Smirnov tests.

The paper then demonstrates that the direct Omori law (an increase in the rate of small events immediately after a main shock) and the inverse Omori law (a gradual decay of activity before a main shock) are present in seizure recordings. After a large seizure, the rate of minor EEG spikes rises sharply and then decays with a power‑law exponent comparable to that observed after major earthquakes. Conversely, a decrease in spike activity precedes the main seizure, mirroring the inverse Omori pattern. This dual observation strongly supports the idea that stress redistribution in the Earth’s crust has an analogue in the redistribution of excitability across neuronal networks.

To probe the role of coupling, the authors manipulate the rodent model. At low coupling (high inhibition), the seizure‑like events obey a pure power‑law without any discernible characteristic scale. As inhibition is reduced, the network becomes more tightly coupled; the power‑law exponent modestly shifts, and a distinct peak emerges in the energy distribution, reflecting a characteristic scale (e.g., a dominant 0.5 Hz oscillation). This coexistence of scale‑free and scale‑specific statistics reproduces the “earthquake‑driven” hypothesis that a system can retain criticality while developing synchronized modes as it approaches a phase transition.

Complementary numerical simulations of a network of threshold oscillators (a Kuramoto‑type model with heterogeneous thresholds) reproduce the empirical findings. By varying the coupling constant, the model traverses three regimes: (i) weak coupling – pure power‑law statistics, (ii) intermediate coupling – power‑law plus a nascent characteristic peak, and (iii) strong coupling – dominant synchronized bursts with reduced power‑law tails. The simulated scaling exponents match the empirical β and τ values, reinforcing the claim that the observed universality arises from generic features of coupled threshold systems rather than domain‑specific mechanisms.

The discussion emphasizes the translational implications. Because the power‑law exponents are robust across individuals and species, a probabilistic forecasting framework based on these exponents could extend the prediction horizon beyond the few seconds currently achievable with threshold‑based methods. However, practical deployment would require individualized estimates of network coupling (e.g., via functional connectivity metrics) and continuous high‑resolution EEG monitoring to maintain statistical stationarity. The animal experiments also suggest that therapeutic strategies aimed at modulating coupling—through pharmacology, neuromodulation, or closed‑loop stimulation—might shift the system away from the synchronized regime, thereby reducing the likelihood of large seizures.

In conclusion, the study provides compelling evidence that the statistical physics of earthquakes and epileptic seizures are governed by the same universality class of coupled threshold oscillators. By demonstrating identical power‑law distributions, Omori‑type temporal clustering, and the emergence of characteristic scales with increasing coupling, the authors bridge geophysics and neuroscience, opening new avenues for seizure forecasting and for the broader application of universality concepts to biological catastrophes.