Preserved Intrinsic Neural Timescale Organization with Hierarchical Variation in Autism Spectrum Disorder

This study reveals that while the large-scale hierarchical organization of intrinsic neural timescales is preserved in autism spectrum disorder, individual deviations from this hierarchy—driven primarily by demographic factors like sex rather than diagnosis—modestly correlate with sensory traits characterized by reduced sensory registration.

The Gist

The Big Picture: The Brain's "Rhythm" and Autism

Imagine your brain isn't just a static computer, but a massive orchestra playing a complex piece of music. Every instrument (or brain region) has its own natural speed or "rhythm."

• Fast instruments: The drums and violins in the front row represent your sensory areas (sight, sound, touch). They react instantly to the world. They have a short rhythm.
• Slow instruments: The deep cellos and basses in the back represent your thinking areas (planning, memory, self-reflection). They hold onto information longer to make sense of the big picture. They have a long rhythm.

Scientists call this the Intrinsic Neural Timescale (INT). It's basically how long a specific part of the brain "remembers" a signal before moving on to the next one.

The Study's Question

For a long time, researchers thought that in people with Autism Spectrum Disorder (ASD), this orchestra was "out of tune" in specific, broken spots. Maybe the drums were too slow, or the violins were too fast.

This study asked a different question: Is the whole orchestra playing the same song, just at a slightly different volume or tempo? Or is the sheet music itself different?

The Findings: The Song is the Same, But the Tempo Shifts

The researchers looked at brain scans from 182 people (67 with autism, 115 without). Here is what they found, translated into everyday terms:

1. The Hierarchy is Intact (The Sheet Music is the Same)

The Analogy: Imagine a ladder. The bottom rungs are fast (sensory), and the top rungs are slow (thinking).
The Finding: In both autistic and non-autistic brains, the ladder looks exactly the same. The sensory areas are still fast, and the thinking areas are still slow. The brain's "map" of time is preserved. Autism does not break the fundamental structure of how the brain organizes time.

2. The "Long" Rhythms are a Bit Longer

The Analogy: Imagine the slow cellos at the back of the orchestra. In the autistic group, these cellos were playing slightly slower than usual.
The Finding: While the structure is the same, the parts of the brain that are supposed to be slow (the high-level thinking areas) were even slower in the autistic group. It wasn't a broken instrument; it was just a subtle shift in the tempo of the slowest notes.

3. It's Not About "Broken" Parts, It's About "Personal Style"

The Analogy: Think of the brain's rhythm as a standard uniform. Most people wear it perfectly. Some people wear it slightly loose, some slightly tight.
The Finding: The researchers realized that looking for "average" differences between groups wasn't telling the whole story. Instead, they looked at individual deviations.

• They found that how much a person's brain rhythm deviated from the "standard" pattern was linked to their sensory experiences.
• The Connection: People whose brains had more "wobbly" or unique rhythms (deviating from the standard ladder) tended to report specific sensory traits: they often felt like they missed sensory details (Low Registration) but didn't necessarily try to avoid them (Sensation Avoiding).

Why This Matters

The Old Way of Thinking:
"Autism is a broken brain where specific parts don't work right."

The New Way of Thinking (from this paper):
"Autism is a brain that follows the same general rules as everyone else, but with a unique personal tempo. The 'map' is correct, but the 'speed' of the slow-thinking areas is slightly different. This difference in speed might explain why some autistic people experience the world differently—perhaps needing more time to register a sound or a touch before their brain fully processes it."

The Takeaway Metaphor

Think of the brain as a river.

• In a typical brain, the water flows fast near the source (sensory input) and slows down as it reaches the wide, deep lake (thinking/abstract thought).
• In this study, they found that in autistic brains, the river still flows from fast to slow. The path is the same.
• However, the deep lake (the thinking part) is a little bit deeper and holds the water for a longer time.
• This "holding on" might be why some people with autism feel overwhelmed by sensory input (the water is moving too slowly to clear out new information) or why they sometimes miss subtle cues (the water is moving so slowly it takes a while to notice a new drop).

In short: The brain's architecture is healthy and organized, but the "clock speed" of the high-level thinking areas is slightly different, and this difference is deeply connected to how a person experiences their senses.

Technical Summary

1. Problem Statement

Autism Spectrum Disorder (ASD) is characterized by atypical sensory processing and difficulties in temporal integration. Previous neuroimaging studies have reported alterations in Intrinsic Neural Timescales (INTs)—the duration over which neural activity remains autocorrelated—in individuals with ASD. However, findings have been inconsistent, with some studies reporting shortened INTs and others reporting prolonged INTs in specific regions.

The central problem addressed by this study is whether these inconsistencies arise from:

1. Focal abnormalities: Discrete disruptions in specific brain regions.
2. System-level reorganization: A global shift or scaling of the entire cortical temporal hierarchy.

The authors hypothesize that ASD may not involve a collapse of the hierarchical organization itself, but rather a systematic modulation of the hierarchy (e.g., global shifts or scaling) and individual deviations from a canonical template that correlate with sensory traits.

2. Methodology

Participants and Data Acquisition:

• Cohort: 182 participants (67 ASD, 115 Typically Developed Controls [TDC]) from the Brain/MINDS Beyond project.
• Inclusion Criteria: Adults (18–48 years), completed 4 resting-state fMRI runs, full Adolescent/Adult Sensory Profile (AASP) questionnaires, and no structural brain anomalies.
• Scanning: 3T Siemens Skyra fit scanner using the Harmonization Protocol (HARP).
• Data Types: Resting-state fMRI (multiband EPI, TR=800ms), T1w/T2w (myelin maps), and Diffusion-Weighted Imaging (for Neurite Density Index, NDI).

Preprocessing and INT Calculation:

• Pipeline: Human Connectome Project (HCP) minimal preprocessing pipeline, including ICA-FIX denoising.
• INT Estimation: Calculated as the half-life of the autocorrelation function (ACF) of the resting-state time series for each cortical vertex.
• Aggregation: Vertex-wise INTs were averaged within Glasser's 360 cortical parcels and 12 large-scale functional networks.

Analytical Framework:

1. Hierarchy Validation: Correlated INTs with anatomical hierarchies (auditory, visual, somatosensory) and microstructural markers (T1w/T2w myelin content and NDI) to confirm the preservation of the sensory-to-transmodal gradient.
2. Group Comparison: Compared spatial distributions of INTs between ASD and TDC at vertex, parcel, and network levels.
3. Decomposition of Individual Variability:
   • Constructed a TDC-derived INT template.
   • Modeled individual INT profiles as a linear transformation of this template: $INT_{individual} = \alpha + \beta \times INT_{template} + \epsilon$.
   • $\alpha$ (Global Offset): Overall prolongation or shortening.
   • $\beta$ (Hierarchical Scaling): Strength of alignment to the template's gradient.
   • $\epsilon$ (Residual Deviation): Non-linear, distributed deviations from the template (measured as Root Mean Square, RMS).
4. Behavioral Correlation: Analyzed associations between the residual RMS ($\epsilon$) and sensory traits derived from AASP scores using Principal Component Analysis (PCA).

3. Key Contributions

• Systematic Decomposition: Introduced a novel method to separate global shifts ($\alpha$), hierarchical scaling ($\beta$), and distributed residual deviations ($\epsilon$) in neural timescales, moving beyond simple region-wise comparisons.
• Resolution of Inconsistencies: Demonstrated that while focal group differences are negligible, there is a hierarchy-dependent prolongation in ASD, where regions with naturally longer timescales show greater divergence.
• Linking Structure to Function: Confirmed that the INT hierarchy aligns with microstructural markers (myelin and neurite density) in both ASD and TDC, suggesting the structural basis of the hierarchy is intact.
• Sensory Trait Specificity: Identified that individual deviations from the canonical hierarchy (not absolute INT values) are specifically linked to sensory traits characterized by reduced sensory registration and lower sensory avoidance.

4. Key Results

A. Preservation of Cortical Hierarchy

• Sensory Systems: INTs in auditory, visual, and somatosensory systems followed the established anatomical hierarchy (short in primary areas, long in association areas) in both ASD and TDC.
• Microstructural Alignment: INTs showed significant negative correlations with myelin content and NDI in both groups, confirming that the temporal hierarchy is tightly coupled with structural gradients.
• Global Concordance: The spatial distribution of INTs across the whole brain was nearly identical between groups (Pearson's $r > 0.98$). No single region survived multiple comparison correction for group differences.

B. Hierarchy-Dependent Prolongation

• While no focal differences existed, a systematic trend emerged: Regions with longer INTs exhibited larger group differences (ASD > TDC).
• This indicates a "stretching" of the hierarchy where high-order association areas in ASD operate on even slower timescales compared to TDC, while primary sensory areas remain relatively unchanged.

C. Individual Deviations and Sensory Traits

• Demographics: Global offset and hierarchical scaling parameters were primarily driven by sex (females showed longer INTs and shallower slopes) rather than diagnosis.
• Residual Deviations: After accounting for global shifts and scaling, the residual RMS ($\epsilon$) did not differ between ASD and TDC groups.
• Sensory Correlation: The residual deviation was significantly positively correlated with PC3 of the AASP.
  • PC3 Profile: High "Low Registration" (difficulty noticing sensory stimuli) and low "Sensation Avoiding."
  • Interpretation: Individuals with greater deviations from the canonical temporal hierarchy (regardless of diagnosis) tend to exhibit sensory profiles where they fail to register stimuli, suggesting a need for longer temporal integration windows to accumulate sufficient sensory evidence.

5. Significance and Implications

• Reframing ASD Neurobiology: The study suggests that ASD is not defined by a broken temporal hierarchy but by systematic biases (prolongation in high-order areas) and idiosyncratic deviations from the normative template.
• Mechanism for Sensory Atypicalities: The link between residual deviations and "Low Registration" supports a computational model where sensory processing deficits arise from altered temporal integration windows across the cortical hierarchy, rather than isolated regional dysfunction.
• Methodological Advancement: The approach of modeling individual profiles against a canonical template allows researchers to distinguish between group-level shifts and continuous individual variability, offering a more nuanced view of neurodevelopmental heterogeneity.
• Clinical Relevance: The findings imply that sensory interventions might need to target the coordination of temporal processing across the hierarchy rather than focusing on specific sensory regions.

In conclusion, the paper establishes that the large-scale organization of cortical temporal dynamics is preserved in ASD, but subtle, distributed deviations from this hierarchy—particularly in how information is integrated over time—underlie the heterogeneity of sensory experiences in autism.