Dissociable EEG connectivity patterns reflect the latent general and specific dimensions of ADHD in children and adolescents.

Attention-Deficit/Hyperactivity Disorder (ADHD) exhibits substantial phenotypic heterogeneity. While bifactor models successfully isolate a general ADHD factor and specific symptom dimensions, their distinct neurophysiological underpinnings remain largely unexplored. In the current study, clinical and resting-state EEG data from 4,311 participants (aged 5-21 years) from the Healthy Brain Network (HBN) were analyzed. Confirmatory factor analysis evaluated competing structural models of ADHD symptoms. For participants with valid EEG data (n = 2,284), we performed source localization to reconstruct cortical activity, and used partial correlations and permutation tests to link factor and questionnaire scores to global EEG power spectral density and network-level functional connectivity. A restricted bifactor model comprising one general ADHD factor, a specific inattention factor, and a specific hyperactivity factor provided the optimal fit and interpretability (RMSEA = .06, SRMR = .04, CFI = .94, TLI = .93). The general factor reflected overall symptom severity and was linked to widespread increases in delta, theta, and beta power (r range: [.06, .08], ps < .05), as well as global network hyperconnectivity (r range: [.08, .09], ps < .048). For the specific dimensions, the inattention factor mapped onto localized alpha-band hypoconnectivity between the default mode and ventral attention, somatomotor, and limbic networks (r range: [-.10, -.09], ps < .032), whereas the hyperactivity factor was exclusively linked to a single theta-band hyperconnectivity edge between the frontoparietal and limbic networks (r = .09, p = .032). Notably, traditional composite questionnaire scores captured global dysfunctions but entirely obscured these localized, circuit-specific impairments. Age and sex subgroup analyses yielded only partial replication of the factor-connectivity associations, suggesting that these findings should be generalized with caution. The general and specific dimensions of ADHD are associated with distinct neural substrates. This hierarchical framework separates global cortical hypoarousal from localized network dysfunctions. These results highlight candidate neural signatures underlying the phenotypic heterogeneity of ADHD.