Inter- and intra-individual variability in structure-function coupling in human brain

This study reveals that while alpha power correlates positively with microstructural features like myelin and iron across brain regions, it correlates negatively with these same features across individuals, suggesting that excitatory and inhibitory mechanisms drive intra- versus inter-individual variability in brain function, respectively.

The Gist

Imagine your brain is a massive, bustling city. This paper is like a detective story where researchers try to figure out how the physical layout of the city's buildings (the brain's structure) affects the traffic patterns and noise levels (the brain's electrical activity).

Specifically, they looked at the brain's "alpha rhythm"—a steady, humming electrical wave that happens when you are awake but relaxed (like daydreaming). They wanted to know: Does the thickness of the city's walls or the amount of wiring (myelin) in a neighborhood determine how loud that hum is?

Here is the breakdown of their discovery, using some creative analogies:

1. The Two Different Rules of the City

The researchers found a fascinating twist: The rules change depending on whether you are looking at different neighborhoods or different people.

Rule A: Comparing Neighborhoods (Intra-individual)

Imagine you are a city planner looking at a map of one person's brain.

• The Observation: In neighborhoods where the buildings are taller (specifically, a specific layer of the brain called "Layer IV" is thicker) and the wiring is denser, the "hum" (alpha power) is louder.
• The Analogy: Think of a concert hall. If you have a bigger room with more seats (more neurons) and better acoustics (more myelin/wiring), you can get a louder, richer sound.
• The Takeaway: In a single brain, areas with more "hardware" (cells and wiring) produce a stronger signal. It's a "more is more" situation.

Rule B: Comparing People (Inter-individual)

Now, imagine you are comparing two different people, Person A and Person B.

• The Observation: This is where it gets weird. For Person A, if their brain has more wiring and thicker layers in the middle of the cortex, their "hum" is actually quieter than Person B's.
• The Analogy: Imagine two different orchestras.
  • Orchestra A has a huge, thick section of violins (lots of excitatory neurons). They play loud and proud.
  • Orchestra B has a conductor who is very strict and tells the violins to play softer to keep the peace (stronger inhibitory control). Even if Orchestra B has the same number of instruments, the strictness of the conductor makes the overall volume lower.
• The Takeaway: When comparing different people, the difference isn't just about how many instruments they have; it's about how much they are being "reined in" by the brain's internal brakes (inhibition).

2. The "Why" Behind the Twist

The researchers used computer simulations to figure out why these two rules are opposite.

• The Excitatory Engine: Think of the brain's "excitatory" neurons as the gas pedal. When you have more of them (thicker layers), the brain revs up, and the alpha rhythm gets louder. This explains the Neighborhood rule.
• The Inhibitory Brake: Think of "inhibitory" neurons as the brake pedal. When you have more of these, or when they are more active, they slow the engine down. This explains the Person rule.

The Big Reveal:

• Between regions: The differences are mostly about how many "gas pedals" (excitatory cells) a specific area has. More gas = louder hum.
• Between people: The differences are mostly about how hard the "brakes" (inhibitory cells) are being pressed. Stronger brakes = quieter hum.

3. Why This Matters

For a long time, scientists thought that if you saw a strong brain signal, it just meant "more brain cells." This paper says, "Not so fast!"

• If you look at one person's brain map, a strong signal means a busy, well-wired area.
• If you look at two different people, a strong signal might mean one person has a "looser" brain with less braking power, while a quieter signal might mean a person has a very efficient, well-regulated brain with strong brakes.

Summary in a Nutshell

Think of the brain's alpha rhythm like the volume on a radio.

• Across the brain (different spots): Turning up the volume is like adding more speakers (more cells/wiring).
• Across people (different owners): Turning down the volume isn't because they have fewer speakers; it's because they have a better volume control knob (inhibition) that keeps the noise down.

This study is a crucial step in understanding that our brains aren't just static maps; they are dynamic systems where the balance between "gas" (excitation) and "brakes" (inhibition) creates the unique electrical signature of every single human being.

Technical Summary

1. Problem Statement

While macroscopic functional recordings (EEG/MEG) are widely used to link brain dynamics to cognition, the mechanistic link between these signals and the underlying cortical microstructure (cytoarchitecture and myeloarchitecture) remains poorly understood. Previous research has been limited by:

• Inter-laminar studies: Often restricted to animal models or small cortical patches, limiting generalizability.
• Inter-regional studies: Typically correlating averaged metrics across different individuals for structural and functional data, preventing subject-specific inference.
• Inter-individual studies: Often lacking the spatial resolution to resolve fine-grained laminar structures.

The core problem addressed is the lack of a unified framework to understand how cortical microstructure (specifically layer thickness, cell density, and myelin/iron content) shapes macroscopic electrophysiological signals (like alpha rhythms) both within an individual (across brain regions) and between different individuals.

2. Methodology

The study employed a multimodal approach integrating four distinct datasets and two analytical perspectives (across regions vs. across participants).

Datasets:

1. EEG (LEMON): 209 participants (resting-state, eyes open/closed).
2. MEG (Cam-CAN): 507 participants (resting-state, eyes closed).
3. Post-mortem Histology (BigBrain): Ultrahigh-resolution (20µm) 3D reconstruction of a human brain for cytoarchitectonic variables (layer thickness, staining profiles).
4. High-Resolution MRI (7T):
   • Dataset A: 10 participants (McColgan et al., 2021) for cross-regional structural analysis.
   • Dataset B (Multimodal): 31 participants with both resting-state MEG and 7T MRI (Multiparametric Maps - MPMs) to enable within-subject analysis.

Variables & Processing:

• Electrophysiology: Derived power and frequency for theta, alpha, and beta bands, 1/f slope, and alpha rhythm shape (nonsinusoidality, baseline shift).
• Microstructure:
  • Cytoarchitecture: Layer thickness (I–VI), supragranular/infragranular sums, and staining profile statistics (mean, skewness) from BigBrain.
  • Myeloarchitecture: Myelin and iron estimates (R1 and R2* maps) at superficial, mid-cortical, and deep depths. Factor Analysis (FA) was used to reduce dimensionality and decorrelate depth surfaces.
• Statistical Approach:
  • Across Regions: Linear mixed-effects models correlating averaged functional and structural data across vertices/regions. Significance tested via "spin" permutation tests (preserving spatial autocorrelation) and bootstrapping.
  • Across Participants: Vertex-wise and region-wise regression of functional vs. structural data within the same 31 participants, controlling for age, brain volume, skull thickness, and curvature.
  • Computational Modeling: Wilson-Cowan neural mass models within The Virtual Brain framework to simulate how excitatory (pyramidal) vs. inhibitory (interneuron) population dynamics affect alpha power.

3. Key Contributions

• Discovery of a Sign Reversal: The study is the first to demonstrate that the correlation between cortical microstructure (specifically mid-cortical myelin/iron) and alpha power has opposite signs depending on the level of analysis:
  • Positive correlation across brain regions (Intra-individual).
  • Negative correlation across participants (Inter-individual).
• Integration of Modalities: Successfully linked histological ground truth (BigBrain), in-vivo high-resolution MRI (7T MPMs), and large-scale electrophysiology (EEG/MEG) in a single study design.
• Mechanistic Hypothesis: Proposed that intra-individual variability is driven by excitatory factors (total neuron count), while inter-individual variability is driven by inhibitory modulation.

4. Key Results

A. Selection of Variables
Only alpha power, low beta power, and high beta power showed sufficient topographic similarity (correlation > 0.6) between EEG and MEG datasets to be used for cross-modal analysis. Alpha power was the primary focus due to its strong relationship with microstructure.

B. Across-Region Analysis (Intra-individual)

• Alpha Power vs. Layer IV: Alpha power showed a strong positive correlation with the thickness of Layer IV (granular layer).
• Alpha Power vs. Myelin/Iron: Alpha power was positively correlated with mid-cortical R2* (a proxy for myelin and iron).
• Mediation: The correlation between Layer IV thickness and alpha power was almost entirely mediated by the "staining profile mean" (total cell count).
• Interpretation: Regions with thicker Layer IV, higher cell density, and more myelin/iron exhibit stronger alpha rhythms. This suggests alpha power scales with the total number of participating neurons (excitatory drive) in a region.

C. Across-Participant Analysis (Inter-individual)

• Alpha Power vs. Myelin/Iron: In contrast to the regional analysis, participants with higher mid-cortical R2* (myelin/iron) exhibited lower alpha power.
• Spatial Specificity: This negative correlation was most pronounced in somatomotor and frontal regions (pre-central, post-central, superior frontal gyri).
• Interpretation: Higher inter-individual myelin content is associated with reduced alpha amplitude, suggesting a different underlying mechanism than the regional one.

D. Computational Modeling

• Excitatory Drive: Increasing excitatory-to-excitatory coupling ($C_{ee}$) and reactivity increased alpha power. This aligns with the positive regional correlation (more cells = more excitatory drive = more alpha).
• Inhibitory Drive: Increasing the external input to the inhibitory population ($Q$) decreased alpha power. This aligns with the negative inter-individual correlation.
• Conclusion: The sign reversal is explained by the hypothesis that inter-individual differences in alpha power are dominated by variations in inhibitory interneuron activity (or tangential fibers associated with inhibition), whereas regional differences are dominated by excitatory pyramidal neuron density.

5. Significance

• Resolving a Paradox: The study resolves the apparent contradiction in literature regarding structure-function coupling by showing that the direction of correlation depends on whether one looks at spatial variations within a brain or biological variations between brains.
• Biological Mechanism: It provides a biophysical explanation for alpha rhythm generation:
  • Regional differences reflect the "hardware" capacity (number of excitatory neurons).
  • Individual differences reflect the "software" regulation (strength of inhibitory control).
• Clinical Implications: Understanding that inter-individual variability in alpha power may be driven by inhibitory tone (GABAergic systems) rather than just structural density could refine biomarkers for neurological and psychiatric disorders where inhibition is disrupted (e.g., epilepsy, schizophrenia, anxiety).
• Methodological Advancement: Demonstrates the utility of combining 7T MRI, histology, and electrophysiology to move beyond gross anatomical correlations toward laminar-specific functional mechanisms.