Brain’s Fleeting Electrical States Reveal New Clues to ADHD in Children

Every few hundred milliseconds, the resting brain flips between a small set of stable electrical configurations, like a projector cycling through a limited reel of slides. Researchers call these brief episodes EEG microstates, and for decades they have been described as the atoms of thought: the building blocks from which the stream of conscious mental activity is assembled. A new cross-sectional study of children and adolescents now suggests that the way these building blocks are deployed differs measurably in attention-deficit/hyperactivity disorder, offering one of the most concrete neurophysiological fingerprints of the condition to date.

The study, published in BMC Psychiatry by a team led by Ran Wei and Xiaoyan Ke of the Affiliated Brain Hospital of Nanjing Medical University and collaborating institutions, recorded resting-state electroencephalography from 121 children aged 6 to 15 years. Sixty-three of the participants had a diagnosis of ADHD, while 58 were typically developing peers. The researchers used a 24-channel EEG system with the electrode Pz as the online reference and offline average re-referencing, a standard preprocessing choice that minimises bias from any single recording site. Artefacts such as blinks and muscle activity were removed using independent component analysis with the extended Infomax algorithm, and components were classified with ICLabel to ensure that only genuine neural signals entered the analysis.

Microstate analysis works by examining the topography of scalp electrical fields at moments of maximal global field power. When the maps are clustered, they reliably resolve into four canonical classes, labelled A through D, each thought to reflect the transient dominance of a different large-scale brain network. In this sample, k-means clustering identified the four classes with a global explained variance of 54.74 percent, and the solution remained stable across cluster counts from four to seven, satisfying a key criterion for the validity of the microstate model. Class A is commonly associated with phonological processing, class B with visual imagery, class C with salience and interoceptive awareness, and class D with attentional reorientation and executive control.

From each child’s recording, the team extracted twenty-four parameters: how long each microstate lasted, how often it occurred per second, how much total time it covered, and the probabilities of transitions between every pair of classes. These measures were compared between the ADHD and typically developing groups using Welch t-tests, a variant of the t-test that does not assume equal variances, with Benjamini-Hochberg correction for the false discovery rate to guard against spurious findings arising from multiple comparisons.

Two differences survived this statistical gauntlet. Children with ADHD showed significantly reduced per-second occurrence of microstate C, with an effect size of Cohen’s d equal to minus 0.56, and a significantly reduced probability of transitioning from microstate C to microstate D, with d equal to minus 0.58. Both effects were of medium magnitude, and both remained statistically significant after the researchers adjusted for age, sex and intelligence quotient, indicating that the differences were not simply by-products of demographic mismatch between the groups.

The interpretation is intriguing because of what microstates C and D are thought to do. Microstate C has been linked to the brain’s salience network, the circuitry that decides which internal and external signals deserve attention, while microstate D is associated with the frontoparietal attention network that reorients focus toward behaviourally relevant events. A reduction in how often C appears, combined with a reduced likelihood of handing control from C to D, could therefore represent a measurable disruption in the machinery that governs when attention is captured and redirected. In other words, the electrical choreography that normally shuttles the brain between detecting something important and acting on it appears to be less fluid in ADHD.

To probe whether these electrical signatures relate to actual attentional performance, the researchers turned to the Integrated Visual and Auditory Continuous Performance Test, or IVA-CPT, a behavioural assessment in which children respond to targets and withhold responses to non-targets across both visual and auditory modalities. Within the ADHD subgroup alone, linear regression revealed positive associations between microstate C occurrence and the auditory response control quotients of the IVA-CPT, pointing in the same direction as the group-level difference and most consistent for the visual response control measure. The authors are careful to label these within-group findings exploratory: none remained significant after correction for the three tests performed, with the Benjamini-Hochberg adjusted p-value reaching 0.051, just above the conventional threshold.

The team then asked a practical question: could microstate parameters help classify children as having ADHD or not? Three machine-learning classifiers, logistic regression, random forest and support vector machine, were trained on the microstate features and evaluated under repeated five-fold cross-validation, a procedure that repeatedly partitions the data into training and testing subsets to estimate how well a model generalises to unseen individuals. The microstate-only model achieved an area under the receiver operating characteristic curve, or AUC, of 0.744, with a 95 percent confidence interval of 0.650 to 0.830. An AUC of 0.744 indicates discriminative ability clearly better than chance but well short of diagnostic certainty.

For comparison, a model built only on the IVA-CPT behavioural test reached an AUC of 0.913, and a combined model incorporating both microstate and behavioural features reached 0.930. Critically, the microstate features added no statistically significant increment over the behavioural test alone. This is a sobering result for anyone hoping that a quick EEG could replace existing assessment tools, and the authors state plainly that in this sample the microstate parameters did not improve discriminative value beyond the IVA-CPT. The finding tempers enthusiasm without extinguishing it: a candidate neurophysiological correlate need not outperform behavioural tests to be scientifically valuable, because it may illuminate mechanisms that behaviour alone cannot reveal.

The broader significance of the study lies in the persistent gap it tries to fill. ADHD affects roughly 3 to 7 percent of school-age children, yet clinical diagnosis remains entirely behavioural, resting on interviews, rating scales and performance tests, with no objective neurophysiological marker in routine use. By showing that the dynamics of large-scale network activity, indexed by microstate C engagement and C-to-D transitions, differ between children with ADHD and their peers with medium effect sizes, the study strengthens the case that microstate analysis could eventually complement clinical assessment rather than replace it. The authors themselves emphasise the limits of a single cross-sectional sample: replication in independent cohorts, external validation across sites, and longitudinal designs that track how microstate dynamics change with development and treatment are all required before any clinical translation. For now, the flickering electrical states of the resting brain have offered researchers a new, quantifiable window onto one of the most common neurodevelopmental conditions of childhood, and a reminder that the biology of attention may be legible in signals lasting barely a tenth of a second.

Subject of Research: EEG microstate dynamics as candidate neurophysiological correlates of ADHD in children and adolescents

Article Title: EEG microstate dynamics as candidate neurophysiological correlates of attention-deficit/hyperactivity disorder in children and adolescents: a cross-sectional study

Article References: Wei, R., Zhang, D., Zhang, H., Jin, H., Fang, H., Wang, Y., Gao, J., Xu, X., Ma, Z., & Ke, X. (2026). EEG microstate dynamics as candidate neurophysiological correlates of attention-deficit/hyperactivity disorder in children and adolescents: a cross-sectional study. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08555-3

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08555-3

Keywords: ADHD, EEG microstates, electroencephalography, neurodevelopmental disorders, attentional control, IVA-CPT, machine learning, neurophysiology, brain networks, children and adolescents, candidate biomarkers, resting-state EEG