Intrinsic timescale develops hierarchically from sensorimotor to association cortex in youth and stabilizes in adulthood
Golia ShafieiiD, Joëlle BagautdinovaiD, Valerie J. SydnoriD, Dani S. BassettiD, Deanna M. BarchiD, Matthew CieslakiD, Yong Fan AiD, Elizabeth A. FlookiD, Alexandre R. FrancoiD, Gregory KiariD, Audrey C. LuoiD, Michael Peter MilhamiD, Linden ParkesiD, Taylor SaloiD, Leah H. SomervilleiD, Tien T. TongiD, Russell Taki ShinoharaiD, Theodore Daniel SatterthwaiteiD
Abstract
Intrinsic timescale is a commonly used measure of spontaneous neural dynamics that quantifies the temporal window of processing of neuronal populations. Intrinsic timescale displays a hierarchical cortical organization across multiple species and imaging modalities, with shorter timescales in sensorimotor cortex compared to association cortex. However, less is known about how intrinsic timescale evolves during human brain development and whether its cortical maturation patterns generalize to independent developmental samples. Here we estimate the intrinsic timescale in two independent cross-sectional datasets of youth (HCPD: n=565; HBN: n=729; age range 8-22 years) and investigate its neurodevelopmental patterns. We find that developmental patterns in the intrinsic timescale follow a hierarchical pattern that recapitulates an axis spanning sensorimotor to association cortices (S-A axis). Our analysis of an independent healthy young adult dataset (HCPYA: n=973, age range 22-37 years) underscores the specificity of these developmental findings, suggesting that the intrinsic timescale develops along the S-A axis in youth and stabilizes in adulthood. Together, these results reveal convergence between major axes of cortical organization and development, highlighting intrinsic timescale as a principled marker of hierarchical brain maturation in youth.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 1
- FWCI
- 4.13
- Field percentile
- top 5%
- Organism
- Assay
- Research Focus
- Approach
In-depth summary
Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 2 (17 Sept 2026) · CC BY 4.0
Resting-state fMRI intrinsic timescale lengthens with age in association, not sensorimotor, cortex during ages 8-22, following the sensorimotor-association axis in two cohorts; it is largely age-stable in young adults.
Intrinsic timescale is shorter in sensorimotor than association cortex in adults, but its development across the whole cortex in youth has been little characterized with fMRI; earlier developmental work used EEG or electrocorticography (ECoG), reporting overall decreases with age, or studied only the hippocampus or neonates. This work maps cortex-wide fMRI timescale development in two independent youth cohorts and checks specificity in a young-adult sample.
Key results
ShownIn the Human Connectome Project-Development (HCPD) cohort, cortex-wide mean timescale rises with age from 8 to 22 years. The rise comes from association regions while sensorimotor regions stay stable, so regional age effects follow the sensorimotor-association (S-A) axis.
Whole-brain generalized additive model (GAM) partial R^2 = 0.047, p_anova = 1.25 x 10^-7 (n=565; sex and mean framewise displacement as covariates); regional age effects vs S-A axis rank rs = 0.41, p_spin = 0.001; a within-subject S-A differentiation index also rose with age (partial R^2 = 0.025, p_anova = 0.0007; direct t-value version 0.026, p_anova = 9.36 x 10^-5). · Neurodevelopmental patterns in intrinsic…; Generalized Additive Model (GAM); Sensorimotor-association timescale…; Figure legends
ShownThe pattern replicates in the independent Healthy Brain Network (HBN) cohort, but with weaker S-A alignment and fewer significant regions than HCPD. In both cohorts, the S-A alignment of age-related change peaks in early-to-mid adolescence.
HBN (n=729) whole-brain partial R^2 = 0.037, p_anova = 1.13 x 10^-7; age effects vs S-A axis rs = 0.19, p_spin = 0.02; HCPD vs HBN age-effect maps rs = 0.30, p_spin = 0.0002; regions significant after false discovery rate (FDR) correction 46.5% (HCPD) vs 22.5% (HBN), overlap Dice = 0.42, p_spin = 0.0008; peak S-A alignment at 14.6 years (HCPD) and 12.8 years (HBN). · Findings replicate in an independent…; Age of maximum alignment between age-related…; Figure legends
ShownIn Human Connectome Project-Young Adults (HCPYA) participants aged 22-37, the mean timescale map still follows the S-A axis. Age effects, however, are small, mostly non-significant and not organized along the axis.
HCPYA (n=973): mean map vs S-A axis rs = 0.4, p_spin = 0.01; association vs sensorimotor t = 7.5, p_spin = 0.02; whole-brain partial R^2 = −0.004, p_anova = 0.04; 11.7% of regions with significant age effects; age effects vs S-A axis rs = −0.1, p_spin = 0.58. · Timescale remains relatively stable in adulthood; Figure legends
ShownIn both youth cohorts, average timescale is longer in association than in sensorimotor cortex, and the two group maps agree. However, the map's correlation with the continuous S-A axis ranking is significant only in HCPD.
Association vs sensorimotor: HCPD t = 7.8, p_spin = 0.01; HBN t = 6.6, p_spin = 0.04 (two-tailed). HCPD vs HBN mean maps rs = 0.76, p_spin = 0.0002. Mean map vs S-A axis: HCPD rs = 0.37, p_spin = 0.03; HBN rs = 0.27, p_spin = 0.12. · Intrinsic timescale is longer in association…; Figure legends
SuggestedAge effects held across covariate, atlas, site and subsample checks and two other full-autocorrelation definitions. Lag-1 temporal autocorrelation (lag-1 TA) gave a different mean map and age effects without significant S-A alignment, and the main measure's age effects rest mostly on long-lag autocorrelation.
HCPD Schaefer-200: whole-brain partial R^2 = 0.041, p_anova = 9.86 x 10^-7; results consistent with estimated total intracranial volume as covariate, CovBat-GAM site harmonization and two N=600 HBN subsets. HCPD lag-1 TA: whole-brain partial R^2 = 0.008, p_anova = 0.03; 28.2% of regions significant; age effects not significantly related to S-A axis but correlated with the main measure's age effects. · Sensitivity analysis; Intrinsic timescale estimated using lag-1…; Short- and long-lag contributions to intrinsic…; Figure legends
Methods, data and reproducibility
For each participant and Schaefer-400 parcel, intrinsic timescale was the sum of positive autocorrelation function (ACF) values up to the first zero crossing, multiplied by TR, computed from 0.01-0.08 Hz band-passed resting-state fMRI. Age effects were fit with GAMs (age smooth, k=3; sex and mean framewise displacement covariates) and compared with the S-A axis using spin tests with 10,000 rotations.
- Human Connectome Project-Development (HCPD) resting-state fMRI; discovery cohort of typically developing youth aged 8-22 from four US sites (3T Siemens Prisma, TR 800 ms, 26 min per participant) · human · n = 565 (302 female, 263 male)
- Healthy Brain Network (HBN) resting-state fMRI; replication cohort of healthy and help-seeking youth aged 8-22 from three sites (Staten Island mobile-scanner data excluded; TR 800 ms) · human · n = 729 (279 female, 418 male, 32 other)
- Human Connectome Project-Young Adults (HCPYA) resting-state fMRI; young adults aged 22-37 scanned at Washington University (TR 720 ms, about 1 hour per participant) · human · n = 973 (523 female, 450 male)
- Ready-to-plot data and scripts for main and supplementary figures (Zenodo record) · human · 10.5281/zenodo.22799416
- Methods & tools
- fMRIPrep 22.0.2, XCP-D, HCP Minimal Preprocessing Pipelines, FreeSurfer 7.2.0, ANTs 2.3.3, Connectome Workbench, Schaefer atlas (400 and 200 parcels), statsmodels (tsa), mgcv, neuromaps, ComBatFamily (CovBat-GAM), spatiotemporal
- Data & accessions
- https://doi.org/10.5281/zenodo.22799416https://www.humanconnectome.org/study/hcp-young-adulthttps://www.humanconnectome.org/study/hcp-lifespan-developmenthttps://fcon_1000.projects.nitrc.org/indi/cmi_healthy_brain_network/
- Software & environment
- fMRIPrep 22.0.2 (Nipype 1.8.5), ANTs 2.3.3, FSL 6.0.5.1, FreeSurfer 7.2.0, XCP-D v0.3.0 (youth cohorts) and v0.9.1 (HCPYA), Scikit-Learn 1.1.3, R 4.2.2 with mgcv and ComBatFamily.
- Reproducibility gaps
- The code link labelled GitHub is a github.io page rather than a repository address; no container, dependency file or compute requirements are described.
Limitations (6)
- The design is cross-sectional, so age differences are between people; within-person trajectories, and whether they predict later cognition or mental health, need longitudinal data. · stated by the authors
- The youth range of 8-22 years omits infancy and early childhood, and the adult HCPYA cohort differs in acquisition and cohort characteristics, so it tests specificity rather than extending the youth trajectories. · stated by the authors
- fMRI sampling (TR 0.8 s) and the 0.01-0.08 Hz band limit the measure to slow BOLD persistence, so faster dynamics are missed. The authors caution that the association-cortex increase is not a global slowing, given EEG and ECoG reports of decreases with age. · stated by the authors
- Only cortical parcels were analysed; subcortical structures and finer spatial gradients remain unexamined. · stated by the authors
- In HBN the motion threshold removed 831 of 1650 participants who had passed T1 quality control, so the replication sample is a heavily filtered subset of those scanned. · noted in this summary
- Comparisons with alternative metrics (lag-1 TA, exponential fit, zero crossing, short/long-lag split) were run only in HCPD, and several robustness checks are reported as consistent without effect sizes in the text. · noted in this summary
Who should read:Read this if you study functional brain development or fMRI temporal dynamics and want replicated maps of how autocorrelation-based intrinsic timescale changes along the sensorimotor-association axis from age 8 to 22.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Task-evoked flexibility of intrinsic neural timescales is hierarchically organized, behaviorally relevant, and refined across two developmental stages
Linden Parkes, Valerie J. Sydnor, … Bart Larsen·bioRxiv·Oct 22, 2026
- Intrinsic Activity Develops Along a Sensorimotor-Association Cortical Axis in Youth
Valerie J. Sydnor, Bart Larsen, … Theodore Daniel Satterthwaite·bioRxiv·Aug 15, 2022
- Developmental relationships between the human alpha rhythm and intrinsic neural timescales are dependent on neural hierarchy
Jesse T. Miles, Kurt E. Weaver, … Jeffrey G. Ojemann·Journal of NeurophysiologyPeer-reviewed·Dec 13, 2025
- Spatiotemporal patterns in cortical development: Age, puberty, and individual variability from 9 to 13 years of age
Katherine L. Bottenhorn, Jordan D. Corbett, … Megan M. Herting·Journal of NeurosciencePeer-reviewed·Aug 11, 2025
- Frequency-specific maturation of cortical organization and social-cognitive links from childhood to adolescence
Zhu-Qing Gong, Elizabeth Jefferies, Xi‐Nian Zuo·bioRxiv·Sep 1, 2025
- Distinct lifespan trajectories of cortical intrinsic timescales and neural oscillations
Xiaobo Liu, Brendan Harris, … Tengteng Fan·bioRxiv·Nov 6, 2026