Largest brain imaging study shows autism and ADHD follow opposite developmental paths
A January 2026 study from the University of Minnesota analysed brain imaging data from 9,647 individuals aged 5–64 years, including 1,533 people with ADHD, 1,080 autistic people, and 7,034 controls. Data came from six large-scale datasets across 67 acquisition sites, all processed using standardised pipelines and harmonised to account for site-related variability.
The research examined cortical thickness — the measurement of grey matter depth in the brain’s outer layer — and cortical curvature, the inward folds (sulci) and outward folds (gyri) that shape brain surface complexity. These structural features reflect how brains develop, mature, and specialise during childhood and adolescence.
Findings revealed distinct neuroanatomical signatures for autism and ADHD. Autistic individuals exhibited regionally thinner cortex and curvature alterations particularly in the Cingulo-Opercular network, which supports cognitive control and switching between tasks. ADHD individuals displayed regionally thicker cortex, particularly in the default mode network and somatomotor networks, alongside different curvature patterns.
Control participants showed intermediate cortical thickness patterns between autism and ADHD groups. This positioning suggests autism and ADHD may represent diverging extremes of cortical maturation rather than simple deviations from a single “normal” developmental trajectory.
The analysis used the Human Connectome Project atlas, dividing the cortex into 360 regions and mapping them onto large-scale functional networks including default mode, frontoparietal, somatomotor, language, and visual networks. This allowed researchers to identify which brain systems showed the most pronounced differences across diagnostic groups.
Previous neuroimaging studies suffered from small sample sizes, variable analytical approaches, and lack of harmonisation across different scanning sites. Lower-resolution brain atlases potentially obscured subtle, region-specific effects. This study addressed these limitations through standardised processing, large sample size, and statistical correction for site-related batch effects.
Autistic brains show accelerated cortical thinning in language and attention networks
Autistic participants showed distributed cortical thinning across multiple large-scale networks compared to both control and ADHD groups. Thinning appeared in the Language network, Frontoparietal network supporting executive control, Cingulo-Opercular network involved in task switching, Somatomotor network controlling movement, Default Mode Network linked to self-referential thought, and Visual networks processing visual information.
Anatomically, these regions locate within dorsolateral and premotor prefrontal cortices controlling planning and action preparation, mid-cingulate and supplementary motor areas coordinating complex movements, posterior cingulate involved in memory and spatial navigation, superior parietal and occipital cortices processing visual and spatial information, and primary sensorimotor areas directly controlling movement and sensation.
Thinning in occipital regions aligns with prior reports of altered visual integration in autism. The pattern indicates widespread cortical thinning across networks supporting language processing, attention control, sensorimotor function, and visual perception.
Previous longitudinal evidence suggests autism involves atypical developmental trajectory characterised by early cortical overgrowth during childhood, followed by accelerated thinning during adolescence, with partial stabilisation in adulthood. This accelerated refinement pattern contrasts with typical developmental timelines where cortical thinning proceeds more gradually through adolescence.
Cortical thinning traditionally interpreted as tissue loss may instead reflect increased intracortical myelination — the insulation of nerve fibres that speeds signal transmission. Myelination shortens T1 relaxation times in MRI scanning, reducing apparent grey matter thickness without actual neuronal loss. This means structural differences may represent accelerated specialisation rather than deficit.
Curvature findings showed autistic individuals had increased sulcal depth in the left PFcm sulcus within the Cingulo-Opercular network. This region relates to language processing and sensory-motor function, positioned within posterior opercular cortex. The deeper fold suggests altered mechanical forces during development — potentially atypical surface area expansion, dendritic arborisation, or local axonal tension shaping cortical architecture.
ADHD brains show delayed cortical thickening in self-regulation and decision-making regions
ADHD participants showed cortical thickening primarily within the Default Mode Network, including right orbitofrontal cortex, left ventromedial prefrontal cortex, and left anterior rostral prefrontal cortex. These regions integrate sensory information, process reward signals, and shape decision-making based on emotional and motivational inputs.
The Default Mode Network activates during internally-focused thought — mind-wandering, autobiographical memory, future planning, self-referential processing. DMN dysfunction links to impaired self-regulation and executive control difficulties characteristic of ADHD clinical presentations. Thicker cortex in these regions suggests prolonged developmental period rather than arrested development.
Evidence characterises ADHD as delayed cortical maturation potentially due to slower synaptic pruning or microstructural maturation. Synaptic pruning eliminates unnecessary neural connections during adolescence, refining brain networks for efficient adult function. ADHD brains may undergo this refinement process more gradually, maintaining relatively thicker cortex into later developmental stages.
Previous studies reported increased cortical thickness in ADHD during adolescence within somatosensory cortex, anterior cingulate, anterior insula, pre-supplementary motor area, and occipital cortices, with fewer changes observed in adulthood. This suggests developmental delay rather than permanent structural difference — maturation proceeding along similar pathway but offset in timing.
Curvature analyses revealed ADHD-specific alterations within the somatomotor network, most prominently shallower left area 3a sulcus relative to both autism and controls, alongside deeper left Ig sulcus and shallower right 24dd sulcus. Within the Default Mode Network, ADHD individuals showed more prominent right area 10v gyrus compared to other groups.
Shallower sulci in ADHD may reflect reduced cortical tension or delayed mechanical folding, aligning with broader evidence of maturational lag in somatomotor regions. These curvature differences suggest not just timing variations but potentially different mechanical processes shaping cortical architecture during development.
Shared vulnerability in late-developing brain networks with divergent maturation timing in both autism and ADHD
Both autism and ADHD showed significant cortical thickness differences in the Fronto-Parietal Network, particularly left POS2 and left area 7Pm regions. Autism showed thinnest cortex in these areas, ADHD showed thickest, with controls intermediate. This pattern suggests transdiagnostic locus of structural variation — shared vulnerability expressed through divergent developmental paths.
The left POS2 functions as integrative hub engaged in diverse cognitive operations. The left area 7Pm implicates in complex cognitive processing and executive control. Both regions belong to association cortex — evolutionarily expanded brain areas underlying cognition, social behaviour, and self-regulation that mature later than primary sensory regions.
Association cortices remain plastic well into young adulthood due to delayed inhibitory maturation and extended synaptic remodelling. These prolonged developmental windows make them particularly vulnerable to genetic, molecular, and environmental perturbations affecting neurodevelopment. Both autism and ADHD appear to disrupt similar late-developing networks but follow opposite structural trajectories.
The researchers propose unifying neurodevelopmental framework: autism and ADHD share vulnerability within high-plasticity transmodal systems but diverge in direction and timing of cortical maturation. Opposite profiles of cortical thickness and curvature — accelerated refinement in autism, delayed specialisation in ADHD — highlight maturational timing as central principle of neurodevelopmental divergence.
This framework moves beyond simple deficit models. Different timing patterns may serve different adaptive functions. Accelerated refinement might enable early specialisation in particular domains whilst reducing flexibility in others. Delayed maturation might preserve plasticity longer, maintaining adaptability whilst potentially affecting efficiency in institutionally-valued domains.
The study’s limitations prevent distinguishing whether observed structural patterns represent strengths, weaknesses, or neutral differences. Researchers explicitly acknowledge: “Both autism and ADHD may reflect a combination of strengths and weaknesses, and this study does not elucidate how parts of these signatures may reflect behavioural strengths and/or weaknesses.”
Association cortices develop last and remain plastic longest because they demand most environmental calibration. Language networks, executive function systems, self-regulation circuits all require extensive learning and adaptation to specific cultural and institutional contexts. These late-developing regions show most pronounced differences between autism, ADHD, and control groups.
Institutional environments optimised for particular maturational timelines inevitably disadvantage brains following different developmental schedules. Educational systems demand specific cognitive capacities at specific ages. Workplaces require particular executive function profiles. Social contexts expect certain self-regulation patterns. Brains maturing faster or slower than institutional expectations encounter systematic incompatibilities.
The intermediate positioning of control participants between autism and ADHD extremes suggests typical development represents middle ground rather than singular optimal trajectory. Different maturational timing patterns may prove advantageous in different environmental contexts requiring different cognitive profiles at different developmental stages.
Citations
Mahmoudi, M., Moore, L.A., Lundquist, J., Stier, A., Anderson, M., Fayzullobekova, B., Hermosillo, R., Houghton, A., Madison, T.J., McCollum, R., Weldon, K.B., Nelson, S., Esler, A., Miranda-Dominguez, O., Fair, D.A., Tervo-Clemmens, B., & Feczko, E. (2026) — Cortical Thickness and Curvature in Autism and ADHD: A Mega-Analysis
