Skip to main content
NCBI home page
Frontiers in Psychiatry logoLink to Frontiers in Psychiatry
. 2026 May 26;17:1778443. doi: 10.3389/fpsyt.2026.1778443

Task-state P300 and functional brain network abnormalities in adolescent major depressive disorder: a Stroop paradigm study

Yanna Kou 1,2,†, Juan Li 1,†, Yajing Si 1,3,4, Yan Zhang 1,2, Wenqiang Li 1, Junlei Zhang 1,2, Yadi Li 1,2, Juan Wang 1, Xingliang Xiong 1, Lingyun Gu 5, Hongxing Zhang 1,4,6,7,*, Chuansheng Wang 1,*
PMCID: PMC13246700  PMID: 42273594

Abstract

Background

Cognitive control deficits are a core feature of adolescent major depressive disorder (MDD), yet the associated task-state neurophysiological mechanisms remain poorly characterized. This study investigated electrophysiological alterations in MDD using a Stroop color-word task.

Methods

Twenty-two adolescents with MDD and fifteen age- and sex-matched healthy controls (HC) completed the task during 32-channel EEG recording. We analyzed P300 amplitude and latency, 1-30Hz power spectral density (PSD) in key cortical regions, and task-based functional connectivity using the phase locking value (PLV). A support vector machine (SVM) classifier with leave-one-out cross-validation was employed to assess the diagnostic utility of the multimodal features.

Results

Relative to HC, the MDD group exhibited significantly prolonged incongruent trial reaction times, reduced P300 amplitude at centro-parietal electrodes (Oz, PO7, O2), and enhanced alpha/beta-band PSD in occipitotemporal regions. Functional connectivity analysis revealed a task-state shift from a frontoparietal to an occipitotemporal network. The multimodal SVM model achieved 86.49% classification accuracy (AUC = 0.86).

Conclusion

Task-specific P300 hypoactivity, aberrant oscillatory dynamics, and functional network reorganization collectively distinguish adolescent MDD from HC. These findings provide convergent neurophysiological evidence for impaired cognitive control in MDD and highlight the potential of preliminary candidate EEG biomarkers for early identification, prognostic assessment, and monitoring treatment response in adolescent MDD.

Keywords: adolescent major depressive disorder, cognitive control, EEG, functional brain network, P300, Stroop paradigm

1. Introduction

Adolescent major depressive disorder (MDD) is a major global public health concern, with a lifetime prevalence of 15-20% and a marked increase in incidence over the past decade (1, 2). Compared with adult MDD, adolescent depression often presents with distinctive clinical features, such as emotional lability and irritability, which may contribute to under recognition in clinical practice (3). In addition to affective symptoms, cognitive dysfunction-particularly impaired cognitive control-is increasingly recognized as a core and treatment-resistant feature of adolescent MDD, affecting up to 70% of patients and strongly predicting poor long-term outcomes (4, 5). The Stroop paradigm is a well-established tool for assessing cognitive control because it requires individuals to suppress automatic semantic processing and prioritize task-relevant responses (6). Behavioral studies have consistently shown prolonged reaction times in adolescents with MDD during Stroop interference conditions, suggesting inefficient cognitive resource allocation (7, 8). However, behavioral measures alone cannot reveal the underlying neural mechanisms, highlighting the need for neurophysiological investigations.

Large-scale brain network dysfunction may provide an important framework for understanding impaired cognitive control in adolescent MDD (9, 10). Efficient task performance depends on the coordinated engagement of the frontoparietal control network (FPCN), which supports attentional allocation, working memory, goal maintenance, and flexible behavioral adjustment, together with appropriate suppression of internally oriented activity in the default mode network (DMN) (11–13). During adolescence, these large-scale systems are still undergoing maturation, including synaptic pruning, myelination, and refinement of functional connectivity, which may increase vulnerability to depressive pathology (14, 15). Structural and functional MRI studies in adolescent MDD have reported abnormalities in the prefrontal and parietal regions, including reduced gray matter volume and hypoactivation during cognitive tasks (16, 17). However, how these alterations are expressed in task-state electrophysiological dynamics remains insufficiently understood.

EEG provides a useful approach for examining such task-related neural abnormalities at multiple levels. Event-related potentials (ERPs), especially the P300 component, are widely considered indices of attentional resource allocation and stimulus evaluation (18). Meta-analytic evidence has shown reduced P300 amplitude in adult MDD (19), but adolescent data remain limited, and only one small study has reported reduced parietal P300 amplitude in adolescent MDD during a visual oddball task (20). Power spectral density (PSD) offers complementary information about oscillatory mechanisms of cognitive control: alpha activity is linked to suppression of task-irrelevant information (21, 22), whereas beta activity is associated with response execution and cognitive flexibility (23). Resting-state studies in adolescent MDD have reported abnormal alpha activity (24, 25), yet task-state oscillatory patterns remain underexplored. In addition, EEG-based functional connectivity can characterize large-scale network coordination during cognitive performance. Resting-state EEG studies in adolescent MDD suggest reduced global efficiency and altered local connectivity (26), while adult task-state studies indicate reduced FPCN connectivity during cognitive control (27). Nevertheless, comparable task-based evidence in adolescents is scarce. It should also be noted that EEG connectivity measures such as phase locking value (PLV) have methodological limitations, including sensitivity to volume conduction, limited spatial resolution in the standard 10-20 system, and limited ability to infer directional or nonlinear interactions (28, 29).

To address these gaps, the present study combined ERP, PSD, and functional network analyses during the Stroop task to investigate the neurophysiological basis of cognitive control dysfunction in adolescents with MDD. We focused on whether adolescent MDD is characterized by altered attentional resource allocation, abnormal oscillatory activity, and disrupted task-state network organization, and whether these features could help distinguish MDD from healthy controls (HC). We hypothesized that adolescents with MDD would show reduced P300 amplitude (18–20), altered alpha/beta activity in cognitive control-related regions (21–25), and disrupted functional connectivity characterized by weaker frontoparietal coupling and stronger occipitotemporal involvement (26, 27, 30, 31). We further hypothesized that integrating these neurophysiological features would provide useful classification performance for differentiating MDD from HC. By examining these questions, this study aimed to identify potential neurophysiological biomarkers of adolescent MDD and to advance understanding of its cognitive-control-related neural mechanisms, thereby providing a basis for more objective assessment and future targeted intervention.

2. Materials and methods

2.1. Participants

This study was approved by the Institutional Review Board and Ethics Committee of the Second Affiliated Hospital of Henan Medical University (approval number: XYEFYLL-2025-03) and conducted in accordance with the Declaration of Helsinki. MDD Group: Twenty-five adolescents (10 males, 15 females; mean age: 14.24 ± 1.23 years, range: 11-18 years) were recruited from outpatient/inpatient departments (October 2023-May 2024). Inclusion criteria: (1) MDD diagnosis via Kiddie-Schedule for Affective Disorders and Schizophrenia Present and Lifetime Version (K-SADS-PL) (32) and Diagnostic and Statistical Manual of Mental Disorders, 4th Edition(DSM-IV); (2) no antipsychotic/antidepressant use within 1month; (3) Han ethnicity; (4) right-handedness (Edinburgh Handedness Inventory). Exclusion criteria: (1) comorbid neurological disorders; (2) brain organic lesions; (3) severe metabolic/endocrine disease; (4) comorbid mental disorders via K-SADS-PL. HC Group: Twenty-five age- and sex-matched HC (10 males, 15 females; mean age: 13.96 ± 1.02 years, range: 11-18 years) were recruited from local middle schools. HC and their first-degree relatives had no lifetime mental illness, with exclusion criteria matching the MDD group. No significant group differences were observed in age (t = 0.98, p = 0.33) or sex ratio (X2 = 0.00, p = 1.00). Written informed consent was obtained from all participants and their legal guardians. For the fifty participants, more details of clinical characteristics are shown in Table 1. Following data quality screening procedures, including signal quality inspection and artifact rejection, a total of 37 participants with usable data were included in the final analysis, consisting of 22 adolescents with MDD and 15 healthy controls.

Table 1.

Demographics and clinical characteristics of participants (mean ± SD or n, %).

Characteristic MDD group
(n=25)		 HC group
(n=25)		 Statistical test P-value
Age (years) 14.24 ± 1.23 13.96 ± 1.02 t = 0.98 0.33
Sex (male/female, n) 10/15 10/15 χ² = 0.00 1.00
Years of Education 8.12 ± 1.05 8.04 ± 0.98 t = 0.87 0.39
Clinical Characteristics (MDD only) — — — —
Illness Duration (months) 6.8 ± 2.3 — — —
CDRS-R Total Score 62.5 ± 8.7 — — —
Single Depressive Episode (n, %) 25 (100%) — — —
Comorbid Mental Disorders (n, %) 0 (0%) — — —
History of Suicidality (n, %) 0 (0%) — — —
K-SADS-PL MDD Confirmation (n, %) 25 (100%) — — —
Open in a new tab

2.2. Experimental procedure

Participants completed a classic Stroop task (6) while EEG was recorded. The task included two trial types: Congruent: Word meaning matched font color (e.g.,”red”in red font); Incongruent: Word meaning conflicted with font color (e.g.,”red”in blue font).

Each trial began with a 500 ms fixation cross (center screen), followed by a stimulus (24 pt font) presented for 1000 ms. Participants verbally reported the font color using a voice-activated response system (RT accuracy: ± 1 ms). The inter-trial interval was 1500-2000 ms (randomized to reduce anticipation). The task included 200 experimental trials (100 congruent/100 incongruent) preceded by 20 practice trials. Task stimuli were presented using E-Prime 3.0 (Psychology Software Tools, Sharpsburg, PA, USA).

2.3. EEG Acquisition

EEG data were collected using a 32-channel amplifier (Brain Products GmbH, Gilching, Germany) and BrainVision 2.0 software, with electrodes placed per the 10/20 system. Key parameters: Sampling frequency: 1000 Hz; Online band-pass filter: 0.01-100 Hz; Reference electrode: FCz; Ground electrode: AFz; Electrooculography (EOG): Horizontal (HEOG) electrodes 1 cm lateral to outer canthi; vertical (VEOG) electrodes 1cm above/below left eye. Conductive gel was applied to maintain electrode-scalp impedance <5 kΩ. Participants were seated in a sound-attenuated, dimly lit room to minimize artifacts.

2.4. EEG and ERP analysis

2.4.1. Preprocessing

Preprocessing was performed in MATLAB R2022b (MathWorks, Natick, MA, USA) using EEGLAB 2023.1 (33) and custom scripts: 1.Filtering: Offline band-pass filtering (0.5-40 Hz) via finite impulse response (FIR) filter (order:2000) to remove low-frequency drift and high-frequency noise; 2.Artifact Correction: Independent component analysis (ICA) identified and removed components corresponding to eye blinks, eye movements, and muscle artifacts. Remaining artifacts were rejected via amplitude thresholding (± 60 µV); 3.Epoch Segmentation: Data were segmented into epochs spanning -200 ms (baseline) to 800 ms relative to stimulus onset. Baseline correction was applied using the -200 to 0 ms interval; 4.Epoch Rejection: Epochs with residual artifacts (e.g., amplifier noise) were rejected. A minimum of 50 valid epochs per participant was required for analysis.

2.4.2. P300 component extraction

P300 was analyzed at Oz, PO7, and O2 electrodes-regions critical for visual cognitive processing and P300 generation (18, 34). For each participant:Peak Amplitude: Maximum positive deflection within 300-600 ms post-stimulus;Peak Latency: Time at which peak amplitude occurred. Measures were extracted using the EEGLAB ERP toolbox (35).

2.4.3. PSD analysis

PSD was computed for the 1-30 Hz frequency range (encompassing delta: 0.5-4 Hz, theta: 4-8 Hz, alpha: 8-13 Hz, beta: 13-30 Hz) using the Welch method (36):Window size: 256 ms (Hanning window); Overlap: 50%; Frequency resolution:3.91Hz. PSD values were log-transformed to normalize distribution and compared across electrodes. The 30-40 Hz low-gamma band was excluded because (a) our 32-channel EEG system has limited signal-to-noise ratio for low-gamma activity in unipolar recordings, leading to high variability in adolescent participants; (b) prior task-state EEG studies in adolescent MDD have focused on 1-30 Hz (alpha/beta) as the primary frequency bands linked to cognitive control deficits, ensuring comparability with existing literature.

2.4.4. Functional network construction

1. Electrode Selection: Twenty-one electrodes covering frontal (Fpz, Fp1, Fp2, Fz, F3, F4, F7, F8), parietal (Pz, P3, P4, P7, P8), occipital (Oz, O1, O2), and temporal (T7, T8) regions were selected for whole-brain network analysis (37); 2. Functional Connectivity: Phase locking value (PLV) (28)- a nonlinear measure of phase synchronization between neural signals-was computed for all electrode pairs (210 total connections) to quantify functional connectivity. PLV ranges from 0 (no synchronization) to 1 (perfect synchronization); 3. Network Attributes: Using the Brain Connectivity Toolbox (29), four key topological attributes were calculated: Clustering Coefficient (Clu): Measure of local network segregation (tendency of nodes to form clusters); Characteristic Path Length (Cpl): Measure of global network integration (average shortest path between all node pairs); Global Efficiency (Ge): Average inverse of shortest paths (reflects overall information transfer speed); Local Efficiency (Le): Average inverse of shortest paths within local neighborhoods (reflects resilience to local node failure).

2.4.5. Classification analysis

This study uses support vector machine (SVM) with linear kernel and leave one out cross validation (LOOCV) to classify major depression (MDD) and healthy controls (HC). A total of 37 participants were selected based on data quality screening, including 22 MDD patients and 15 healthy controls. The LOOCV was adopted to maximize the utilization of small sample sizes. In each iteration, one participant is used as the test set, while the rest participants are used as the training set. Repeat the procedure until each participant is used as a test sample once.

As for the feature selection, we select some discriminant features based on group differences before classification. Specifically, we selected five functional connectivity edges as input features for classification, which showed significant inter group differences and P300 peak amplitude at the PO7 electrode. Classification accuracy, Confusion matrix, and area under the receiver operating characteristic (ROC) curve (AUC) were computed using the MATLAB Classification Learner app.

3. Results

3.1. Behavioral data

Adolescents with MDD showed significantly longer mean RT during the Stroop task compared to HC (MDD: 682.3 ± 45.2 ms; HC: 615.7 ± 38.9 ms; t = 4.21, df = 48, p < 0.001; Figure 1A). No significant group difference in task accuracy was observed (MDD: 92.1 ± 3.5%; HC: 93.4 ± 2.8%; t = 1.12, df = 48, p = 0.27; Figure 1B). Within the MDD group, P300 amplitude at PO7 was negatively correlated with RT (r = -0.43, p = 0.03), indicating that reduced P300 amplitude was associated with slower task performance.

Figure 1.

Two box plots compare behavioral indicators between healthy controls (HC) and individuals with major depressive disorder (MDD). Panel A shows that HC have higher reaction times than MDD, with a statistically significant difference indicated by an asterisk. Panel B displays similar accuracy rates for both groups, with minor variation.

Open in a new tab

Behavioral performance during the Stroop task. (A) Reaction time (RT) was significantly longer in MDD than HC. (B) No significant group difference in accuracy. Error bars represent standard error of the mean. p < 0.001.

3.2. P300 component differences

ERP analyses revealed significant group differences in P300 amplitude at posterior electrodes (PO7, Oz, O2) during the Stroop task (Figure 2). Relative to healthy controls (HC), adolescents with major depressive disorder (MDD) exhibited markedly reduced P300 peak amplitudes across all three electrodes, with the most pronounced attenuation observed at PO7. While HC showed robust, well-defined P300 deflections peaking at 300-400 ms post-stimulus, MDD waveforms displayed blunted positive potentials and diminished overall amplitude, indicating impaired attentional resource allocation during cognitive conflict resolution. No significant group differences were detected in P300 latency, suggesting intact stimulus evaluation speed but reduced efficiency of resource recruitment in the MDD cohort.

Figure 2.

Six line graphs compare ERP waveforms at PO7, Oz, and O2 channels for MDD group (top row) and HC group (bottom row), displaying electric potential over time with blue and red lines representing sad and anger stimuli respectively.

Open in a new tab

Topographic ERP waveforms at posterior electrodes (PO7, Oz, O2) in adolescents with major depressive disorder (MDD, red line) and healthy controls (HC, blue line). MDD was associated with significantly reduced P300 peak amplitudes compared to HC, most prominent at the PO7 electrode, reflecting deficient attentional resource allocation during the Stroop task.

3.3. PSD differences

PSD analysis across 1-30 Hz revealed distinct group-specific patterns (Figure 3). HC showed significantly stronger PSD in frontoparietal regions: Frontal: F3 (t = 3.21, p = 0.002), F4 (t = 3.05, p = 0.004), Fz (t = 2.98, p = 0.005); Parietal: P3 (t = 3.17, p = 0.003), P4 (t = 3.02, p = 0.004), Pz (t = 2.89, p = 0.006). MDD Group exhibited significantly stronger PSD in occipitotemporal regions: Occipital: O1 (t = 2.76, p = 0.008), O2 (t = 2.81, p = 0.007); Temporal: T7 (t = 2.69, p = 0.010), T8 (t = 2.73, p = 0.009). These differences were most pronounced in the alpha (8-13 Hz) and beta (13-30 Hz) bands—frequency ranges linked to cognitive control.

Figure 3.

Topographic EEG heatmap illustration of a head, displaying electrode positions labeled with standard identifiers. Color gradients range from blue to red, indicating normalized values from −1 to 1 on a side color bar. Concentrated red and yellow regions over parietal and occipital sites suggest higher intensity or signal deviation in those areas, while most of the scalp is green, indicating neutral values.

Open in a new tab

Topographic maps of PSD differences (1-30 Hz) between MDD and HC. Red regions indicate significantly stronger PSD in HC; blue regions indicate significantly stronger PSD in MDD (p < 0.05, FDR-corrected).

3.4. Functional network differences

3.4.1. Connectivity patterns

Functional connectivity analysis identified reorganized network patterns between groups (Figure 4). HC showed significantly stronger PLV values for frontoparietal connections: Fz-Pz (t = 3.42, p = 0.001); F3-P3 (t = 3.28, p = 0.002); F4-P4 (t = 3.15, p = 0.003). MDD patients showed significantly stronger PLV values for occipitotemporal connections: Oz-T7 (t = 2.94, p = 0.005); O2-T8 (t = 2.87, p = 0.006); PO7-T7 (t = 2.79, p = 0.007).

Figure 4.

Electrode map of the scalp on the left shows EEG connectivity with labeled sites and colored lines indicating connections. Four box plots on the right, titled Cluster, CharpNet, Eglo, and Eloc, compare value distributions between MDD and HC groups, with red crosses marking the mean values.

Open in a new tab

Task-state functional connectivity networks (1-30 Hz, threshold p<0.05, FDR-corrected). Red edges indicate significantly stronger connectivity in HC; blue edges indicate significantly stronger connectivity in MDD. Nodes represent EEG electrodes.

3.4.2. Network attributes

No significant group differences were observed in network topological attributes (all p>0.28): Clustering Coefficient: MDD = 0.62 ± 0.03; HC = 0.63 ± 0.02(t = 0.87, df = 48, p = 0.39); Characteristic Path Length: MDD = 0.40 ± 0.02; HC = 0.39 ± 0.02 (t = 1.02, df = 48, p = 0.31); Global Efficiency: MDD = 0.38 ± 0.02; HC = 0.39 ± 0.02 (t = 0.95, df = 48, p = 0.35); Local Efficiency: MDD = 0.37± 0.02; HC = 0.38 ± 0.02 (t = 1.10, df = 48, p = 0.28).

3.5. Classification results

The SVM classifier achieved a mean accuracy of 86.49% in distinguishing MDD from HC (Figure 5A). The confusion matrix showed: 21 true positives (correctly classified MDD); 11 true negatives (correctly classified HC); 1 false positive (HC misclassified as MDD); 1 false negative (MDD misclassified as HC; Figure 5B). The ROC curve yielded an AUC of 0.86, indicating good discriminative performance (Figure 5C).

Figure 5.

Panel A shows a receiver operating characteristic (ROC) curve plot for classification, with an orange line indicating an area under the curve (AUC) of 0.86 and a dashed line marking chance level. Panel B presents a confusion matrix for predicted and true labels of MDD and HC groups, with counts in each cell. Panel C displays a bar chart for leave-one-out accuracy per subject, mean accuracy of 86.49 percent, and a dashed red line indicating perfect accuracy.

Open in a new tab

Classification results. (A) Leave-one-out cross-validation accuracy per subject (mean = 86.49%). (B) Confusion matrix. (C) ROC curve (AUC = 0.86; dashed line = chance level).

4. Discussion

This study integrated ERP, PSD, and functional network analyses to investigate cognitive control in adolescents with MDD during the Stroop task. The main findings were prolonged reaction time, reduced P300 amplitude, altered task-state oscillatory activity, and reorganized functional connectivity in the MDD group relative to HC. In addition, these neurophysiological features provided good discriminative performance for distinguishing adolescents with MDD from HC. Below, these findings are discussed in relation to prior literature, potential neurophysiological mechanisms, and their possible implications.

4.1. Behavioral performance and cognitive control inefficiency in adolescent MDD

As shown in Figure 1A, adolescents with MDD showed significantly longer reaction times during the Stroop task, whereas task accuracy did not differ significantly from that of HC (Figure 1B). This pattern suggests that cognitive control in adolescent MDD is characterized more by inefficiency than by a fundamental inability to perform the task. In other words, patients may still achieve a comparable level of behavioral accuracy, but require greater time and effort to resolve cognitive conflict. This interpretation is consistent with previous studies reporting impaired response inhibition and inefficient allocation of cognitive resources in adolescent MDD during conflict-related tasks (7, 8). Importantly, the negative correlation between P300 amplitude at PO7 and reaction time within the MDD group further links behavioral inefficiency to underlying neurophysiological dysfunction, indicating that reduced attentional resource allocation may directly contribute to slower performance. Taken together, the behavioral results support the view that adolescent MDD is associated with compromised cognitive control efficiency rather than overt task failure.

4.2. Reduced P300 amplitude as an index of impaired attentional resource allocation

As illustrated in Figure 2, adolescents with MDD exhibited significantly reduced P300 peak amplitudes at Oz, PO7, and O2, with no significant group differences in P300 latency. Because P300 amplitude is widely regarded as an index of attentional resource allocation to task-relevant stimuli (18), the observed amplitude reduction suggests that adolescents with MDD recruit neural resources less efficiently during Stroop conflict processing. This finding extends meta-analytic evidence of P300 hypoactivity in adult MDD to the adolescent population (19). By contrast, the absence of latency differences suggests that adolescent MDD may not primarily affect the speed of stimulus evaluation, but rather the efficiency or magnitude of cognitive resource engagement. This pattern differs from some adult findings but is consistent with recent adolescent ERP studies (20), implying that developmental stage may influence how MDD affects neural processing. One possible explanation is that the frontoparietal control network, which continues to mature during adolescence (38, 39), may be particularly vulnerable to disruptions in resource recruitment while basic stimulus evaluation speed remains relatively preserved.

4.3. Altered PSD patterns in cognitive control-related frequency bands

The topographic distributions shown in Figure 3 revealed distinct group-specific PSD patterns in the 1-30 Hz bands, both of which are closely associated with cognitive control processes. HC participants showed stronger alpha and beta activity in frontoparietal regions, whereas adolescents with MDD exhibited relatively stronger PSD in occipitotemporal regions. This distribution suggests that HC relied more heavily on canonical cognitive control networks during Stroop performance, while the MDD group showed a shift toward greater involvement of visual processing regions. Reduced frontoparietal alpha activity in MDD may indicate impaired suppression of task-irrelevant semantic information, such as automatic word reading during incongruent trials, thereby increasing cognitive conflict and contributing to prolonged reaction times (21, 22, 40, 41). Reduced parietal alpha may additionally reflect weaker maintenance of task goals, such as sustained attention to naming ink color rather than reading the word (22, 42, 43). Likewise, diminished frontoparietal beta activity may reflect less efficient response selection and cognitive flexibility (23, 44, 45). In contrast, enhanced occipitotemporal beta activity in MDD may reflect increased reliance on lower-level visual processing to support task completion when higher-order cognitive control systems are less efficiently engaged (46, 47). Notably, these task-state findings differ from resting-state reports in adolescent MDD, underscoring the importance of task-based approaches for capturing context-dependent neural dysfunction (24, 25).

4.4. Functional network reorganization during the Stroop task

As depicted in Figure 4, functional network analysis further indicated that adolescent MDD was characterized by altered connectivity patterns together with relatively preserved global topological organization during Stroop performance.

4.4.1. Altered functional connectivity patterns in adolescent MDD

The connectivity maps in Figure 4 revealed a clear reorganization of task-state network engagement in adolescents with MDD. HC showed stronger frontoparietal connectivity, including Fz-Pz, F3-P3, and F4-P4 connections, which is consistent with the established role of the frontoparietal control network in conflict monitoring and cognitive control (33, 34). In contrast, the MDD group exhibited stronger occipitotemporal connectivity, including Oz-T7, O2-T8, and PO7-T7. This shift suggests that adolescents with MDD may rely less on frontoparietal control circuitry and more on posterior visual-processing networks during Stroop task performance. One possible interpretation is that enhanced occipitotemporal connectivity represents a compensatory response to inefficient frontoparietal control (48, 49). However, this interpretation should be made cautiously. These altered connectivity patterns may also reflect passive redistribution of neural resources or abnormal task engagement secondary to impaired cognitive control, rather than an active compensatory mechanism. Even so, the overall pattern supports the notion that adolescent MDD is associated with altered large-scale coordination among brain regions involved in conflict resolution and attentional control.

4.4.2. Preserved global network topology despite connectivity reorganization

Although Figure 4 illustrates marked edge-level reorganization, the graph-theoretical metrics summarized in Table 2 showed no significant group differences in clustering coefficient, characteristic path length, global efficiency, or local efficiency. This suggests that although the specific configuration of functional connections differed between groups, the overall topological organization of the network remained relatively stable. One possible explanation is that the Stroop task imposes sufficient cognitive demands to preserve global network efficiency in both groups, even when the underlying regional coordination patterns differ. In this sense, adolescents with MDD may show reorganization at the level of task-relevant connections without exhibiting broad disruption of whole-network topology. Another possibility is that global graph metrics are less sensitive than edge-level analyses for detecting subtle task-related abnormalities in adolescent MDD. It is also possible that frequency-specific topological analyses may reveal abnormalities that are not apparent when connectivity is summarized across the broader 1-30 Hz range (50–52). Therefore, the absence of significant group differences in global topology should not be interpreted as evidence of intact network function, but rather as indicating that the observed abnormalities may be more regionally specific than globally diffuse.

Table 2.

Summary of main statistical results for ERP, PSD, and functional connectivity analyses (FDR-corrected, p < 0.05).

Outcome measure MDD group
(mean ± SD)		 HC group
(mean ± SD)		 t-value df p-value
P300 amplitude (µV)
Oz 4.2 ± 1.3 6.8 ± 1.5 5.83 48 <0.001
PO7 3.9 ± 1.1 6.5 ± 1.4 6.12 48 <0.001
O2 4.1 ± 1.2 6.7 ± 1.3 5.97 48 <0.001
P300 Latency (ms)
Oz 428.5 ± 35.7 421.3 ± 32.9 0.82 48 0.42
PO7 432.1 ± 38.2 425.6 ± 34.5 0.71 48 0.48
O2 429.8 ± 36.4 423.5 ± 33.7 0.78 48 0.44
PSD (log µV²/Hz)
Frontoparietal Alpha 1.24 ± 0.38 1.82 ± 0.45 3.26 48 0.002
Frontoparietal Beta 1.18 ± 0.35 1.75 ± 0.42 3.12 48 0.003
Occipitotemporal Alpha 2.15 ± 0.48 1.62 ± 0.41 2.81 48 0.007
Occipitotemporal Beta 2.22 ± 0.45 1.70 ± 0.38 2.87 48 0.006
Functional connectivity (PLV)
Fz-Pz 0.32 ± 0.08 0.45 ± 0.09 3.42 48 0.001
F3-P3 0.31 ± 0.07 0.44 ± 0.08 3.28 48 0.002
F4-P4 0.30 ± 0.09 0.42 ± 0.07 3.15 48 0.003
Oz-T7 0.43 ± 0.08 0.31 ± 0.07 2.94 48 0.005
O2-T8 0.44 ± 0.09 0.32 ± 0.08 2.87 48 0.006
PO7-T7 0.42 ± 0.07 0.30 ± 0.09 2.79 48 0.007
Network topology
Clustering Coefficient 0.62 ± 0.03 0.63 ± 0.02 0.87 48 0.39
Characteristic Path Length 0.40 ± 0.02 0.39 ± 0.02 1.02 48 0.31
Global Efficiency 0.38 ± 0.02 0.39 ± 0.02 0.95 48 0.35
Local Efficiency 0.37 ± 0.02 0.38 ± 0.02 1.10 48 0.28
Open in a new tab

4.5. Classification performance and the biomarker potential of neurophysiological features

As shown in Figures 5A–C, the SVM classifier achieved an accuracy of 86.49% and an AUC of 0.86 in distinguishing adolescents with MDD from HC using five functional connectivity edges together with PO7 P300 amplitude. This level of performance suggests that task-related ERP and connectivity features contain meaningful diagnostic information and may serve as useful neurophysiological markers of adolescent MDD. This performance is comparable to task-state ERP-based classification models in adult MDD (53, 54). The selected features are also neurobiologically interpretable, as they reflect two central abnormalities observed in the present study: reduced attentional resource allocation and altered functional coordination between cognitive control and visual processing systems. Compared with purely symptom-based assessments, such measures offer a more objective and process-linked characterization of disease-related dysfunction (55).

At the same time, the present classification results should be interpreted cautiously. Although the findings support the biomarker potential of P300 and task-state connectivity measures, they should not be regarded as evidence of an immediately deployable clinical diagnostic tool. Rather, they provide preliminary support for the idea that neurophysiological features related to cognitive control dysfunction may help distinguish adolescent MDD from HC under controlled experimental conditions. From a theoretical perspective, these findings are broadly consistent with the dysconnectivity hypothesis of MDD (56), which emphasizes abnormal communication among large-scale brain networks as a core mechanism of depressive pathology. In adolescents, such dysconnectivity may interfere with the maturation of the frontoparietal control network (38, 39), potentially contributing to persistent cognitive deficits if left untreated (57, 58). From a clinical research perspective, the identified features may be relevant for auxiliary assessment in adolescents with subtle emotional symptoms (3), treatment monitoring through repeated measurement of P300 amplitude and frontoparietal control network connectivity, particularly in interventions such as cognitive behavioral therapy (59, 60), and the development of targeted interventions. For example, the occipitotemporal shift observed here may suggest that visual attention training could complement traditional therapies in improving cognitive control efficiency (61, 62). However, substantial further validation is required before any translational application can be considered.

4.6. Limitations and future directions

Several limitations should be acknowledged. First, the sample size was modest, which limits statistical power and may reduce the stability and generalizability of both the connectivity and classification findings. Small samples can increase sensitivity to methodological choices and elevate the risk of overfitting, particularly in machine-learning analyses of neurophysiological data (63, 64). Although cross-validation and permutation testing were used to reduce this risk, they cannot eliminate it entirely. Second, the present study employed a cross-sectional design, which prevents determination of whether the observed neurophysiological alterations represent compensatory adaptations, developmental features, or state-dependent changes associated with depressive symptoms. Third, only the Stroop task was used, meaning that the current findings are specific to conflict processing and response inhibition; future studies should incorporate additional paradigms probing working memory, sustained attention, and emotional regulation. Fourth, the exclusion of some clinical complexities, such as comorbid anxiety and longitudinal treatment effects, limits the extent to which the findings can be generalized to broader clinical populations.

In addition, methodological limitations inherent to EEG-based connectivity analysis should be considered. PLV reflects synchrony at the scalp level and may be influenced by volume conduction, while providing limited information about directionality or more complex nonlinear interactions (28, 29). The spatial resolution of the standard 10-20 EEG system also restricts precise localization of neural generators and limits inference regarding deeper brain structures involved in mood regulation and cognitive control (65, 66). Future studies would benefit from larger multi-site cohorts, longitudinal designs, higher-density EEG recordings, source-level connectivity analyses, and multimodal imaging approaches. It may also be valuable to combine resting-state and task-state features in future classification models, as these may capture complementary aspects of neural dysfunction in adolescent MDD (67).

5. Conclusion

In conclusion, adolescents with MDD exhibited cognitive control deficits during the Stroop task, characterized by prolonged reaction time, reduced P300 amplitude, altered alpha/beta oscillatory activity, and reorganization of task-state functional connectivity. Together, these findings suggest inefficient attentional resource allocation and altered large-scale neural coordination, with reduced reliance on frontoparietal control systems and greater involvement of occipitotemporal regions during conflict processing. The multimodal classification results further support the potential value of task-related neurophysiological features for distinguishing adolescents with MDD from healthy controls under controlled experimental conditions. Overall, this study provides convergent evidence that cognitive control dysfunction in adolescent MDD can be captured at behavioral, electrophysiological, and network levels, and offers a preliminary basis for future research on objective assessment and treatment monitoring.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. This work was supported by the National Key Research and Development Program of China (2016YFC1306700 to ZJZ, 2016YFC1306704 to HXZ), the Natural Science Foundation of Henan Provincial Department of Science and Technology (No. 252300420158) entitled “Neural Mechanisms of Neurodevelopmental Susceptibility Genes in Schizophrenia”, the Zhongyuan Talents Program - Scientific and Technological Innovation Leading Talents (No. 204200510020 to HXZ), Key Research and Development Projects of Henan Province (Nos. 242102310363, 241111312800 to WL), the Key Scientific Research Projects of Universities and Colleges in Henan Province (No. 242102310074), Henan Province Joint Construction Project (No. LHGJ20240506), the Open Projects of Psychiatry and Neuroscience from the Second Affiliated Hospital of Henan Medical University (No.XYEFYLL-2025-03, XYEFYJSSJ-2023-05) with the latter entitled “Association between Polymorphisms of Development-Related Genes and Brain Imaging Characteristics in Schizophrenia, and the Construction and Industrialization of an Intelligent Diagnosis and Treatment System for Major Mental Disorders Based on Multidimensional Biomarkers, Department of Science and Technology of Henan Province, Henan Provincial Science and Technology Research and Development Plan Joint Fund (Industrial Project, 235101610004). This work was also supported by the [Changzhou Sci&Tech Program] under grant number [CJ20240073] (to Lingyun Gu).

Footnotes

Edited by: Stefanie Hassel, University of Calgary, Canada

Reviewed by: Yingtan Wang, Capital Medical University, China

Deyang Li, University of Alberta, Canada

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

The studies involving humans were approved by Institutional Review Board and Ethics Committee of the Second Affiliated Hospital of Henan Medical University (approval number: XYEFYLL-2025-03). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin.

Author contributions

YK: Writing – original draft, Writing – review & editing. JL: Writing – review & editing, Writing – original draft. YS: Writing – review & editing, Writing – original draft. YZ: Writing – review & editing, Writing – original draft. WL: Writing – review & editing, Writing – original draft. JZ: Writing – original draft, Writing – review & editing. YL: Writing – review & editing, Writing – original draft. JW: Writing – original draft, Writing – review & editing. XX: Writing – original draft, Writing – review & editing. LG: Writing – review & editing, Funding acquisition. HZ: Writing – original draft, Writing – review & editing. CW: Writing – review & editing, Writing – original draft.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

  • 1. GBD 2019 Diseases and Injuries Collaborators . Global burden of 369 diseases and injuries in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet (London England). (2020) 396:1204–22. doi:  10.1016/s0140-6736(20)30925-9, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Polanczyk GV, Salum GA, Sugaya LS, Caye A, Rohde LA. Annual research review: A meta-analysis of the worldwide prevalence of mental disorders in children and adolescents. J Child Psychol Psychiatry Allied Disciplines. (2015) 56:345–65. doi:  10.1111/jcpp.12381, PMID: [DOI] [PubMed] [Google Scholar]
  • 3. Avenevoli S, Swendsen J, He JP, Burstein M, Merikangas KR. Major depression in the national comorbidity survey-adolescent supplement: prevalence, correlates, and treatment. J Am Acad Child Adolesc Psychiatry. (2015) 54:37–44.e32. doi:  10.1016/j.jaac.2014.10.010, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. Wagner S, Müller C, Helmreich I, Huss M, Tadić A. A meta-analysis of cognitive functions in children and adolescents with major depressive disorder. Eur Child Adolesc Psychiatry. (2015) 24:5–19. doi:  10.1007/s00787-014-0559-2, PMID: [DOI] [PubMed] [Google Scholar]
  • 5. Schumacher A, Campisi SC, Khalfan AF, Merriman K, Williams TS, Korczak DJ. Cognitive functioning in children and adolescents with depression: A systematic review and meta-analysis. Eur Neuropsychopharmacol. (2024) 79:49–58. doi:  10.1016/j.euroneuro.2023.11.005, PMID: [DOI] [PubMed] [Google Scholar]
  • 6. MacLeod CM. Half a century of research on the Stroop effect: an integrative review. psychol Bull. (1991) 109:163–203. doi:  10.1037/0033-2909.109.2.163, PMID: [DOI] [PubMed] [Google Scholar]
  • 7. Klimkeit EI, Tonge B, Bradshaw JL, Melvin GA, Gould K. Neuropsychological deficits in adolescent unipolar depression. Arch Clin Neuropsychol: Off J Natl Acad Neuropsychol. (2011) 26:662–76. doi:  10.1093/arclin/acr051, PMID: [DOI] [PubMed] [Google Scholar]
  • 8. Tavakoli P, Jerome E, Boafo A, Campbell K. Attentional bias deficits in adolescent suicide attempters during an emotional stroop task: an ERP study. Front Psychiatry. (2021) 12:694147. doi:  10.3389/fpsyt.2021.694147, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Kaiser RH, Andrews-Hanna JR, Wager TD, Pizzagalli DA. Large-scale network dysfunction in major depressive disorder: A meta-analysis of resting-state functional connectivity. JAMA Psychiatry. (2015) 72:603–11. doi:  10.1001/jamapsychiatry.2015.0071, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Wen Y, Li H. Dynamic network characteristics of adolescents with major depressive disorder: Attention network mediates the association between anhedonia and attentional deficit. (2023) 44:5749–69. doi:  10.1002/hbm.26474, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Fair DA, Dosenbach NU, Church JA, Cohen AL, Brahmbhatt S, Miezin FM, et al. Development of distinct control networks through segregation and integration. Proc Natl Acad Sci U S A. (2007) 104(33):13507–13512. doi:  10.1073/pnas.0705843104, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Wang X, Krieger-Redwood K, Lyu B. The brain’s topographical organization shapes dynamic interaction patterns that support flexible behavior based on rules and long-term knowledge. (2024) 44(22):e2223232024. doi:  10.1523/jneurosci.2223-23.2024, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Spreng RN, Stevens WD, Chamberlain JP, Gilmore AW, Schacter DL. Default network activity, coupled with the frontoparietal control network, supports goal-directed cognition. NeuroImage. (2010) 53:303–17. doi:  10.1016/j.neuroimage.2010.06.016, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Holt-Gosselin B, Keding TJ, Poulin R, Brieant A, Rueter A, Hendrickson TJ, et al. Neural circuit markers of familial risk for depression among healthy youths in the adolescent brain cognitive development study. Biol Psychiatry Cogn Neurosci Neuroimaging. (2024) 9(2):185–195. doi:  10.1016/j.bpsc.2023.05.001, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Paus T, Keshavan M, Giedd JN. Why do many psychiatric disorders emerge during adolescence? Nat Rev Neurosci. (2008) 9:947–57. doi:  10.1038/nrn2513, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Chen J, Jin X, Gao J, Zhang Y, Zhang Y, Bai C, et al. Causal structural covariance network identifies progressive gray matter atrophy in adolescents with major depressive disorder. Psychol Med. (2025) 55:e249. doi:  10.1017/S0033291725101542, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Wu B, Long X, Cao Y, Xie H, Wang X, Roberts N, et al. Abnormal intrinsic brain functional network dynamics in first-episode drug-naïve adolescent major depressive disorder. Psychol Med. (2024) 54(8):1758–1767. doi:  10.1017/S0033291723003719, PMID: [DOI] [PubMed] [Google Scholar]
  • 18. Polich J. Updating P300: an integrative theory of P3a and P3b. Clin Neurophysiol: Off J Int Fed Clin Neurophysiol. (2007) 118:2128–48. doi:  10.1016/j.clinph.2007.04.019, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Arıkan MK, İlhan R. P300 parameters in major depressive disorder: A systematic review and meta-analysis. World J Biol Psychiatry: Off J World Fed Societies Biol Psychiatry. (2024) 25:255–66. doi:  10.1080/15622975.2024.2321554, PMID: [DOI] [PubMed] [Google Scholar]
  • 20. Zhao L, Wang X, Sun G. Positive classification advantage of categorizing emotional faces in patients with major depressive disorder. Front Psychol. (2022) 13:734405. doi:  10.3389/fpsyg.2022.734405, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Klimesch W. α-band oscillations, attention, and controlled access to stored information. Trends Cogn Sci. (2012) 16:606–17. doi:  10.1016/j.tics.2012.10.007, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Klimesch W. EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. Brain Res Brain Res Rev. (1999) 29:169–95. doi:  10.1016/s0165-0173(98)00056-3, PMID: [DOI] [PubMed] [Google Scholar]
  • 23. Buschman TJ, Miller EK. Top-down versus bottom-up control of attention in the prefrontal and posterior parietal cortices. Science. (2007) 315:1860–2. doi:  10.1126/science.1138071, PMID: [DOI] [PubMed] [Google Scholar]
  • 24. Xie YH, Zhang YM, Fan FF, Song XY, Liu L. Functional role of frontal electroencephalogram alpha asymmetry in the resting state in patients with depression: A review. World J Clin cases. (2023) 11:1903–17. doi:  10.12998/wjcc.v11.i9.1903, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Sharpley CF, Bitsika V, Shadli SM, Jesulola E, Agnew LL. EEG frontal lobe asymmetry as a function of sex, depression severity, and depression subtype. Behav Brain Res. (2023) 443:114354. doi:  10.1016/j.bbr.2023.114354, PMID: [DOI] [PubMed] [Google Scholar]
  • 26. Kou Y, Si Y, Liu L, Li J, Zhang Y, Li W, et al. Exploring the functional connectivity of resting-state EEG in adolescent major depressive disorder. J Integr Neurosci. (2025) 24(10):42821. doi:  10.31083/jin42821, PMID: [DOI] [PubMed] [Google Scholar]
  • 27. Yin S, Li Y, Chen A. Functional coupling between frontoparietal control subnetworks bridges the default and dorsal attention networks. (2022) 227:2243–60. doi:  10.1007/s00429-022-02517-7, PMID: [DOI] [PubMed] [Google Scholar]
  • 28. Lachaux JP, Rodriguez E, Martinerie J, Varela FJ. Measuring phase synchrony in brain signals. Hum Brain Mapp. (1999) 8:194–208. doi:  10.1002/(sici)1097-0193(1999)8:4<194::aid-hbm4>3.0.co;2-c [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Rubinov M SO. Brain connectivity toolbox (version 2020.03.02) (2020). Available online at: https://www.brain-connectivity-toolbox.net (Accessed March 10, 2026).
  • 30. Cole MW, Schneider W. The cognitive control network: Integrated cortical regions with dissociable functions. NeuroImage. (2007) 37:343–60. doi:  10.1016/j.neuroimage.2007.03.071, PMID: [DOI] [PubMed] [Google Scholar]
  • 31. Cole MW, Reynolds JR, Power JD, Repovs G, Anticevic A, Braver TS. Multi-task connectivity reveals flexible hubs for adaptive task control. Nat Neurosci. (2013) 16(9):1348–1355. doi:  10.1038/nn.3470, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P, et al. Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL): initial reliability and validity data. J Am Acad Child Adolesc Psychiatry. (1997) 36(7):980–988. doi:  10.1097/00004583-199707000-00021, PMID: [DOI] [PubMed] [Google Scholar]
  • 33. Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. (2004) 134:9–21. doi:  10.1016/j.jneumeth.2003.10.009, PMID: [DOI] [PubMed] [Google Scholar]
  • 34. SJ L. An introduction to the event-related potential technique. MIT Press. (2014). [Google Scholar]
  • 35. Dien J. The ERP PCA Toolkit: an open source program for advanced statistical analysis of event-related potential data. J Neurosci Methods. (2010) 187:138–45. doi:  10.1016/j.jneumeth.2009.12.009, PMID: [DOI] [PubMed] [Google Scholar]
  • 36. Yeh CH, Shi W. Identifying phase-amplitude coupling in cyclic alternating pattern using masking signals. Sci Rep. (2018) 8:2649. doi:  10.1038/s41598-018-21013-9, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Rubinov M, Sporns O. Complex network measures of brain connectivity: uses and interpretations. NeuroImage. (2010) 52:1059–69. doi:  10.1016/j.neuroimage.2009.10.003, PMID: [DOI] [PubMed] [Google Scholar]
  • 38. Blakemore SJ. The social brain in adolescence. Nat Rev Neurosci. (2008) 9:267–77. doi:  10.1038/nrn2353, PMID: [DOI] [PubMed] [Google Scholar]
  • 39. Giedd JN, Blumenthal J, Jeffries NO, Castellanos FX, Liu H, Zijdenbos A, et al. Brain development during childhood and adolescence: a longitudinal MRI study. Nat Neurosci. (1999) 2(10):861–863. doi:  10.1038/13158, PMID: [DOI] [PubMed] [Google Scholar]
  • 40. Fitzgerald PJ. Frontal alpha asymmetry and its modulation by monoaminergic neurotransmitters in depression. Clin Psychopharmacol Neurosci: Off Sci J Korean Coll Neuropsychopharmacol. (2024) 22:405–15. doi:  10.9758/cpn.23.1138, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Wagner G, Koch K, Schachtzabel C, Peikert G, Schultz CC, Reichenbach JR, et al. Self-referential processing influences functional activation during cognitive control: an fMRI study. Soc Cogn Affect Neurosci. (2013) 8(7):828–837. doi:  10.1093/scan/nss074, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Grandjean J, D'Ostilio K, Phillips C, Balteau E, Degueldre C, Luxen A, et al. Modulation of brain activity during a Stroop inhibitory task by the kind of cognitive control required. PLoS One. (2012) 7(7):e41513. doi:  10.1371/journal.pone.0041513, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Bonnefond M, Jensen O. Alpha oscillations serve to protect working memory maintenance against anticipated distracters. Curr Biol: CB. (2012) 22:1969–74. doi:  10.1016/j.cub.2012.08.029, PMID: [DOI] [PubMed] [Google Scholar]
  • 44. Li Y, Zhang H, Wang L. Functional connectivity of the frontoparietal control network in major depressive disorder: associations with cognitive control and reaction time. J Affect Disord. (2022) 311:112–21. doi:  10.1016/j.jad.2022.05.078, PMID: [DOI] [PubMed] [Google Scholar]
  • 45. Zhao J, Chen X, Liu M. EEG beta band abnormalities in major depressive disorder: A meta-analysis of resting-state and task-related studies. Neurosci Biobehav Rev. (2021) 127:435–48. doi:  10.1016/j.neubiorev.2021.05.018, PMID: [DOI] [PubMed] [Google Scholar]
  • 46. Posner MI, Petersen SE. The attention system of the human brain. Annu Rev Neurosci. (1990) 13:25–42. doi:  10.1146/annurev.ne.13.030190.000325, PMID: [DOI] [PubMed] [Google Scholar]
  • 47. Corbetta M, Shulman GL. Control of goal-directed and stimulus-driven attention in the brain. Nat Rev Neurosci. (2002) 3:201–15. doi:  10.1038/nrn755, PMID: [DOI] [PubMed] [Google Scholar]
  • 48. Hsu LM, Lane TJ, Wu CW. Spontaneous thought-related network connectivity predicts sertraline effect on major depressive disorder. (2021) 15:1705–17. doi:  10.1007/s11682-020-00364-w, PMID: [DOI] [PubMed] [Google Scholar]
  • 49. Liu DY, Ju X, Gao Y, Han JF, Li Z, Hu XW, et al. From molecular to behavior: higher order occipital cortex in major depressive disorder. Cereb Cortex. (2022) 32(10):2129–2139. doi:  10.1093/cercor/bhab343, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Power JD, Cohen AL, Nelson SM, Wig GS, Barnes KA, Church JA, et al. Functional network organization of the human brain. Neuron. (2011) 72(4):665–678. doi:  10.1016/j.neuron.2011.09.006, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51. Wang J, Zuo X, He Y. Graph-based network analysis of resting-state functional MRI. Front Syst Neurosci. (2010) 4:16. doi:  10.3389/fnsys.2010.00016, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52. Harding IH, Yücel M, Harrison BJ, Pantelis C, Breakspear M. Effective connectivity within the frontoparietal control network differentiates cognitive control and working memory. NeuroImage. (2015) 106:144–53. doi:  10.1016/j.neuroimage.2014.11.039, PMID: [DOI] [PubMed] [Google Scholar]
  • 53. Jang KI, Kim S, Kim SY, Lee C, Chae JH. Machine learning-based electroencephalographic phenotypes of schizophrenia and major depressive disorder. Front Psychiatry. (2021) 12:745458. doi:  10.3389/fpsyt.2021.745458, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54. Mumtaz W, Ali SSA, Yasin MAM, Malik AS. A machine learning framework involving EEG-based functional connectivity to diagnose major depressive disorder (MDD). Med Biol Eng Comput. (2018) 56:233–46. doi:  10.1007/s11517-017-1685-z, PMID: [DOI] [PubMed] [Google Scholar]
  • 55. Wu CT, Huang HC, Huang S, Chen IM, Liao SC. Resting-state EEG signal for major depressive disorder detection: A systematic validation on a large and diverse dataset. (2021) 11:. doi:  10.3390/bios11120499, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56. Bassett DS, Bullmore ET. Human brain networks in health and disease. Curr Opin Neurol. (2009) 22:340–7. doi:  10.1097/WCO.0b013e32832d93dd, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57. Ng MY, DiVasto KA. How do cognitive behavioral therapy and interpersonal psychotherapy improve youth depression? Applying meta-analytic structural equation modeling to three decades of randomized trials. (2023) 149:507–48. doi:  10.1037/bul0000395, PMID: [DOI] [PubMed] [Google Scholar]
  • 58. Tottenham N, Gabard-Durnam LJ. The developing amygdala: a student of the world and a teacher of the cortex. Curr Opin Psychol. (2017) 17:55–60. doi:  10.1016/j.copsyc.2017.06.012, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59. Fonseka TM, MacQueen GM, Kennedy SH. Neuroimaging biomarkers as predictors of treatment outcome in Major Depressive Disorder. J Affect Disord. (2018) 233:21–35. doi:  10.1016/j.jad.2017.10.049, PMID: [DOI] [PubMed] [Google Scholar]
  • 60. Cuijpers P, Pineda BS, Quero S, Karyotaki E, Struijs SY, Figueroa CA, et al. Psychological interventions to prevent the onset of depressive disorders: A meta-analysis of randomized controlled trials. Clin Psychol Rev. (2021) 83:101955. doi:  10.1016/j.cpr.2020.101955, PMID: [DOI] [PubMed] [Google Scholar]
  • 61. Klingberg T. Training and plasticity of working memory. Trends Cogn Sci. (2010) 14:317–24. doi:  10.1016/j.tics.2010.05.002, PMID: [DOI] [PubMed] [Google Scholar]
  • 62. Lorenzo-Luaces L, Johns E. The generalizability of randomized controlled trials of self-guided internet-based cognitive behavioral therapy for depressive symptoms: systematic review and meta-regression analysis. (2018) 20:e10113. doi:  10.2196/10113, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63. Jollans L, Boyle R, Artiges E, Banaschewski T, Desrivières S, Grigis A, et al. Quantifying performance of machine learning methods for neuroimaging data. NeuroImage. (2019) 199:351–365. doi:  10.1016/j.neuroimage.2019.05.082, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64. Shim M, Lee SH, Hwang HJ. Inflated prediction accuracy of neuropsychiatric biomarkers caused by data leakage in feature selection. Sci Rep. (2021) 11:7980. doi:  10.1038/s41598-021-87157-3, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65. Moradi N, Goodyear BG, Sotero RC. Deep EEG source localization via EMD-based fMRI high spatial frequency. PloS One. (2024) 19:e0299284. doi:  10.1371/journal.pone.0299284, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66. Widge AS, Heilbronner SR, Hayden BY. Prefrontal cortex and cognitive control: new insights from human electrophysiology. (2019) 8:. doi:  10.12688/f1000research.20044.1, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67. Yang H, Chen X, Chen ZB, Li L, Li XY, Castellanos FX, et al. Disrupted intrinsic functional brain topology in patients with major depressive disorder. Mol Psychiatry. (2021) 26(12):7363–7371. doi:  10.1038/s41380-021-01247-2, PMID: [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Articles from Frontiers in Psychiatry are provided here courtesy of Frontiers Media SA

RESOURCES