A longitudinal study of school-aged children with autism spectrum disorder (ASD) has revealed that two distinct neurofunctional subtypes, identified at baseline through resting-state functional MRI, are associated with divergent brain-state dynamics despite comparable overall executive function decline. Research published in the journal Molecular Autism tracked 68 children with ASD, aged 6 to 9 years at baseline, over a four-year period. By integrating standardized amplitude of low-frequency fluctuations (ALFF) and co-activation pattern (CAP) analyses, the researchers identified that similar clinical deterioration in executive function may arise from fundamentally different underlying neurobiological processes. This study addresses a critical gap in understanding how brain-state dynamics during late childhood shape the development of cognitive flexibility and emotional regulation as these children transition into early adolescence, offering a pathway for more precise, biologically informed clinical interventions.
Executive function impairments, encompassing domains such as cognitive flexibility, inhibition, and emotional control, are core features in many individuals with ASD and are frequently linked to increased severity in emotional and behavioral comorbidities, including anxiety and sleep disturbances. Although clinical research has extensively documented these deficits, the developmental trajectories of executive function—particularly during the transition from childhood to adolescence—remain inconsistent, and the neurobiological correlates of this heterogeneity are poorly understood. The current investigation adopted a longitudinal follow-up design to determine if clinical variability in EF development could be mapped onto distinct neurodynamic profiles, treating baseline neurofunctional stratification as a complementary approach to traditional clinical metrics. The study enrolled 68 children with ASD and 50 age-matched typically developing (TD) controls at baseline, following the ASD group for an average of 3.77 years to characterize longitudinal outcomes using the Behavior Rating Inventory of Executive Function (BRIEF).
The study employed a normative-deviation framework to classify the ASD participants into two distinct neurofunctional subtypes based on baseline ALFF patterns, which measure the intensity of spontaneous brain activity. By projecting voxel-wise ALFF maps onto the canonical Yeo-7 network parcellation and calculating deviation Z-scores relative to the TD control group, researchers derived two stable clusters through hierarchical clustering. Subtype 1 (n=32) and Subtype 2 (n=36) exhibited distinct patterns of atypical baseline activity: Subtype 1 showed relative hyperactivity in the visual (VIS) and somatomotor (SMN) networks alongside hypoactivity in prefrontal control circuits, whereas Subtype 2 displayed the inverse pattern, characterized by decreased activity in posterior sensory areas and increased activity in the frontoparietal network (FPN). These subtypes did not differ significantly in their baseline core autism symptom severity, indicating that these neurofunctional profiles capture a dimension of biological heterogeneity not fully captured by diagnostic phenotypes alone.
To investigate the temporal nature of these neural deviations, the researchers used co-activation pattern (CAP) analysis, a framewise dynamic functional connectivity approach that characterizes recurring, transient configurations of large-scale network interactions. K-means clustering of the baseline fMRI data yielded five recurring CAP states, which were used to derive metrics including occupancy rate, mean persistence (dwell time), and transition probability for each participant. This approach revealed that the static ALFF abnormalities observed in the two subtypes were not isolated regional phenomena but were tied to biased engagement of specific CAP states over time. Subtype 1 exhibited a “sensory stickiness” profile, characterized by prolonged occupancy and persistence of a visual-related CAP state, whereas Subtype 2 displayed a “control instability” profile, characterized by frequent reconfiguration and transitions between brain states, particularly involving the default mode and frontoparietal networks.
Longitudinal clinical assessments demonstrated that, across the ASD cohort, executive function declined significantly over the four-year follow-up period, particularly within behavioral regulation subdomains such as inhibition, shift, and emotional control. When analyzed through the neurofunctional subtype lens, Subtype 2—the control-instability profile—was associated with a broader range of EF impairments across longitudinal measures. Crucially, the researchers identified distinct brain-behavior associations: the probability of specific state transitions correlated with changes in executive function scores in opposite directions for the two subtypes. For instance, CAP3-to-CAP5 transition probability was positively correlated with changes in emotional control in Subtype 1, whereas this same transition showed an inverse correlation with emotional control in Subtype 2. These findings provide evidence for a form of equifinality in ASD, where seemingly uniform phenotypic deterioration is driven by divergent neurodynamic mechanisms.
The study also utilized principal component analysis to identify three clinical trajectory clusters—declining, stable, and improving groups—to test whether these clusters showed distinct baseline brain-state dynamics. This supplementary analysis revealed limited baseline CAP differentiation across the clinical trajectory groups, suggesting that while longitudinal clinical outcomes are meaningful, they are likely shaped by a multifaceted combination of baseline neural organization, subsequent developmental compensation, and environmental factors. This finding underscores the necessity of a complementary stratification framework that integrates both baseline neurobiological features and longitudinal clinical progression to enhance risk prediction and identify individuals most in need of targeted support during the critical transition to adolescence.
Spatial correlation and linear reconstruction models further validated the link between transient CAP dynamics and static ALFF patterns, suggesting that regional activity deviations represent long-timescale footprints of aberrant network-level switching. The observed prefrontal-occipital antagonism, which appeared in both the static ALFF maps and the transient CAP state definitions, suggests that these patterns reflect fundamental organizational features of large-scale network function rather than isolated regional deficits. This interpretation aligns with network neuroscience accounts that define executive control as an emergent property of cross-network coordination, where efficient function relies on the stability of adaptive hubs rather than simple, strong regional activation. Dysfunction in these processes may thus manifest as either a failure to flexibly reorient attentional resources, as seen in the sensory-stickiness subtype, or as inefficient, noisy regulation due to over-frequent network reconfiguration in the control-instability subtype.
Despite the insights provided, the study acknowledges several limitations, including the observational nature of the resting-state measures and the lack of a longitudinal TD control group, which prevents the precise quantification of neurodynamic developmental paths relative to normative maturation. The observed IQ differences between the ASD and TD cohorts, while controlled for as a covariate in the analyses, suggest that some network-level effects might reflect broader cognitive-developmental variation rather than ASD-exclusive signatures. Furthermore, the attrition of participants during the four-year follow-up, though not introducing systemic bias in baseline clinical characteristics, warrants caution regarding the generalizability of these findings to more cognitively diverse populations. The study emphasizes that labels like “sensory stickiness” and “control instability” remain descriptive summaries of CAP dynamics, rather than direct proxies for cognitive mechanisms, and calls for future interventional research—such as neuromodulation or tailored cognitive training—to validate these potential targets.
The implications for precision medicine in autism are substantial, as the research suggests that interventions targeting executive function should not be considered one-size-fits-all. Instead of assuming a uniform biological target, clinical strategies might be refined by aligning intervention approaches with subtype-specific neurodynamic features. For individuals presenting with a sensory-dominant dynamic profile, interventions that emphasize context-embedded cognitive flexibility and rule-updating tasks may be particularly effective. Conversely, individuals exhibiting the control-instability profile may benefit from targeted training aimed at stabilizing the recruitment of control-related networks, potentially through methods such as real-time fMRI neurofeedback. By moving toward a framework that maps mechanistic subtypes onto specific intervention targets, researchers may be able to reduce the heterogeneity of treatment response that frequently complicates autism-related clinical trials.
In conclusion, this study demonstrates that longitudinal clinical heterogeneity in ASD is accompanied by distinct neurodynamic signatures at the level of instantaneous brain-state configuration. The evidence supports the integration of spatiotemporal neuroimaging markers into the diagnostic and assessment toolkit to delineate between mechanistically distinct subtypes that share common behavioral outcomes. This integrative framework provides a robust foundation for future research aiming to move beyond categorical diagnostic labels, enabling the development of personalized intervention strategies that account for the individual’s underlying brain-state regulation. Future large-scale longitudinal investigations will be essential to further clarify the predictive power of these neurofunctional subtypes and to optimize clinical translation in the service of supporting executive function development in autistic youth.
Subject of Research: Neurofunctional stratification of autism spectrum disorder using longitudinal executive function trajectories and resting-state brain-state dynamics.
Article Title: Spatiotemporal brain-state dynamics delineate executive function subtypes in school-aged autism: evidence from co-activation patterns and a four-year follow-up
Article References: Yue, Z., Zhu, J., Wang, Y., Guan, L., Yin, S., Fang, H., & Ke, X. (2026). Spatiotemporal brain-state dynamics delineate executive function subtypes in school-aged autism: evidence from co-activation patterns and a four-year follow-up. Molecular Autism, 17(1), Article 40. https://doi.org/10.1186/s13229-026-00728-x
Image Credits: AI Generated
DOI: 10.1186/s13229-026-00728-x
Keywords: autism spectrum disorder, executive function, resting-state fMRI, co-activation patterns, longitudinal study, ALFF, neurofunctional subtyping, brain-state dynamics, neurodevelopment, behavior regulation, precision psychiatry, network neuroscience, BRIEF, adolescent transition, cognitive flexibility, clinical heterogeneity, functional connectivity, neural markers
News Source: Cassandra Pierce. (November 19, 2026). Distinct Brain-State Dynamics Mark Executive Function Decline in School-Aged Autism. Scienmag.



