Maturational trajectories of neurophysiological indices of inhibitory control in typically developing children

Abstract

The development of cognitive control during childhood is characterized by progressive improvements in attentional processing and response inhibition. The present study examined neural activity associated with visual attention and inhibitory control in typically developing children aged 6–12 years, using quantitative high-density electroencephalography (HD-EEG). A total of 172 typically developing children aged 6–12 years participated in the study. Participants completed a computerized Go/NoGo task while EEG activity was recorded. Behavioral performance, event-related potentials (ERPs), and time-frequency EEG features were analysed, including median frequency (MF), spectral entropy (SE), and relative power (RP) in the theta (θ) and beta1 (β₁) frequency bands. Differences were observed both between task conditions and across age groups. P300 amplitude and latency decreased linearly with age across conditions. NoGo trials elicited larger P300 amplitudes and longer latencies than Go trials. Time-frequency analyses showed reductions in MF and SE with increasing age. During task performance, RP (θ) increased and RP (β₁) decreased relative to the pre-stimulus baseline. These time-frequency changes were observed across all age groups but were more pronounced in older children. All time-frequency measures differentiated Go and NoGo trials, with stronger effect observed during NoGo trials. These patterns suggest that with age, children become progressively more capable of synchronizing their frequency-specific neural activity to meet the functional demands of the task. Together, these findings indicate that HD-EEG biomarkers are useful to study the maturation of large-scale coordination mechanisms and inhibitory control networks across childhood.

Graphical Abstract

Highlights

• •P300 amplitude and latency decrease linearly with age in the visual Go/NoGo task.
• •Changes in ERP morphology may reflect progressive increases in neural efficiency.
• •Developmental shift of spectral power toward lower frequencies with increasing age.
• •Increase in mid-central theta power with age may reflect maturation of inhibitory control.
• •Inhibitory control becomes more selective, automatic, and efficient with age.

1. Introduction

Inhibition, a concept from cognitive psychology, is considered a foundational component of executive functioning and cognitive control. It refers to the ability to intentionally suppress automatic or dominant responses, which is essential to filter irrelevant information, ignore competing stimuli, and regulate emotion and behavior (Anderson, 2002, Miyake et al., 2000). Inhibition and cognitive control are therefore essential for adaptive functioning across multiple domains of life, particularly for social and emotional development and academic achievement (Anderson, 2002, Best and Miller, 2010, Diamond, 2013, Miyake et al., 2000). There is consistent evidence that inhibition emerges very early in life and undergoes continuous refinement throughout childhood, adolescence, and early adulthood. Its progressive development is closely linked to the structural and functional maturation of the frontal lobes, driven by key neurobiological processes such as myelination, synaptogenesis, apoptosis, and synaptic pruning (Casey et al., 2005, Romine and Reynolds, 2005). The neural dynamics of inhibitory control across developmental populations have been extensively examined using the Go/NoGo paradigm and event‑related potentials (ERPs) derived from electroencephalography (EEG) (Brooker et al., 2020, Hoyniak, 2017, Lamm et al., 2006, Peters et al., 2023). The Go/NoGo task measures the ability to provide a fast motor response to frequent stimuli designated as “Go”, and withhold the response for infrequent stimuli designated as “NoGo” (Halperin et al., 1991). EEG has been widely used to investigate inhibitory control due to its high temporal resolution, which enables the characterization of the rapid temporal neural dynamics and the mechanisms underlying these functions. Previous studies have consistently shown that ERPs elicited during visual Go/NoGo paradigms differentiate between task conditions in both adult and paediatric populations. In particular, the P300 component (defined as a positive deflection occurring approximately 200–600 ms after stimulus onset and typically maximal over centroparietal regions) consistently exhibits larger amplitudes and longer latencies for infrequent NoGo stimuli compared to frequent Go stimuli (Folstein and Van Petten, 2008, Hoyniak, 2017). These variations in the P300 waveforms are generally interpreted as reflecting increased demands on inhibitory control and greater allocation of attentional resource during NoGo trials (Bokura et al., 2001, Bruin and Wijers, 2002, Pfefferbaum et al., 1985, Polich, 2012).

In recent years, increasing interest has focused on describing the developmental changes of P300 component from childhood to adolescence, as it may provide insight into the neural generators and maturation of inhibition and cognitive control (Peters et al., 2023, Riggins and Scott, 2020). Most studies investigating age-related changes in the P300 component have predominantly used auditory oddball paradigms (Čeponienė et al., 2004, Dinteren et al., 2014, Määttä et al., 2005), while relatively few have applied visual paradigms (Stige et al., 2007), and even fewer have specifically assessed P300 development within visual Go/NoGo task (Peters et al., 2023, St. John et al., 2019, Willner et al., 2015). Findings regarding age-related changes in P300 amplitude during visual Go/NoGo tasks remain inconsistent. Several cross-sectional studies examining preschoolers have reported no significant developmental differences in P300 amplitude with age (St. John et al., 2019, Willner et al., 2015). In contrast, Peters et al. (2023), who conducted a longitudinal study in a large cohort of preschool children, observed consistent age-related reductions in both P300 latency and amplitude during a visual Go/NoGo task. These discrepancies likely reflect substantial methodological heterogeneity within childhood ERP literature, including wide variation in recording montages, reference schemes, data acquisition protocols, and experimental paradigms.

Importantly, there remains a gap in the ERP literature concerning visual Go/NoGo paradigms in middle childhood, as most studies examining developmental trajectories on P300 component have focused on preschool populations. To our knowledge, the only study that has specifically focused on assessing visual Go/NoGo task in middle childhood is Lamm et al. (2006). However, this study reported development changes in other ERP components and did not specifically analyse the P300.

Overall, evidence indicates that P300 morphology exhibits age-related changes in preschoolers that could extent from childhood to adolescence. These changes may reflect neuronal maturation and increasing efficiency of neural processes supporting executive functioning (Broyd et al., 2005, Epstein et al., 2014, Hoyniak, 2017, Lamm et al., 2006). However, the P300 presents several limitations as an index of developmental changes in neuronal dynamics. The functional significance of ERP components remains a matter of ongoing debate, given that the P300 can be elicited across a wide range of tasks involving stimulus processing or discrimination, potentially reflecting multiple overlapping cognitive processes (Kappenman and Luck, 2012). These processes may include attention, working memory, problem-solving (Polich, 2007, Riggins and Scott, 2020), as well as motor inhibition and information processing (Enriquez-Geppert et al., 2010).

Complementary to ERPs analysis, the study of neural oscillations provides additional insight into ongoing brain dynamics and large-scale network interactions underlying the development of inhibition and cognitive control. Neural oscillations represent fundamental mechanisms for coordinating information processing across distributed brain systems, and their properties undergo substantial developmental changes throughout childhood. However, EEG time-frequency analysis has been rarely used in developmental paediatric studies on inhibitory control (Morales and Buzzell, 2025). The available evidence indicates that increases in P300 amplitude in NoGo trials are accompanied by enhanced mid-frontal theta power activity (Kappenman and Luck, 2012), supporting the interpretation of theta oscillations as neural markers of inhibitory control (Barry, 2009, Kamarajan et al., 2004, Kamarajan et al., 2006, Kirmizi-Alsan et al., 2006, Yamanaka and Yamamoto, 2009). Although these patterns have been documented in healthy adults performing the Go/NoGo and related cognitive control tasks, evidence in paediatric populations remains limited (Xu et al., 2025). Mapping age-related changes in inhibition and cognitive control, together with their neural correlates, may contribute to delineating typical and atypical developmental trajectories of executive functioning. These developmental trajectories may, in turn, provide valuable insight to identify sensitive periods for intervention and educational support in neurodevelopmental disorders characterized by executive dysfunction, including Attention Deficit Hyperactivity Disorder (ADHD) and Autism Spectrum Disorder (ASD) (Fossum et al., 2021).

The aim of the present study was to characterize the neural dynamics underlying visual attention and response inhibition in a cohort of typically developing children aged 6–12 years. We analysed developmental changes in ERPs components and time-frequency features derived from HD-EEG recordings acquired during a visual attention task based on the Go/NoGo paradigm. In addition, we assessed the relationship between ERP components, neural oscillatory frequency patterns, and behavioral performance during an inhibitory control task.

2. Method

2.1. Participants

Participants were recruited from a collaborating local primary school from Barcelona, Spain. Prior to data collection, participants were screened for neurodevelopmental disorders by a qualified psychologist. 10 participants were excluded from the study due to suspected learning disorders. Following EEG acquisition, 8 participants were excluded due to poor EEG signal quality or technical issues. Poor signal quality was defined as EEG recordings containing excessive artifacts throughout most of the task, preventing reliable preprocessing and analysis. Technical issues included poor electrode-scalp contact affecting more than 10% of the electrodes and the case of one participant whose responses were not consistently registered because the response button was not pressed firmly enough during the task.

Following these exclusions, the final cohort comprised 172 children (84 boys and 88 girls) aged 6–12 years. All participants had normal or corrected-to-normal vision. Children were grouped by age (6, 7, 8, 9, 10, 11, and 12 years), with sample sizes ranging from 22 to 27 participants per each age group and a balanced sex distribution (detailed information is exposed in Table 1). The sample size was determined based on a target statistical power of 0.80, a significance level of α = .05 (two-tailed), and effect size estimates derived from preliminary analyses, which suggested a moderate to large effect of age (Cohen’s f ≈.31) and task condition (Cohen’s f ≈.25) on ERP and time-frequency measures. Power calculations indicated that approximately 140–180 participants would be required to detect effects of this magnitude in between-subject designs, with fewer participants needed for within-subject factors (approximately 50–70). Therefore, the achieved sample size of 172 provided adequate power to detect effects of this size.

Table 1.

Sample characteristics and behavioral measures of task performance.

Age (years)	n	Sex	Go trials accuracy (%)	NoGo trials accuracy (%)	Omission errors	Commission errors
		M:F	M (SD)	M (SD)	M (SD)	M (SD)
6	22	13:09	95.34(4.57)	75.94(13.6)	5.58(5.49)	9.62(5.44)
7	25	14:11	95.38(6.44)	73.07(15.13)	5.54(7.73)	10.80(6.06)
8	27	10:17	97.59(4.21)	81.48(8.48)	2.89(3.06)	7.41(3.53)
9	25	11:14	97.50(4.83)	84.48(11.07)	3.00(3.79)	6.32(4.43)
10	24	11:13	97.90(0.97)	82.9(8.41)	2.52(1.39)	6.84(3.36)
11	24	12:12	99.20(2.97)	88.23(6.82)	0.96(1.16)	4.71(2.73)
12	25	13:12	98.77(4.01)	86.48(8.06)	1.48 (2.81)	5.41(3.23)

Note: Values are presented as Mean (M) and Standard deviation (SD). Sex is reported as M:F (male:female). n = number of participants.

The study was approved by the local institutional Ethics Committee of Sant Joan de Déu Hospital (reference PIC-44–22). All research was conducted in accordance with the principles of the Declaration of Helsinki (2024). Written informed consent was obtained from the parents or legal representatives of all participants.

2.2. Procedure

Visual attention and inhibition were assessed using a computerized version of the zoo Go/NoGo task (Halperin et al., 1991, Lamm et al., 2006), which was programmed using the PsychoPy 2023.1.1 software (Peirce et al., 2019). Participants were seated in front of a computer monitor in a quiet room, with viewing distance and alignment standardized across individuals. Screen had an active display of 400 × 300 mm, a resolution of 1024 × 768 pixels, and a refresh rate of 85 Hz. Participants were instructed to remain still throughout the task and to minimize unnecessary head and hand movements to reduce motor artifacts.

The visual stimuli consisted of colored pictures of animals presented on a black background on the screen. Each trial began with a white fixation cross displayed at the center of the screen for a variable interval of 200–300 ms. This was followed by an animal image displayed for 750 ms. Subsequently, a black screen was presented for 500 ms. Children were instructed to monitor the visual stimuli and to press a button as quickly as possible whenever they saw an animal (Go trials), while withholding their response when an orangutan was presented (NoGo trials). Task responses were recorded using a push-button panel and were considered valid if they occurred at any time from the stimulus onset until the end of the 500 ms black-screen interval. The task consisted of 320 trials divided into two blocks. The duration of each block was 4 min, with a short pause of approximately 3 min between them. Each block contained 160 trials, comprising 120 Go trials (75%) and 40 NoGo trials (25%) distributed randomly.

2.3. EEG signal recording

Non-invasive EEG recordings were acquired using a high-density 64-channel Neuroscan Quik Cap connected to a Neuvo 64-channel Amplifier, and the CURRY9X software (Compumedics Limited, Abbotsford, VIC, Australia). The head circumference was measured, and an appropriately sized EEG net was selected for each participant. The cap was positioned according to the international 10–10 system. Participants were seated in a comfortable position to facilitate electrode placement and impedance adjustment, in a dimly lit and acoustically isolated room. Prior to EEG recording, the impedance of each electrode was measured and maintained below 10 kΩ. Data were acquired at a sampling frequency of 1000 Hz using a montage referenced between CPz and Cz electrodes. To verify proper signal acquisition, a band-pass filter between 1 and 50 Hz was applied at the beginning of the session to enable visual inspection of the EEG signal.

2.4. Data analysis

2.4.1. Behavioral measures

Performance on the Go/NoGo task was assessed using three behavioral measures: (1) Accuracy on the task was calculated as the percentage of correct responses, defined as the number of trials in which participants provided a motor response to Go stimuli or a withheld response to NoGo stimuli, (2) Omission errors were calculated as the number of times participants failed to provide a motor response to Go stimuli, and (3) Commission errors were calculated as the number of times participants provided motor response to NoGo stimuli.

2.4.2. EEG signal preprocessing

The EEG recordings were pre-processed using MATLAB R2024b (MathWorks, Natick, MA, USA) and EEGLAB 2024.2, a software package designed for electrophysiological data analysis. The EEG signals were segmented into two separate blocks to allow independent analysis and to minimize potential intra-session effects. All pre-processing steps were applied identically to each block.

First, 8 out of 64 EEG channels were excluded prior to analysis due to poor scalp contact or persistent noise (M1, M2, F11, FP11, F12, FP12, Cb1 and Cb2), as these channels could compromise the quality of the signal and the ICA decomposition. Additional bad channels were identified using a Tukey-based outlier detection method, whereby channels with values exceeding 1.5 times the interquartile range (IQR) below the first quartile or above the third quartile were considered outliers and excluded from further analysis. Subsequently, EEG data were decimated from 1000 Hz to 500 Hz using a prior Chebyshev type II band-pass filter (0.3–140 Hz). Additionally, 4th order Butterworth notch filters were applied at 50 Hz and harmonics. After filtering, EEG channels were re-referenced using the common average reference (CAR) method. Independent Component Analysis (ICA) was applied using the AMICA algorithm (Delorme et al., 2012, Palmer and Kreutz-Delgado, 2011). Independent components were classified using the ICLabel plugin, and components corresponding to ocular, muscular, cardiac, or channel noise artifacts were identified and removed prior to reconstructing the cleaned EEG signals.

From artifact-free EEG data, spherical interpolation was applied to reconstruct signals from bad or missing electrodes using information from neighboring channels (Perrin et al., 1989). The continuous EEG data were then segmented into stimulus-locked epochs also named trials, covering a time window from −1 s pre-stimulus to 1.5 s post-stimulus. Trial matrices were extracted separately for each condition (Go/NoGo) and for each block. Then, a bad trial rejection was applied. For each block and condition, trials were evaluated across all channels using four features: amplitude, kurtosis, variance, and signal energy. Trials showing extreme deviations in any of these metrics were identified using a Tukey-based outlier method and were rejected if more than 25% of channels were marked as poor quality. Supplementary Table 1 presents descriptive statistics of the artifact-free trials, including the mean, standard deviation, and range for each age group and trial type.

Baseline correction was then applied to each EEG epoch by removing the mean amplitude in the time window between −500 and −300 ms prior to stimulus onset, which was used as the baseline interval.

After baseline correction, two complementary analyses were conducted:

• I.Event - related potential (ERP) analysis

  For each participant and condition, ERPs were obtained as the averaged amplitude among trials. The resulting ERPs were then averaged across a predefined fronto-central electrode region of interest (ROI) (F3, F4, Fz, C3, C4, Cz, FC3, FC4 and FCz). The choice of this ROI is consistent with previous studies that detected differences in relevant ERP components (Baijot et al., 2017, Gajewski and Falkenstein, 2013, Vanderwert et al., 2016). P300 amplitudes and latencies were extracted from the averaged ERP waveforms for each participant and condition. Given that a fronto-central cluster was assessed, peak amplitude was defined as the largest positive deflection within the predefined P300 time window (250–500 ms), and peak latency as the time point at which this peak occurred after the stimulus. Mean amplitude was calculated as the average voltage within the same window.

• II.Time-frequency analysis

The time-frequency analysis was performed using the Continuous Wavelet Transform (CWT) with a complex Morlet wavelet (the bandwidth and the wavelet center frequency parameters were set to 1 which provides a suitable trade-off between frequency and temporal resolution). Spectral content was examined from 4 Hz to 48 Hz in 1 Hz bins. CWT coefficients were computed independently for each channel and each trial, and the resulting time-frequency representations were normalized within the 4–45 Hz range according to the time-frequency resolution of the wavelet at each frequency bin. From the normalized CWT power spectrum, three metrics were computed: median frequency (MF), Shannon spectral entropy (SE) and relative power (RP). MF was defined as the frequency that divided the power spectrum into two equal halves, with 50% of the total power below and 50% above that frequency (Tonner and Bein, 2006). SE quantifies the degree of spectral dispersion or order in a signal (Powell and Percival, 2001, Rosso, 2007). RP indicates the ratio of the power within a specific frequency band to the total band spectral power (Zhou et al., 2021). For spectral power analysis, the following frequency bands were selected according to previous studies (Wang et al., 2016): theta (θ; 4–8 Hz), alpha (α; 8–13 Hz), beta1 (β₁; 13–19 Hz), beta2 (β₂; 19–30 Hz) and gamma (ɣ; 30–45 Hz). Finally, to assess the task effect (Go vs. NoGo), baseline correction was performed using the percent of change method, with the baseline defined as the window of −500 to −300 ms prior to stimulus onset (Bachiller et al., 2014).

2.5. Statistical analysis

For each participant, the distribution of electrode-level values within the predefined region of interest (ROI) was examined. While distributions were generally symmetric across participants, occasional extreme electrode values were observed in a small subset of cases (≤8% across EEG measures). To reduce the influence of these isolated extreme values, median values were computed across electrodes within the ROI for each EEG feature, providing a robust and stable summary of regional activity.

Participants were then grouped by age, and median values for each age group and task condition were calculated.

Assumptions of normality of residuals and homoscedasticity were assessed prior to hypothesis testing. A mixed‑design ANOVA was performed to evaluate both between‑group and within‑group differences in EEG features and task performance. Partial eta squared (ηp²) was reported as a measure of effect size for all significant main effects (small effect, ηp²≈.01; medium effect, ηp²≈.06; large effect, ηp²≈.14). When effects were detected, post hoc pairwise comparisons were conducted using estimated marginal means. False Discovery Rate (FDR) correction was applied to adjust for multiple comparisons.

For time-frequency measures, FDR corrected p‑values were transformed to the log10 scale and visualized using topographic maps, enabling spatial interpretation of age‑related differences across task conditions and frequency bands, particularly in the RP analysis. Statistical significance was set at α = .05, and all statistical tests were two‑tailed. All statistical analyses were performed using RStudio 2025.09.2.

3. Results

3.1. Behavioral measures

Task performance significantly improved with age (see Table 1). Accuracy in Go trials, reflecting sustained attention and task compliance, was consistently high across all participants (≥ 90%) and showed a small age-related increase (F(6, 165) = 3.45, p < .05, ηp² = .11). In contrast, NoGo trial accuracy, reflecting inhibitory control, exhibited greater age-related improvements (F(6, 165) = 6.06, p < .001, ηp² = .18). Correspondingly, both omission and commission error rates decreased significantly across age groups. Pairwise comparisons indicated that these effects were primarily driven by differences between younger children (6–7 years) and older age groups. Specifically, children aged 6 and 7 years showed significantly lower accuracy rates than those aged 8, to 12 years (ps ≤.023). A similar pattern was observed for omission and commission errors, where children aged 6 and 7 years exhibited higher error rates than children aged 8–12 years (ps ≤.039). No significant differences were observed among older age groups (9–12 years) for either accuracy or error rates (ps >.05).

3.2. ERP Analysis: P300 amplitude and latency for the Go/NoGo task

The effects of age and task condition on the P300 component were examined at fronto-central ROIs. Both P300 amplitude and latency linearly decreased with age, with older children showing progressively smaller amplitudes (F(6, 154) = 6.75, p < .001, ηp² = .2) and shorter latencies (F(6, 154) = 5.07 p < .001, ηp² = .16) (see Fig. 1). Pairwise comparisons indicated that age-related differences in P300 amplitude emerged after age 8. Children aged 6–8 years showed significantly larger P300 amplitudes than those aged 10–12 years in both task conditions (ps <.015). No significant differences were observed between adjacent age groups (e.g., 6 vs. 7 or 8 years; 10 vs. 11 or 12 years) (ps >.05).

Fig. 1.

Grand-average ERP waveforms. ERPs represented for each age group and task condition recorded at frontocentral electrode ROI (F3, F4, Fz, C3, C4, Cz, FC3, FC4 and FCz). Waveforms for Go trials are represented in red and NoGo trials in blue. Shaded areas represent the standard error of the mean (SEM) for each condition. The P300 component, defined as the maximum peak within the ∼250–500 ms, appears inverted in polarity, consistent with expected dipole orientation at frontocentral locations.

Regarding task condition effects, NoGo trials elicited larger P300 amplitudes (F(1, 154) = 75.6, p < .001, ηp² = .33) and longer latencies compared to Go trials (F(1, 154) = 34.3 p < .001, ηp² = .18). This shift between Go and NoGo remained consistent across age groups with no significant interaction between age and task condition (ps > 0.05).

3.3. Time-frequency analysis

An overview of the effects of age and condition for each variable is presented in Fig. 2, illustrating the developmental trajectories of the four time-frequency features analysed in this study: MF, SE, RP (θ), and RP (β₁). These trajectories reflect the median change relative to baseline across participants within each age group and task condition.

Fig. 2.

Developmental trajectories of median frequency (MF), spectral entropy (SE) and relative power in the theta (RP (θ)) and beta1 (RP (β₁)) bands across ages. Each point represents the median value of each feature within the predefined ROI for each age group and task condition.

3.3.1. Median frequency (MF)

Time-frequency analysis relative to the baseline revealed that MF was significantly reduced with increasing age (F(6, 154) = 13.6, p < .001, ηp² = .34). Topographic maps of MF showed an age-related reduction in MF across the scalp (see Fig. 3a). In younger children (6–7 years), MF showed minimal deviation from baseline, whereas older children exhibited a more pronounced reduction particularly over mid-central and posterior regions. In addition, a significant effect of task condition on MF was observed (see Fig. 2a), with NoGo trials showing higher deviations from baseline than Go trials (F(1, 154) = 38.9, p < .001, ηp² = .31). This effect was consistent across all age groups, with no interaction between age and task condition (F(6, 154) = 0.85, p = .59, ηp² = .03).

Fig. 3.

Developmental differences in median frequency (MF): topographic maps and statistical comparisons. a) Topographic maps of baseline-relative changes in MF for each age cluster. b) Topographic maps showing the spatial distribution of statistically significant effects between age groups for Go (left) and NoGo (right). Color scales indicate the magnitude of changes (Fig. 3a) and significance levels (Fig. 3b).

Fig. 3b depicts the topographic maps of corrected p-values, displaying the spatial distribution of statistically significant effects between age groups. Significant differences emerged between groups with larger age gaps (e.g. 6–7 vs 10–12 years; ps <.001). Younger children (6–7 years) differed consistently from older groups across both Go and NoGo conditions, particularly over mid-central and posterior electrode sites. In contrast, comparisons between children of adjacent or similar age groups (e.g., 6 vs 7 or 10 vs 12 years) revealed no significant differences in either condition (ps >.71). Although both conditions showed significant differences across the same age-group comparisons, the Go condition exhibited a higher significant effect than the NoGo condition.

3.3.2. Spectral entropy (SE)

Analysis of SE revealed a significant effect of age (F(6, 154) = 7.02, p < .001, ηp² = .21) and condition (F(1, 154) = 32.1, p < .001, ηp² = .17). SE decreased with age relative to baseline in both Go and NoGo conditions, with a more focused pattern observed for NoGo trials (see Fig. 1 in Appendix A). This SE reduction was observed across all age groups, although it was less pronounced in younger children than in older age groups. Regarding the effect of task condition, significant differences were observed between them (see Fig. 2b), consistent with the pattern reported for MF (see Figs. 2a and 3, with NoGo trials showing greater deviations from baseline than Go trials.

3.3.3. Relative power analysis (RP)

The analysis of baseline‑referenced RP indicated significant age-related variation across frequency bands. Specifically, significant effects of age were observed in the theta band (θ) (F(6, 154) = 11.85, p < .001, ηp² = .31) as well as in the beta1 (β₁) (F(6, 154) = 10.36, p < .001, ηp² = .29), and beta2 (β₂) bands (F(6, 154) = 4.903, p < .001, ηp² = .16). In contrast, no significant differences of age were found in alpha (α) (F(6, 154) = 0.369, p = .897, ηp² = .29) or gamma (ɣ) bands (F(6, 154) = 1.273, p = .286, ηp² = .04). Therefore, α and ɣ bands were excluded from subsequent analyses.

Although a main effect of age was observed for RP (β₂), post-hoc pairwise comparisons showed significant differences in only a limited number of group comparisons. Accordingly, subsequent analyses focused exclusively on the RP (θ) and RP (β₁) bands, which showed more consistent age-related effects.

Topographic maps visualization showed an age-related increase in theta band power (RP (θ); 4–8 Hz) across all age groups (F(6, 154) = 11.85, p < .001, ηp² = .31) (see Fig. 4a). As age increases, RP (θ) progressively rises and spreads across the scalp, predominantly over centro-parietal and posterior regions, resulting in larger deviations from baseline in both Go and NoGo conditions. Additionally, significant differences were found between task conditions (F(6, 154) = 25.19, p < .001, ηp² = .14); the NoGo exhibits an increase in RP (θ) magnitude compared to the Go condition (see Fig. 2c). This effect was found across all age groups, with no significant interaction between age and task condition (F(6, 154) = .46, p = .65, ηp² = .01).

Fig. 4.

Developmental differences in theta band relative power (RP (θ)): topographic maps and statistical comparisons. a) Topographic maps of baseline-relative changes in RP (θ) (4–8 Hz) for each age cluster. b) Topographic maps showing the spatial distribution of statistically significant effects between age groups for Go (left) and NoGo (right). Color scales indicate the magnitude of changes (Fig. 4a) and significance levels (Fig. 4b).

To statistically assess age-related differences in RP (θ), pairwise comparisons were conducted between age groups separately for the Go and NoGo conditions (see Fig. 4b). Comparisons between adjacent age groups (e.g., 6 vs 7 years) showed no significant differences in either condition (ps >.815). In contrast, comparisons between groups with larger age gaps (e.g., 6 vs. 11 or 7 vs. 12) showed significant differences in both conditions (ps <.001). Although both conditions exhibit a similar age-related pattern, these effects were more pronounced in the Go condition.

Topographic map visualization of RP (β₁) revealed an age-related decrease compared to baseline across all age groups (F(6, 154) = 10.36, p < .001, ηp² = .29) (see Fig. 5a). As age increases, the reduction in RP (β₁) becomes stronger and more spatially extended, particularly over the central and posterior regions. Regarding condition effects, the reduction in RP (β₁) was more pronounced in the NoGo condition (F(1, 154) = 47.03, p < .001, ηp² = .23) (see Fig. 2d).

Fig. 5.

Developmental differences in theta band relative power (RP (β₁)): topographic maps and statistical comparisons. a) Topographic maps of baseline-relative changes in RP (β₁) (13–19 Hz) for each age cluster. b) Topographic maps showing the spatial distribution of statistically significant effects between age groups for Go (left) and NoGo (right). Color scales indicate the magnitude of changes (Fig. 5b).

Fig. 5b shows the results of the parametric t-test applied to RP (β₁). In both conditions, comparisons between widely separated age groups reveal significant differences, reflected by lower p-values. In contrast, comparisons between closely aged groups showed a marked attenuation of these effects, both in spatial distribution and statistical significance, indicating smaller differences in RP (β₁).

Interestingly, although the overall reduction in RP (β₁) is greater in the NoGo condition, the Go condition appears more sensitive to age-related changes, both in terms of cortical extent and statistical significance.

4. Discussion

The present study characterized developmental changes in electrophysiological markers of visual attention and inhibitory control in typically developing children aged 6–12 years using high-density EEG during a Go/NoGo task. This age range is critical for cognitive development, particularly the development of inhibitory control, attentional allocation, and processing speed (Anderson, 2002, Diamond, 2013, Romine and Reynolds, 2005). Across ERP and time-frequency measures, the results revealed robust age and task related changes consistent with progressive maturation of neural systems supporting executive-attentional processing. Importantly, these developmental effects were observed in both stimulus evaluation processes indexed by the P300 and ongoing oscillatory dynamics reflected in time-frequency EEG measures.

4.1. Developmental changes in the P300 component

Our ERP analysis examined the impact of age and task condition on the P300 component. As expected, NoGo trials elicited both larger P300 amplitudes and longer latencies compared to Go trials across all age groups. This observation is consistent with an extensive body of ERP research in both paediatric and adult populations demonstrating that P300 morphology is sensitive to changes in stimuli characteristics and task difficulty (Enriquez-Geppert et al., 2010, Kappenman and Luck, 2012, Peters et al., 2023, Polich, 2007, Riggins and Scott, 2020). In the context of inhibitory control paradigms, enhanced P300 amplitudes during infrequent NoGo trials are commonly interpreted as reflecting the increased attentional demands required to suppress rapid automatic responses (Bruin and Wijers, 2002, Enriquez-Geppert et al., 2010, Falkenstein et al., 1999, Salisbury et al., 2004). Furthermore, the prolonged P300 latencies observed for NoGo trials are thought to reflect the additional processing time necessary for stimulus evaluation and decision-making prior to successful response inhibition (Bruin and Wijers, 2002, Falkenstein et al., 1999, Salisbury et al., 2004, Smith et al., 2008). Thus, more demanding task conditions, elicit greater neural resource allocation and slower processing dynamics (Salisbury et al., 2004).

In terms of age, the results revealed developmental changes in ERP waveforms across childhood. Specifically, P300 amplitudes and latencies in both the Go and the NoGo conditions decreased linearly with increasing age. Our year‑by‑year analysis from ages 6–12 showed that P300 amplitude decreases until about age 9, after which it remains relatively stable (see Fig. 1). These developmental effects on the P300 were accompanied by improvements in behavioral performance on the task. Older children demonstrated higher accuracy rates and fewer omission and commission errors, indicating enhanced attentional control and more consistent response monitoring with increasing age (see Table 1). Overall, our findings align with previous research suggesting reductions in P300 amplitude and latency with age may reflect greater neural efficiency and faster cognitive processing as the brain matures (Dinteren et al., 2014, Peters et al., 2023, Riggins and Scott, 2020). The reduction in latency likely reflects faster stimulus evaluation and response selection processes, whereas the decrease in amplitude may indicate reduced load on attentional resources as cognitive control networks become more functionally specialized and efficient with development (Dinteren et al., 2014, Lamm et al., 2006).

Our findings extend previous work by Peters et al. (2023), who reported a linear decrease in P300 amplitude and latency between ages 3–5 years old in a longitudinal study using a visual Go/NoGo task. The present results indicate that this trend may continue through middle childhood. Furthermore, we provide relevant evidence that cross sectional designs with sufficient statistical power may also be able to characterize developmental changes in ERP components.

Taken together, the present results extend prior literature and are consistent with the possibility that age-related changes in P300 morphology may, at least in part, reflect the maturation of inhibitory control in childhood (Bledowski et al., 2004, Polich, 2012). However, the association is likely to be indirect, as P300 component is not only modulated in behavioral paradigms requiring response inhibition but is also elicited in a wide range of tasks involving stimulus evaluation, processing, or discrimination (Bruin and Wijers, 2002, Kappenman and Luck, 2012, Pfefferbaum et al., 1985, Polich, 2012). Furthermore, the P300 shows limited neuroanatomical specificity, with its neural generators thought to comprise a distributed fronto-parietal network rather than a single cortical source (Halgren et al., 1998, Linden, 2005, Polich, 2007). In this context, developmental changes in the P300 observed in the Go/NoGo task are more likely to reflect the maturation of broad neural systems supporting stimulus processing and evaluation.

4.2. Developmental modulation of time-frequency features

Complementary to ERP findings, time-frequency analysis revealed significant main effects of age and task condition in all examined features (MF, SE and RP). A robust task-related effect was observed, with all time-frequency measures successfully differentiated between Go and NoGo trials. Specifically, NoGo trials, reflecting response inhibition and increased load on attentional, showed a stronger time-frequency features modulation, with higher baseline-relative changes, compared to Go trials. Results showed a task-related increase in RP (θ), which reflects a shift in the balance of spectral power toward lower frequencies, as evidenced by greater reduction of MF and SE in NoGo compared to Go trials. In addition, this shift is accompanied by a higher decrease in RP (β₁) in NoGo trials, reflecting a reduction in high-frequency ranges. These findings are consistent with previous literature on response inhibition, as increased theta activity in frontal-midline regions during NoGo trials has been widely reported and linked to attentional allocation, conflict monitoring, and inhibitory control processes in adult populations (Adam et al., 2020, Beltrán et al., 2019, Casula et al., 2024, Cavanagh and Shackman, 2015, Gratton et al., 2018, Hong et al., 2020, Nigbur et al., 2011, Xu et al., 2025, Yamanaka and Yamamoto, 2009). Similarly, developmental research across early childhood and adolescence, have reported modulations in theta activity in response to behavioral paradigms that require inhibition (Morales and Buzzell, 2025). Together, this evidence suggests that enhanced theta reflects both the detection of situations requiring control and the implementation of response inhibition across the lifespan.

Regarding age-related effects, the present findings revealed a significant progressive reduction in MF and SE with age, indicating a developmental shift of spectral power toward lower frequencies. In addition, results showed an age-related modulation of power in specific frequency bands; there is an increase in RP (θ) activity and a decrease in RP (β₁) over the midfrontal region with increasing age in response to both task conditions. From a neurophysiological perspective, a spectral shift toward a specific frequency range has been associated with increased large-scale synchronization of neural activity and enhanced coordination across distributed neuronal populations (Buzsáki and Draguhn, 2004). In this context, synchronization refers to neural activity that is more concentrated within specific frequency bands, resulting in a more coherent and less dispersed signal. Thus, the reduction in MF and SE with increasing age may indicate a developmental reorganization and optimization of neural dynamics in response to the functional demands of the Go/NoGo task (Morales and Buzzell, 2025).

Importantly, our results further support evidence that these theta activity modulations increase in magnitude with age (Adam et al., 2020, Liu et al., 2014, Xu et al., 2025). Thus, our results reinforce that enhanced midline theta may serve as a neuroelectrical signature of the maturation of inhibition cognitive control.

To further explore developmental effects, Fig. 2 in Appendix A illustrates the p-values in a logarithmic scale of each age group compared to the 6 years old group, obtained by selecting the lowest decile of p-values across channels and computing their median for each age group. A clear developmental pattern can be observed, with significant differences emerging from 8 years in the Go condition and from 9 years in the NoGo condition. This suggests a progressive developmental shift in time-frequency features, with earlier changes during response execution and slightly later maturation of processes associated with inhibition.

The results further support that middle childhood is critical for cognitive development and rapid maturation, particularly of inhibitory control, attentional allocation, and processing speed (Anderson, 2002, Diamond, 2013, Romine and Reynolds, 2005). Moreover, our findings may reflect a developmental transition toward more stable and automatized executive-attentional processing during late childhood. The absence of significant differences between adjacent age groups further suggests that maturation of inhibitory control follows a gradual and continuous trajectory rather than abrupt stage-like changes. The absence of interaction between age and task condition indicates that although neural efficiency improves with age, the functional distinction between response execution and inhibition remains stable across development.

Importantly, the observed developmental trajectories may have clinical relevance for identifying atypical patterns of executive functioning in neurodevelopmental disorders such as ADHD or autism spectrum disorder, conditions frequently associated with altered P300 responses and abnormal theta/beta dynamics.

In contrast to previous studies with lower statistical power to assess age-specific developmental changes, our sample size enabled a year-by-year analysis, revealing significant effects at specific ages. This allowed the identification of potentially relevant time windows, particularly at 8–9 years of age, which may reflect a relevant point in the development of inhibition and cognitive control. Furthermore, the analysis of time-frequency features offers valuable complementary information about the neural dynamics associated with cognitive functions, and expands the current literature, which has predominantly focused on ERP components.

4.2.1. Limitations and future directions

This study has several limitations that should be considered when interpreting the present findings. One limitation is the broad age range of the sample (from 6 to 12 years). Although the task was adapted to be appropriate across this range, it is possible that participants aged 10 years and older might have near ceiling performance, suggesting that this task may not have been sufficiently challenging for older age groups.

Additionally, the cross-sectional design limits conclusions regarding intra-individual developmental trajectories over time. Longitudinal studies would be necessary to determine the temporal stability and predictive value of these electrophysiological markers across development. Furthermore, because the current analysis focuses on a visual Go/NoGo task, the generalizability of our findings to other behavioral paradigms and contexts may be limited. Future research should explore whether similar patterns are observed in other behavioral paradigms with other stimuli modalities. Along with, additional ERP components could be examined to explore their potential functional roles.

Although the present analyses focused on broad frequency bands and fronto-central ROIs, future work incorporating source localization and functional connectivity analyses may provide more precise characterization of the neural networks underlying developmental changes in inhibitory control.

Despite these limitations, the present findings provide robust evidence that maturation of inhibitory control during childhood is associated with coordinated changes in ERP dynamics and oscillatory activity, reflecting increasing neural efficiency and functional organization of executive-attentional systems.

5. Conclusion

This study examined the neural correlates of visual attention and inhibition using ERP and time-frequency analyses during a visual Go/NoGo task in typically developing children aged 6–12 years. Our results demonstrate significant maturational changes in both ERP waveforms and time-frequency features across childhood in response to the task. P300 amplitudes and latencies showed a linear decrease with age in both conditions, accompanied by improvements in behavioral performance. The reduction in P300 latency may reflect faster stimulus evaluation and response selection processes, whereas decreases in P300 amplitude could indicate greater neural efficiency and reduced load on attentional resources. Time-frequency analysis revealed a developmental shift of spectral power toward lower frequencies with increasing age in both conditions, reflected by reductions in MF and SE, increased RP (θ) and decreased RP (β₁) over mid‑central electrodes. Together with stronger differentiation between Go/NoGo task in older children, the findings reflect the maturation of large-scale coordination mechanisms and may indicate greater functional specialization, network stabilization, and more flexible allocation of attentional resources during executive control.

Overall, this study highlights the value of combining ERP and time-frequency approaches to characterize, respectively, the developmental changes in stimulus processing and the maturation of cognitive control during childhood. Therefore, may provide critical insight for identifying sensitive periods for intervention and educational support in neurodevelopmental disorders characterized by executive dysfunction, such as ADHD and ASD.

Funding

This work was supported by grants from the Instituto de Salud Carlos III, grants (PI22/01441), Ajuts a la Recerca IRSJD - Carmen de Torres 2023 and Torrons Vicens, and grant PID2024–155526OB-I00 funded by MICIU/AEI/10.13039/501100011033 and by the European Regional Development Fund (ERDF), ‘A way to make Europe’.

CRediT authorship contribution statement

Carmen Fons: Writing – review & editing, Validation, Supervision, Project administration, Investigation, Funding acquisition, Conceptualization. Mañanas Miguel Angel: Writing – review & editing, Validation, Investigation. Elena Condominas: Methodology, Data curation. Alejandro Bachiller: Writing – review & editing, Validation, Methodology, Investigation, Conceptualization. Anna López-Sala: Writing – review & editing, Validation. Alonso-Colmenero Itziar: Writing – review & editing, Validation. Berta Mateu-Yus: Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Júlia Estorach-Segarra: Writing – review & editing, Writing – original draft, Visualization, Validation, Formal analysis. Anna Aguilar-Ros: Writing – review & editing, Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Conceptualization.

Declaration of Competing Interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests. Carme Fons reports a relationship with Longboard Pharmaceuticals Inc that includes: consulting or advisory. Carme Fons reports a relationship with Jazz Pharmaceuticals Inc that includes: speaking and lecture fees. Carme Fons reports a relationship with UCB Pharma SA that includes: speaking and lecture fees. Carme Fons reports a relationship with Nutricia Ltd that includes: speaking and lecture fees. Anna Aguilar-Ros reports a relationship with Nutricia Ltd that includes: speaking and lecture fees. Anna Lopez-Sala reports a relationship with Jazz Pharmaceuticals Inc that includes: speaking and lecture fees. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

Supplementary material

Supplementary material

Supplementary material

Data availability

Data will be made available on request.