Early psychosis shows deviations in scaling behaviour within a critical regime

This study reveals that early psychosis is characterized not by a loss of critical-like brain dynamics, but by a systematic reorganization of scaling exponents within a preserved scale-invariant regime, as demonstrated by combining phenomenological renormalization group, power spectral density, and detrended fluctuation analysis on resting-state fMRI data.

The Gist

The Big Idea: The Brain as a Symphony Orchestra

Imagine your brain is a massive orchestra playing a piece of music while you are resting. In a healthy brain, this music isn't just random noise, nor is it a rigid, robotic march. Instead, it operates in a "sweet spot" called criticality.

Think of this sweet spot like a campfire. If the fire is too small (too ordered), it's quiet and doesn't spread. If it's a raging inferno (too chaotic), it burns out of control. But a healthy campfire has a perfect balance: sparks fly, embers glow, and the fire spreads in a way that is self-similar. A small spark looks like a tiny version of a big fire. This is what scientists call scale-invariance: the patterns look the same whether you zoom in on a single neuron or zoom out to look at the whole brain.

This paper asks: Does this "perfect campfire" still exist in people with early psychosis, or has the fire gone out?

The Experiment: Zooming In and Out

The researchers looked at brain scans (fMRI) from two groups:

1. Healthy Controls: People without psychosis.
2. Early Psychosis: People recently diagnosed with conditions like schizophrenia or bipolar disorder with psychotic features.

They used a special mathematical tool called the Phenomenological Renormalization Group (PRG). You can think of PRG as a "zoom lens" for the brain.

• Step 1: They looked at individual brain regions (like individual musicians).
• Step 2: They grouped the most connected regions together (like grouping the violin section).
• Step 3: They kept grouping them into larger and larger chunks (strings, then brass, then the whole orchestra).

By doing this, they could see if the "music" of the brain kept its special, self-similar pattern as they zoomed out. They also used other tools (PSD and DFA) to measure how long the brain's "echoes" lasted over time.

What They Found: The Fire is Still Burning, But the Wind Changed

1. The Pattern Still Exists
The most important finding is that the "campfire" didn't go out. Even in people with early psychosis, the brain still showed that special, self-similar scaling behavior. The brain wasn't broken or chaotic; it was still operating in that critical "sweet spot."

2. The "Wind" Changed Direction
However, while the fire was still burning, the way it burned had shifted. The researchers found systematic differences in the "scaling numbers" (exponents) between the healthy group and the psychosis group.

Here is the analogy for what changed:

• In Healthy Brains: The orchestra coordinates perfectly. When the violins start, the brass follows quickly, and the whole group moves together efficiently. The "silence" between notes and the "loudness" of the music follow a specific, balanced rule.
• In Early Psychosis Brains: The orchestra is still playing, but the coordination is slightly off.
  • Weaker Group Coordination: The brain seemed to have a harder time keeping the large groups of regions perfectly synchronized. It's as if the sections of the orchestra are a bit more independent of each other than they should be.
  • Stronger Persistence: However, once a pattern started, it seemed to "stick" longer. The "echoes" of brain activity lasted longer than in healthy brains. It's like a note that keeps ringing out a bit too long, making the music feel a bit more rigid or "stuck" in time.

The Takeaway: Reorganization, Not Collapse

The paper concludes that early psychosis is not a case of the brain losing its ability to organize itself. The brain is still using the same "critical" rules as a healthy brain.

Instead, it is a case of reorganization. Imagine a dance floor where everyone is still dancing to the same beat (the critical regime), but in the psychosis group, the dancers are taking slightly longer steps and holding their poses a bit longer before moving to the next step. The dance is still happening, but the style has shifted.

Why This Matters (According to the Paper)

The authors suggest that looking at these "scaling rules" gives us a new way to understand the brain. Instead of saying "the brain is broken," we can say "the brain's large-scale dynamics have reorganized."

They also note that these changes might be linked to how the brain balances excitement and inhibition (like the volume knobs on the orchestra). If the "volume" of the inhibitory signals is turned down slightly, it could explain why the brain's patterns stick around longer (persistence) but don't coordinate as tightly across the whole system.

In short: The brain in early psychosis isn't a broken machine; it's a machine that has been tuned to a slightly different, yet still functional, frequency. The "music" is still there, but the rhythm and the way the instruments blend together have changed.

Technical Summary

Technical Summary: Early psychosis shows deviations in scaling behaviour within a critical regime

Problem Statement
Large-scale brain activity is increasingly understood to operate near a critical regime, characterized by scale-invariant dynamics, long-range correlations, and efficient information processing. While altered criticality-related measures have been reported in psychiatric disorders such as schizophrenia and psychosis, previous findings remain methodologically fragmented. Studies rely on disparate observables (e.g., neuronal avalanches, power spectral density, detrended fluctuation analysis) and modalities, making it unclear whether these different measures capture a common alteration in large-scale brain dynamics or unrelated aspects of neural activity. Furthermore, it remains unknown whether early psychosis is characterized by a complete loss of critical-like dynamics or a reorganization within a preserved scaling regime.

Methodology
The study analyzed resting-state fMRI (rs-fMRI) data from the Human Connectome Project for Early Psychosis (HCP-EP). The final sample included 59 individuals with early psychosis (diagnosed within 5 years of onset) and 25 healthy controls.

The authors employed a multi-faceted analytical framework to characterize collective dynamics across spatial and temporal scales:

1. Phenomenological Renormalization Group (PRG): Inspired by statistical physics, this approach iteratively coarse-grains the system by clustering maximally correlated brain regions. The study tracked four static and dynamic observables across increasing cluster sizes ($K$):
   • Silence Probability ($\beta$): The scaling of the probability of silence ($-\ln P_0 \propto K^\beta$).
   • Variance Scaling ($\alpha$): The growth of variance in non-normalized coarse-grained variables ($Var(K) \propto K^\alpha$).
   • Eigenspectrum Scaling ($\mu$): The power-law decay of eigenvalues in the covariance matrices of clusters ($\lambda \propto (rank/K)^{-\mu}$).
   • Dynamical Scaling ($z$): The scaling of autocorrelation times with cluster size ($\tau_c \propto K^z$).
2. Temporal Scaling Analyses: To complement PRG, the authors analyzed the global fMRI signal using:
   • Power Spectral Density (PSD): Estimating the power-law exponent ($\beta_{PSD}$) of the BOLD signal ($P(f) \propto f^{-\beta_{PSD}}$).
   • Detrended Fluctuation Analysis (DFA): Estimating the scaling exponent ($\alpha_{DFA}$) to quantify long-range temporal correlations.
3. Robustness Checks: Analyses were repeated across multiple binarization thresholds (1.0 to 2.0 SD) and alternative event definitions (upward crossings, suprathreshold peaks) to ensure results were not artifacts of preprocessing choices. Null models involving phase-shuffling were used to verify that observed statistics arose from intrinsic spatiotemporal organization rather than random features.

Key Results

• Preservation of Critical-like Organization: Both healthy controls and early psychosis patients exhibited non-trivial scaling behavior across all observables. In healthy controls, the data showed signatures consistent with a near-critical regime: non-Gaussian fixed-point distributions, intermediate variance scaling ($\alpha \approx 1.61$), and scale-free organization of eigenmodes.
• Systematic Shifts in Scaling Exponents: While the overall phenomenology of scale invariance was preserved in the psychosis group, systematic shifts in scaling exponents were observed compared to controls:
  • Static Measures: The silence probability exponent ($\beta$) was significantly increased in patients, while the variance exponent ($\alpha$) was significantly reduced. The eigenspectrum exponent ($\mu$) was also decreased.
  • Dynamic Measures: The dynamical scaling exponent ($z$) was significantly increased in patients.
  • Global Signal: Patients showed significantly larger DFA exponents ($\alpha_{DFA}$) and a trend toward larger PSD exponents ($\beta_{PSD}$), indicating stronger long-range temporal correlations and enhanced temporal persistence.
• Robustness: These group differences remained consistent across different binarization thresholds and event definitions, particularly for $\alpha$ and $z$.
• Null Model Validation: When cross-correlations were destroyed via phase-shuffling, the non-trivial scaling behaviors vanished, confirming that the observed signatures depend on the intrinsic spatiotemporal organization of the BOLD signals.

Significance and Claims
The paper claims that early psychosis is not characterized by a simple loss of critical-like dynamics or a breakdown of scale invariance. Instead, the findings suggest a reorganization of collective dynamics within a preserved scaling regime. The dissociation between the observed shifts—specifically, weaker large-scale coordination (indicated by changes in $\alpha$ and $\beta$) alongside stronger temporal persistence (indicated by changes in $z$ and $\alpha_{DFA}$)—implies a redistribution between collective integration and temporal rigidity.

The study positions the combination of coarse-graining approaches (PRG) with temporal scaling analyses (PSD, DFA) as a principled framework for studying large-scale brain dynamics in psychiatric disorders. By demonstrating that different scaling observables converge on a coherent characterization of altered dynamics, the authors argue that scaling measures can serve as quantitative markers for the altered organization of large-scale dynamical states in psychosis, potentially bridging microscale circuit theories (e.g., excitation-inhibition balance) with macroscale functional alterations. The authors explicitly note that these findings do not permit direct inference about specific neurobiological mechanisms but offer a systems-level signature of disturbed collective dynamics.