Reliability of Electrophysiological Measures of Cognitive Control and Sociodemographic Correlates in a Large Adolescent Cohort

Decreased cognitive control, encompassing processes like response inhibition and error processing, has been identified as a potential neurocognitive risk factor for psychopathology. Although electroencephalography (EEG) shows promise as a cost-effective and high temporal resolution method for linking brain function to behavior, large-scale EEG research to realize this potential remains limited. In this study, a population-based sample of 2994 adolescents (M_age = 18.4 years, range 16-21 years) was utilized to first assess reliability of EEG markers of cognitive control (ERN/Pe, No-Go N2/P3) measured during a Go/No-Go task. This involved computing split-half reliability, Cronbach’s Alpha (CA) and Pearson’s correlations for an increasing number of error/No-Go trials. Given the low-density setup, our analyses focused on the electrode locations FCz and Cz. Moderate reliability (split-half reliability and CA ≥ 0.6) was obtained with a minimum of 9/10 errors for ERN, 7 errors for Pe, and 15 No-Go trials for No-Go N2/P3. Second, linear regression models were used to estimate univariate and multivariate associations between child and family characteristics and EEG parameters (ERN/Pe, No-Go N2/P3 and midfrontal theta power). We report robust associations between higher IQ and larger Pe, No-Go P3 and midfrontal theta power. Maternal education and household income showed a similar association, but seemed partly explained by IQ. Smaller No-Go P3 was observed in adolescents with a migration background. Finally, larger ERN and No-Go N2 were observed in female as compared to male participants. These findings highlight the need to adjust for these variables in future EEG studies in adolescent populations. In conclusion, this study highlights the feasibility of incorporating EEG into large-scale population-based research, even with relatively short task duration. Importantly, researchers should balance higher reliability against retaining more participants and minimizing potential bias.