Tract-explainable and underexplained synchrony play complementary roles in the functional organization of the brain

This study introduces a framework that disentangles brain functional synchrony into tract-explainable and underexplained components, revealing that while tract-based connectivity supports global integration, non-tract-mediated synchrony driven by multiscale cortical features becomes increasingly prominent in higher-order regions to support modularity, individual variability, and behavioral relevance.

The Gist

The Big Idea: Two Types of Brain "Chat"

Imagine your brain is a massive, bustling city with millions of people (neurons) talking to each other. For a long time, scientists thought these conversations happened almost entirely because of roads (white matter tracts) connecting different neighborhoods. If two neighborhoods had a highway between them, they would talk a lot. If they didn't, they wouldn't.

However, this paper argues that the city is more complex than just roads. The researchers discovered that brain activity is actually a mix of two different types of "synchrony" (people talking in unison):

1. The "Road-Dependent" Chat (SynTE): This is communication that happens because the physical cables (roads) are there. It's like neighbors talking because they have a direct phone line.
2. The "Vibe-Dependent" Chat (SynTU): This is communication that happens without a direct road. It's like two people in different parts of the city starting to dance to the same music because they both love jazz, or because they are both influenced by the same wind blowing through the streets.

The paper's main goal was to separate these two types of chatter to see what each one actually does.

---

The Analogy: The Orchestra and the Radio

To understand how they did this, imagine a massive orchestra playing a symphony.

• The Score (Structural Wiring): The sheet music tells the violin section exactly when to play with the cello section. This is the physical road.
• The Radio (Non-tract Influences): But, imagine the whole orchestra is also listening to a radio broadcast of a conductor giving cues, or the musicians are all wearing the same uniform and feeling the same mood. They might start playing together even if they aren't looking at the same sheet music.

The researchers built a computer model (a "digital twin" of the brain) to act like a sound engineer. They asked: "How much of this music is coming from the sheet music (roads), and how much is coming from the radio vibe (non-roads)?"

They split the brain's activity into two separate tracks:

• Track 1 (SynTE): The part explained by the roads.
• Track 2 (SynTU): The part the roads can't explain (the "underexplained" part).

---

What They Found: Two Different Jobs

Once they separated the tracks, they discovered these two types of chatter do very different jobs:

1. The "Road" Track (SynTE) is the Glue

• What it does: It keeps the whole brain connected and stable. It's like the foundation of a building.
• Where it works best: In the "basic" parts of the brain, like the areas controlling your hands, feet, and eyes.
• Who it connects: It's very similar in everyone. Your "road" connections look a lot like mine because we all have similar wiring. It's the shared, stable backbone of the human brain.

2. The "Vibe" Track (SynTU) is the Spark

• What it does: It allows for flexibility, creativity, and individuality. It's like the jazz improvisation in the orchestra.
• Where it works best: In the "high-level" parts of the brain (the association cortex) responsible for thinking, planning, emotions, and social skills.
• Who it connects: This is where you are different from me. Your "vibe" connections are unique to your personality, your experiences, and your genes. It's the individual, flexible layer.

The "City Hierarchy" Discovery

The researchers found a fascinating pattern as they moved from the bottom of the brain to the top:

• At the bottom (Sensory/Motor): The "Roads" and the "Vibe" are tightly linked. The physical cables and the chemical/molecular signals work together perfectly to move your hand quickly.
• At the top (Thinking/Feeling): The "Roads" and the "Vibe" start to drift apart. The physical cables are still there, but the "Vibe" (chemical signals, gene expression, receptor types) takes over. This is why your ability to think about complex emotions or solve a math problem isn't just about having a big highway; it's about the unique chemical "flavor" of your brain cells.

Why This Matters for You (and Medicine)

This discovery is a game-changer for understanding mental health and individual differences.

• Why we are different: If you want to know why you are good at math and your friend is great at art, you shouldn't just look at the "roads" (which are mostly the same for everyone). You need to look at the "Vibe" track (SynTU). This track captures your unique personality and cognitive style.
• Detecting Disease: The paper tested this on people with Depression and Anxiety. They found that the "Vibe" track (SynTU) was much better at spotting the subtle changes in the brain caused by these illnesses than the standard "Road" track.
  • Analogy: If a city has a pothole, the "Road" map shows it. But if the city's mood changes (everyone is sad), the "Vibe" map shows it. Depression changes the brain's "mood" and chemical signals before it might even change the physical roads. SynTU catches this early.

The Takeaway

The brain isn't just a machine built on wires. It is a dynamic system where physical roads provide a stable, shared foundation, but chemical and molecular "vibes" provide the flexibility, individuality, and complexity that make us who we are.

By separating these two, scientists can now better understand why we all think differently and how to detect mental health issues earlier by listening to the "vibe" of the brain, not just the "roads."

Technical Summary

1. Problem Statement

Functional connectivity (FC), typically measured via resting-state fMRI, describes statistical dependencies between brain regions. While it is widely accepted that FC emerges from structural connectivity (SC)—the physical white matter tracts revealed by diffusion MRI—a significant "structure-function mismatch" persists.

• The Gap: SC explains FC well in primary sensorimotor regions but fails to account for functional coupling in higher-order transmodal regions (e.g., the Default Mode Network).
• The Challenge: Current methods treat FC as a monolithic entity. It is difficult to disentangle the specific contributions of tract-mediated interactions (directed by axonal fibers) from non-tract influences (such as cortical geometry, gene expression, receptor similarity, and volume transmission) because these factors are intrinsically entangled in empirical data.
• Goal: To develop a framework that decomposes empirical FC into two distinct, complementary components: one explainable by structural tracts and one that remains "underexplained" by them, thereby revealing their unique roles in brain organization.

2. Methodology

The authors introduced a large-scale brain modeling-based framework using a heterogeneous Dynamic Mean-Field (DMF) model.

• Model Architecture:
  • The model simulates local dynamics of each brain region using a nonlinear differential equation.
  • Heterogeneity: Regional parameters (self-feedback and baseline input) are constrained by macroscale gradients (myelin content and functional connectivity gradients) to capture biological variability.
  • Linearization: To enable efficient individual-level optimization, the model dynamics were analytically linearized around a fixed point.
• Decomposition Strategy:
  • Step 1 (Optimization): Model parameters were optimized to minimize the Mean Absolute Error (MAE) between simulated and empirical FC, assuming a diagonal background noise covariance matrix ($C_v$).
  • Step 2 (Solving for Residuals): Recognizing that SC alone cannot fully explain empirical FC, the authors allowed the background noise covariance matrix ($C_v$) to be non-diagonal. This non-diagonal $C_v$ captures intrinsic synchrony not driven by direct structural links.
  • Step 3 (Extraction):
    • SynTE (Tract-Explainable Synchrony): Simulated FC where $C_v$ is constrained to be diagonal (only tract-mediated coupling).
    • SynTU (Tract-Underexplained Synchrony): Simulated FC where structural connectivity is set to zero, but the optimized non-diagonal $C_v$ is retained. This isolates synchrony driven by non-tract factors.
• Validation:
  • Datasets: Validated across two large human cohorts (HCP and BANDA, total $N=1214$) and a marmoset dataset ($N=24$).
  • Metrics: Compared against structural connectivity (SC), microstructure profile covariance (MPC), receptor similarity (RS), gene expression (GS), and cortical geometry.
  • Network Analysis: Used graph theory (modularity, participation coefficient) and cortical gradients (diffusion map embedding) to characterize topological differences.
  • Behavioral/Clinical: Tested predictive power for latent behavioral dimensions (cognition, mental health, substance use) and disease classification (MDD, Anxiety) using Connectome-Based Predictive Modeling (CBPM).

3. Key Contributions

1. Novel Decomposition Framework: A mathematical approach to separate FC into SynTE (structurally constrained) and SynTU (non-tract mediated), moving beyond the binary view of structure vs. function.
2. Identification of a Hierarchical Axis: Discovery of a systematic "SynTE–SynTU Interaction Gradient" that aligns with the sensorimotor-to-association cortical hierarchy.
3. Biological Substrates: Linking SynTU to specific molecular and synaptic features (receptor diversity, synaptic development genes) rather than just white matter density.
4. Clinical Relevance: Demonstrating that SynTU is a superior biomarker for individual variability, complex behaviors, and psychiatric disorders compared to raw FC or SynTE.

4. Key Results

A. Distinct Biological Alignments

• SynTE is strongly correlated with Structural Connectivity (SC) ($r=0.49$) and supports global integration.
• SynTU is weakly correlated with SC ($r=0.25$) but strongly correlated with multiscale cortical similarity features: Microstructure Profile Covariance (MPC), Receptor Similarity (RS), and Gene Expression (GS). It also aligns with cortical geometry-driven synchrony.

B. Topological and Gradient Dissociation

• Network Properties: SynTU exhibits significantly higher modularity (functional segregation/specialization), while SynTE exhibits a higher participation coefficient (global integration).
• Cortical Hierarchy:
  • Sensorimotor Regions: High consistency between SynTE and SynTU; both rely heavily on structural wiring.
  • Association Regions (e.g., Default Mode Network): Strong dissociation. SynTU becomes dominant, showing greater individual variability and supporting higher-order cognition.
• Gradient: A principal component analysis revealed a gradient where SynTE dominance shifts to SynTU dominance along the sensorimotor-association axis. This gradient correlates with synaptic development, receptor diversity, and cortical expansion.

C. Individual Variability and Behavior

• Shared vs. Specific: SynTE captures the shared, stable anatomical backbone common across individuals. SynTU captures individual-specific functional idiosyncrasies.
• Behavioral Prediction:
  • SynTE performed poorly in predicting individual behaviors.
  • SynTU outperformed empirical FC in predicting personality-emotion, mental health, and illicit substance use.
  • Empirical FC was best for general cognition, but SynTU showed superior sensitivity to domain-specific cognitive factors.

D. Clinical Sensitivity

• Disease Classification: SynTU achieved the highest accuracy in distinguishing Major Depressive Disorder (MDD) and Anxiety Disorders from healthy controls.
• Gene Co-expression: SynTU networks in healthy controls showed the strongest similarity to normative depression-related gene co-expression networks. This similarity was significantly reduced in patient groups, indicating SynTU captures disease-related molecular perturbations.
• Atrophy Propagation: SynTU identified "epicenters" of cortical atrophy propagation that were more consistent with observed disease patterns than those derived from raw FC or SynTE.

5. Significance and Implications

• Mechanistic Insight: The study proposes that large-scale brain function arises from the interplay of two complementary systems: a rigid, tract-based scaffold (SynTE) ensuring stability and global integration, and a flexible, non-tract-mediated system (SynTU) enabling regional specialization, adaptability, and individuality.
• Neurobiological Interpretation: SynTU likely reflects volume transmission (neurotransmitters diffusing through extracellular fluid/CSF) and shared subcortical inputs, driven by molecular homophily (regions with similar receptors/genes synchronizing without direct wires).
• Precision Psychiatry: By filtering out the "noise" of stable anatomical constraints, SynTU reveals functional signatures that are highly sensitive to psychiatric conditions and individual behavioral traits. This suggests that future biomarkers for mental health should focus on non-tract-mediated synchrony.
• Cross-Species Conservation: The findings were replicated in marmosets, suggesting this dual-layer organization is a fundamental principle of primate brain evolution.

In summary, this paper provides a rigorous computational framework to deconstruct functional connectivity, revealing that the "unexplained" portion of FC is not noise, but a biologically distinct, clinically relevant, and hierarchically organized component essential for complex human cognition.