Executive functions are the brain’s top-down control processes, governed largely by prefrontal cortex networks, that allow people to plan, organize, inhibit impulses, update information in working memory, and adapt behavior flexibly rather than relying on automatic responses. Researchers typically divide them into three core components—inhibitory control, working memory, and cognitive flexibility—alongside higher-level functions such as reasoning, planning, and problem-solving. These capacities are often selectively impaired across different neurodevelopmental disorders, and the resulting deficits contribute to behavioral difficulties and mental health problems. Without effective treatment, executive dysfunction can persist into adulthood, which is why identifying interventions that genuinely work in childhood has become a pressing scientific priority.

The challenge has been that previous reviews produced inconsistent findings. Most focused on a single diagnosis, usually ADHD, or examined one intervention category in isolation, and relied on conventional pairwise meta-analyses that can only compare two treatments at a time. The new study, led by Xiao Liang and Ming Chen with colleagues including Raymond Chan and Shirley Li, overcame this limitation by using network meta-analysis, a statistical technique that integrates both direct evidence from head-to-head trials and indirect evidence to compare many interventions simultaneously, even those never tested against each other in the same experiment. The team searched six major databases from inception through September 2025 and registered their protocol prospectively.

The headline finding is encouraging: overall, non-pharmacological interventions were associated with improved performance on task-based executive function measures, with a pooled effect size of 0.521 standardized mean difference across 141 intervention arms—a moderate effect. Every broad category showed significant benefits, but physical activity training produced the largest pooled estimate (0.621), followed by computerized cognitive training (0.501), non-computerized executive function training (0.442), neurofeedback (0.434), and mindfulness training (0.381). Notably, meta-regression revealed that the total number of sessions significantly moderated outcomes, with more sessions associated with larger effects, underscoring that sustained, repeated practice matters more than any single magic bullet.

When the researchers drilled down into specific executive function domains using twelve finer intervention categories, a more nuanced picture emerged. For cognitive flexibility, computerized serious games ranked highest, showing a large effect compared with treatment as usual (standardized mean difference 1.30). For working memory, serious games again topped the rankings (0.79), ahead of computerized working memory training (0.67) and computer-based learning programs (0.59). For planning, serious games led once more with an effect of 1.34, followed by computer-based learning programs at 1.10. The authors suggest that the high interactivity and appeal of game-based formats may boost engagement and provide repeated opportunities to practice updating, monitoring, and flexible switching—processes at the heart of these cognitive skills.

Inhibitory control told a different story. Here, cognitively engaging exercise—physical activities that combine movement with cognitive demands, such as exergaming, table tennis, judo, and martial arts—achieved the highest probability ranking, with a SUCRA score of 92.2 percent and an effect size of 0.65 against treatment as usual. The researchers propose a plausible neurobiological mechanism: physical activity enhances neuroplasticity and cerebral blood flow, particularly in the prefrontal cortex, which is critical for executive functions and continues developing throughout childhood and adolescence. Exercise also promotes co-activation between the cerebellum and dorsolateral prefrontal cortex, and children with neurodevelopmental disorders often show delayed frontal maturation and reduced cerebellar activity, leaving more room for improvement.

The technical rigor of the analysis deserves attention. The team assessed risk of bias using the Cochrane RoB 2.0 tool, finding that 24.6 percent of studies were at low risk, 71.2 percent raised some concerns, and 4.2 percent were at high risk, with concerns centered on randomization and allocation concealment. Confidence in the network estimates was evaluated with the CINeMA framework, which applies GRADE principles to network meta-analysis. Transitivity assumptions were checked by comparing the distribution of potential effect modifiers across interventions, and no significant global or local inconsistency was detected between direct and indirect estimates. Sensitivity analyses removing individual studies, entire intervention nodes, or high-risk studies generally supported the main ranking patterns.

Yet the authors are emphatic that these rankings come with heavy caveats. The certainty of evidence was rated low or very low for most domain-specific estimates, and Egger’s regression suggested possible small-study effects or publication bias, though an exploratory trim-and-fill adjustment actually yielded a slightly larger effect size of 0.584. More importantly, the evidence base is heavily skewed: 83 of the 118 trials involved children with ADHD, compared with only 16 for autism spectrum disorder, 11 for intellectual disability, and 4 each for developmental coordination disorder and learning disorders. The sample was also predominantly male, at 76.1 percent, and sex-disaggregated outcome data were inconsistently reported. The researchers stress that the findings represent group-level estimates and should not be interpreted as individualized treatment effects across all neurodevelopmental conditions.

There are also measurement subtleties worth noting. The analysis deliberately restricted itself to performance-based neurocognitive tasks rather than parent- or teacher-reported questionnaires, treating these as complementary rather than interchangeable measures of executive function, in order to improve comparability across trials and reduce rater-related bias. Domain-specific networks showed substantial statistical heterogeneity, with I-squared values ranging from about 61 to 69 percent, and only four mindfulness studies met inclusion criteria, making conclusions about that approach particularly tentative. Computerized working memory training and neurofeedback showed mixed estimates across domains, suggesting that not every technology-based intervention delivers uniform benefits.

The practical message is one of cautious optimism. Non-pharmacological interventions—from game-based cognitive training to cognitively demanding physical exercise—appear capable of improving measurable executive function performance in children and adolescents with neurodevelopmental disorders, and the comparative rankings offer a roadmap for which approaches merit priority in future trials. The authors call for adequately powered, diagnosis-specific randomized trials with standardized executive function outcomes, transparent control conditions, and more diverse samples before these rankings can inform clinical recommendations. Until then, the study suggests that what happens outside the pharmacy—on the sports field and on the screen—may matter far more for young developing minds than previously appreciated, provided the practice is sustained, engaging, and given enough sessions to take hold.

Subject of Research: Comparative effectiveness of non-pharmacological interventions on executive functions in children and adolescents with neurodevelopmental disorders

Article Title: Non-pharmacological executive function interventions for children and adolescents with neurodevelopmental disorders: a systematic review and network meta-analysis

Article References: Liang, X., Chen, M., Chan, R. C., Chang, J. R., Yeung, M. K., Hsu, C. L., Liu, C., Chen, Y., Xu, R. H., Yang, B.-R., Yan, C., Qiu, H., Tse, A. C., Shum, D. H., Chan, R. C., & Li, S. X. (2026). Non-pharmacological executive function interventions for children and adolescents with neurodevelopmental disorders: a systematic review and network meta-analysis. eClinicalMedicine, 100, Article 104228. https://doi.org/10.1016/j.eclinm.2026.104228

Image Credits: AI Generated

DOI: 10.1016/j.eclinm.2026.104228

Keywords: executive function, neurodevelopmental disorders, ADHD, autism spectrum disorder, network meta-analysis, computerized serious games, cognitively engaging exercise, working memory, inhibitory control, cognitive flexibility, physical activity, children and adolescents