Neural correlates of aggression in schizophrenia: An event-related potential study using the competitive reaction time task

Abstract

BACKGROUND

The neural mechanisms underlying aggressive behavior in schizophrenia (SCZ) remain poorly understood. To date, no studies have reported on the event-related potential (ERP) characteristics of aggression in SCZ using the competitive reaction time task (CRTT). Further investigation into the ERP correlates of aggression in SCZ would provide valuable insights into the neural processes involved.

AIM

To explore the neural mechanism of aggressive behavior in SCZ.

METHODS

Participants of this study included 40 SCZ patients and 42 healthy controls (HCs). The Reactive Proactive Aggression Questionnaire was used to assess trait of aggression. The Barratt Impulsiveness Scale 11 was used to measure impulsiveness. The Positive and Negative Symptom Scale (PANSS) was used to evaluate psychopathological features and disease severity. All participants were measured with ERP while performing the CRTT. Data of behavior, ERP components (P2, N2, and P3), and feedback-related negativity (FRN) were analyzed.

RESULTS

Analysis of the behavioral data revealed that compared with HCs, SCZ patients exhibited higher punishment choices. Analysis of ERP components showed that compared with HCs, SCZ patients exhibited higher N2 amplitudes and P2 amplitudes during the decision phase of the CRTT; however, SCZ patients exhibited lower FRN amplitudes and lower P3 amplitudes during the outcome phase of the CRTT. The N2 amplitudes evoked by high-intensity provocation were positively related to PANSS-P scores. And the P3 amplitudes evoked in the winning trials were negatively correlated with the PANSS-G scores.

CONCLUSION

SCZ patients exhibit abnormal ERP characteristics evoked by the CRTT, which suggests the neural correlates of aggressive behavior in SCZ.

Key Words: Schizophrenia; Event-related potential; Aggressive behavior; Competitive reaction time task; Neural mechanism

Core Tip: This study represents the first investigation employing the competitive reaction time task paradigm to examine the neurocognitive mechanisms underlying aggressive behavior in schizophrenia (SCZ) patients. Our findings demonstrate a complex interplay among positive psychotic symptoms, theory of mind deficits, and negative affect dysregulation in mediating aggression in SCZ. These findings provide valuable insights into the understanding of the neural mechanisms of aggression and may guide targeted interventions in SCZ.

INTRODUCTION

Schizophrenia (SCZ) is defined as a group of severe mental disorders of unknown etiology with an onset typically occurring during adolescence or early adulthood. Moreover, the condition is characterized by abnormalities in perception, thinking, emotions, and behaviors[1]. Although the symptoms of SCZ are varied, aggression is one of the most prominent and has significant social impact[2]. Aggression is defined as any behavior directed toward another individual that is performed with the proximate intent to cause harm[3]. Research has indicated a global prevalence rate of aggression among SCZ patients of 33.3%[4], and the incidence of aggressive behaviors in SCZ patients is higher than that in the general population[5]. Aggression in SCZ patients can lead to severe negative outcomes, including serious harm to other individuals, excessive treatment burdens[6], extended hospitalizations[7], increased antipsychotic dosages, and greater stigmatization[8].

Aggression in clinical populations can be operationally divided into two neurobehaviorally distinct subtypes: Proactive aggression and reactive aggression[9]. Proactive aggression refers to planned, goal-oriented behaviors without provocation, whereas reactive aggression is associated with increased levels of autonomic arousal, and is triggered by negative emotions such as anger, frustration, or perceived threat[10]. Aggression in SCZ typically presents as a behavior that is marked by a lack of clear intent or planning[11].

Studies have demonstrated that aggression may be a precursor to SCZ and may be closely related to positive symptoms, which is a potential new dimension of SCZ[12]. Many studies have demonstrated that cognitive deficits are a major driver of aggression in patients with SCZ[13-16]. The brain regions implicated in aggression include the orbitofrontal cortex, dorsolateral prefrontal cortex[17], anterior cingulate cortex, posterior cingulate cortex, insula, hippocampus, amygdala[18], and striatum[19]. In addition, recent studies have emphasized the potential importance of the precuneus and inferior parietal lobule in various cognitive processes[20,21]. Impaired microstructural integrity in these regions or their connected white matter tracts may contribute to aggression by disrupting emotion recognition and regulation, reward and avoidance learning, and decision-making processes[22].

Event-related potential (ERP) technology provides high temporal resolution electrophysiological data by recording brain electrical activity that occurs during specific events, thus capturing millisecond-level dynamic changes. Compared with other neuroimaging techniques [such as functional magnetic resonance imaging (fMRI) and positron emission tomography], ERP offers noninvasive advantages without long-term psychophysiological impacts, and is suitable for widespread research and clinical applications[23].

Previous ERP studies focusing on aggression in patients with SCZ have utilized the Go/No-Go task or the emotional facial recognition paradigm[24]. These passive response tasks poorly simulate real-world dynamic interactions between motivational conflicts and behavioral decisions. In fact, SCZ patients exhibit significant deficits in their ability to suppress the dominant response, and response inhibition has been shown to be only weakly associated with aggression[25,26].

The Taylor Aggression Paradigm (TAP) is one of the most frequently used measures of reactive aggression[27]. In the TAP, participants believe that they are competing with another participant to achieve the fastest completion on a competitive reaction time (RT) task (CRTT), with the winner choosing the level of an aversive stimulus to be administered to the loser. This stimulus level represents the measure of aggression. Previous studies have shown that ERP components (N2, P2, and P3) and feedback-related negativity (FRN) evoked by the TAP are associated with aggression in healthy individuals. For example, a previous study demonstrated that at high provocation, adult participants with trait aggression had increased frontal N2 amplitudes during the decision phase; another study indicated that N2 amplitudes were increased in nonviolent men and individuals with high psychopathic traits in the decision phase[28]. Many studies have reported increased P2 amplitudes in violent participants during the decision phase[29], as well as increased P2 amplitudes in response to emotional images in provoked groups compared with unprovoked groups[30]. In addition, during the outcome phase, a study reported that the FRN in the violent adolescent group was significantly lower overall than that in the normal adolescent group, regardless of whether the participants were winning or losing[31]. Furthermore, reduced (hypo-sensitive) P3 amplitudes have been linked to various psychopathologies including antisocial behavior as well as depression and SCZ[32].

The CRTT is a widely implemented variant of the TAP[33]. In this competitive paradigm, participants engage in sequential trials against opponents. Upon winning a trial, the losing opponent receives noise punishment; conversely, if they lose, the participants will be punished. Evidence indicates that the CRTT exhibits high internal, convergent, discriminant, and external validity[34].

To date, no studies on the ERP characteristics of aggression with the use of the CRTT in SCZ patients have been reported. Further investigations of the ERP characteristics of aggression in SCZ would be helpful in understanding the neural process of aggression. Moreover, insights into the neural mechanisms underlying these deficits could provide novel targets for pharmacological and neuromodulatory treatments in SCZ.

In the present study, the participants included SCZ patients and healthy controls (HCs), and the measurements of ERPs during the CRTT were used to investigate the neural process of aggression. The purpose of this study was to investigate the ERP characteristics of aggression and to further explore the neural mechanism of the cognitive processing of aggression in patients with SCZ.

MATERIALS AND METHODS

Time and setting

This study was conducted in the Department of Clinical Psychiatry, the Affiliated Mental Health Center of Jiangnan Medical University from January 2025 to April 2025. The study protocol was approved by the Ethics Committee of the Affiliated Mental Health Center of Jiangnan Medical University and was conducted according to the Declaration of Helsinki (Reference No. WXMHCIRB2025 LLky020).

Participants

The inclusion criteria for the SCZ group were as follows: (1) Met the criteria for SCZ according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5); (2) Were aged 18-65 years; and (3) Had no current or history of neurological illness or any other kind of severe physical illness that would affect his or her cognitive function. The exclusion criteria were as follows: (1) Met the criteria for any other mental disorder according to DSM-5; (2) Was treated by electroconvulsive therapy (ECT) or modified ECT within 6 months before recruitment; (3) Had nicotine/other substance misuse or dependence within 6 months before recruitment; and (4) Had hearing and visual impairments that cannot be corrected to satisfy the demands of the current study. The inclusion criteria for the HCs were as follows: (1) No diagnosis of psychiatric disorders according to the DSM-5 criteria; (2) No family history of any mental disorders; and (3) Were aged 18-65 years. The exclusion criteria were as follows: (1) Had hearing and visual impairments that could not be corrected to satisfy the demands of the current study; and (2) Had nicotine/other substance misuse or dependence within 6 months before recruitment. HCs were selected from residents of Wuxi City, Jiangsu Province, China via local media advertising.

The G*Power software (version. 3.1.9.7; Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany) was used to estimate the sample size in this study. In the F tests, α = 0.05 was considered to be statistically significant, and a minimum requirement of 32 subjects per group met the sample size requirement. A total of 40 SCZ patients and 42 HCs were included in this study. For SCZ patients, the dose of antipsychotic medication was calculated based on chlorpromazine equivalence[35]. Prior to the study, participants or their guardians were informed about the study's purpose and procedures, and then provided written informed consent.

Clinical assessments

The Reactive Proactive Aggression Questionnaire was used to assess the trait of aggression[36]. The Barratt Impulsiveness Scale Version 11 was used to measure impulsiveness[37]. Additionally, the Modified Overt Aggression Scale and Buss-Perry Questionnaire were used to evaluate aggression[38]. The Positive and Negative Symptom Scale (PANSS) was used to assess psychopathological features and disease severity[39]. Furthermore, the Annett handedness scale was used to rate the participant’s handedness[40].

Experimental paradigm

The CRTT was conferred from the TAP[27-29,31,32,41-43], and was programmed on computer using E-Prime 3.0 (Psychology Software Tools Incorporated, Pittsburgh, United States). Participants were instructed to believe that they were engaging in a competitive task against an opponent. Each trial commenced with a focus point shown in the center of the screen (“+”, 1000 milliseconds), followed by a red question mark “?” shown for 3 seconds, then the participants were asked to choose the duration and volume of the noise blast on a 10-point scale (volume: 0-85 decibels, duration: 0-5 seconds). The first provocation (first noise feedback not being zero) was given in the seventh trial. After selection, they had to try to outperform the opponent by pressing the space key when the rectangle turned from a green color into a red color on the screen. The duration was randomized, ranging from 1000 to 2000 milliseconds. Subsequently, the intensity of the punishment chosen by the opponent was presented on the screen (1000 milliseconds). The visual feedback indicated whether the participants had won or lost the trial by showing a smiling face with “win” or a sad face with “lose” for 1500 milliseconds. Finally, the participants received the noise stimulus chosen by the opponent. The participants avoided the punishment if they had won the competition. Furthermore, at the end of the study, participants were carefully probed for suspicion of the real existence of the opponent. We used the phrase “What was your impression of your opponent?” to avoid participants becoming suspicious of the questions themselves (Figure 1).

Figure 1 Schematic illustration of the competitive reaction time task. Participants were instructed to play against the opposing participant who was in another room. The player who responded faster won. Before each trial, both participants and their opponents were provided with the option to select the punishment (volume: 0-85 dB; duration: 0-5 s) that the opponent would receive in case of losing the trial. In the reaction phase, both players were asked to press the space key as quickly as possible when a green rectangle turned to red. The screen showed a smiling face when the participants won, and a crying face when the participants lost and received noise punishment. ms: Milliseconds.

In reality, we predetermined the frequencies of the wins and losses, and the effective RTs were 0-2000 milliseconds. When participants reacted more slowly, they automatically lost the trial to ensure credibility. Furthermore, the volume and duration of the noise feedback that the participants experienced were preprogrammed and standardized for each participant. We considered volumes 1-5 as indicating low noise provocation levels and volumes 6-10 as indicating high noise provocation levels.

The task included a total of 60 trials that were divided into two blocks, with a 10-minute rest occurring between the blocks. Before the formal task, a practice session was conducted to ensure that all of the participants were familiar with the test. After completing the entire experiment, each participant was compensated with a payment of 150.0 RMB. The noise level utilized in the trial meets China’s occupational safety and health standards for full-time workers exposed to 85 decibels for a duration of 8 hours, which is within the pain threshold of 125 decibels. The software “GOLDWAVE” (v7.2, Goldwave Inc., St. John’s, Newfoundland, Canada, 2010) was used to assess the noise levels.

Electroencephalograph recording and processing

Continuous electroencephalograph (EEG) recordings (bandpass filtered between 0.05 to 100 Hz, digitized at 500 Hz) were conducted using a 64-channel Ag-AgCl elastic cap, adhering to the International 10/20 System, via the BioSemi Active Two system (https://www.biosemi.com/Products_ActiveTwo.htm). The impedance across all electrodes was kept under 5 kΩ during the sessions. The EEG was epoched from -200 milliseconds to 1000 milliseconds before and after the stimulus marker and the baseline was corrected by using the mean (ranging from -200 milliseconds to 0 milliseconds). Data processing was performed offline using MATLAB version 2020b (MathWorks, Inc., Natick, MA, United States) in conjunction with the EEGLAB toolbox. All of the EEG data were re-referenced to the average of the bilateral mastoids and were bandpass filtered from 0.1 to 30 Hz (via a low-pass cutoff at 12 Hz for N2 and P2) utilizing a notch filter at 50 Hz[42]. Both electric eye artifact elimination and data correction were performed by the independent component analysis statistical procedure. For the subsequent analysis, trials that demonstrated EEG voltages exceeding the ± 80 μV threshold were excluded.

According to the grand averaged ERP waveforms and the results of previous literature[29,31,42-44], in the CRTT decision phase, mean amplitudes were quantified on three electrodes (F3, F4, and Fz) in a time window of 150-250 milliseconds for the P2, and in a time window of 200-300 milliseconds for N2. In the CRTT outcome phase, the activity was recorded at six electrodes (F3, F4, C3, C4, Fz, and Pz) during the time window of 300-500 milliseconds for the P3, and during the time window of 250-280 milliseconds following stimulus onset was used to calculate the FRN.

Statistical analysis

Statistical analyses were performed using SPSS 22.0 (IBM Corp, Armonk, NY, United States). Normality assumptions were verified through the Kolmogorov-Smirnov test, confirming normal distribution for all continuous variables. For continuous variables, independent sample t-tests were used to compare the differences between the SCZ group and HC group, whereas the χ² test was employed for categorical variables. Two-way mixed analysis of variance (ANOVA) on the ERP components (N2, P2, and P3) and FRN was used to compare the differences between the SCZ group and the HC group. Pearson correlation analysis was employed to investigate the relationships between the ERP components and psychopathology parameters. The degrees of freedom for the F ratio were adjusted using the Greenhouse Geisser correction. Additionally, effect sizes were calculated using the partial eta-squared (η²). The threshold for significance was set to P < 0.05 (two-tailed). In cases of significant findings, post hoc Bonferroni corrections were conducted.

RESULTS

Demographic and clinical characteristics

Table 1 presents the demographic characteristics of the participants. There were no significant differences between the SCZ and HC groups in terms of age, sex, education, or handedness. Forty patients were treated with antipsychotics: Six were treated with risperidone (mean dose 4.0 ± 1.3 mg/day), ten treated with olanzapine (mean dose 12.75 ± 2.15 mg/day), nine with quetiapine (mean dose 500.0 ± 100.65 mg/day), six with aripiprazole (mean dose 15.0 ± 9.37 mg/day), six with amisulpride (mean dose 250.50 ± 26.25 mg/day), and four with paliperidone (mean dose 7.50 ± 2.20 mg/day). For SCZ patients, the mean chlorpromazine-equivalent dose was 250.50 ± 10.66 mg/day, as calculated according to the previous report[35].

Table 1 Demographic and clinical characteristics of participants.

Variable	HCs (n = 42)	SCZ (n = 40)	t/χ2	P value
Age, years	34.5 ± 8.0	35.8 ± 9.6	1.373	0.245
Education, years	13.4 ± 2.4	12.4 ± 2.5	1.735	0.087
Handedness, right/middle/left	17/14/11	16/14/10	0.029	0.986
Sex, male/female	15/27	16/24	0.160	0.689
BIS-11 score	65.93 ± 8.23	84.45 ± 11.02	-8.650	< 0.001
RPQ proactive score	0.83 ± 0.82	2.48 ± 2.40	-4.108	< 0.001
RPQ reactive score	3.67 ± 1.92	7.60 ± 2.34	-8.366	< 0.001
RPQ total score	4.57 ± 2.17	10.07 ± 3.83	-8.050	< 0.001
PANSS total score	NA	87.35 ± 12.22	NA	NA
PANSS positive score	NA	29.87 ± 5.15	NA	NA
PANSS negative score	NA	22.35 ± 6.44	NA	NA
PANSS general score	NA	35.17 ± 5.75	NA	NA
MOAS score	NA	7.89 ± 1.77	NA	NA
BPAQ score	34.76 ± 7.89	NA	NA	NA

SCZ: Schizophrenia; HC: Healthy control; PANSS: Positive and Negative Syndrome Scale; BIS-11: Barratt Impulsiveness Scale-11; RPQ: Reactive Proactive Aggression Questionnaire; MOAS: Modified Overt Aggression Scale; BPAQ: Buss-Perry Questionnaire; NA: Not available.

Analysis of behavioral data

The factor-analytically derived scoring method was used to compute the total score[33]. The first seven trials of punishment responses have been used to reflect unprovoked aggression in previous research[43,44].

An ANOVA utilizing provocation (unprovoked vs provoked) as a within-subjects factor and group (SCZ vs HC) as a between-subjects factor for punishment selection revealed a significant main effect of provocation (F_{(1, 80)} = 72.850, P = 0.000, η² = 0.477), indicating greater reactive aggression in participants in the provoked condition than in those in the unprovoked condition. The main effect of the group was significant (F_{(1, 80)} = 119.579, P = 0.000, η² = 0.599), thus indicating that the SCZ patients exhibited greater reactive aggression compared with the HCs. However, the interaction effect between provocation and group was not significant (F_{(1, 80)} = 3.285, P = 0.074, η² = 0.039; Figure 2).

Figure 2 Average punishment selections for healthy control and schizophrenia patient groups under unprovoked and provoked conditions. SCZ: Schizophrenia; HC: Healthy control.

Analysis of ERP data

N2 and P2 components: Two-way mixed ANOVA was performed on the N2 and P2 components with the provocation level (high vs low) being used as the within-subjects factor and group (HC vs SCZ) as the between-subjects factor.

As shown in Figure 3, for the P2 amplitudes, the main effect of provocation was significant (F_{(1, 80)} = 5.820, P = 0.018, η² = 0.068), and the main effect of group was also significant (F_{(1, 80)} = 7.164, P = 0.009, η² = 0.082). Compared with the HC group, the SCZ group exhibited markedly higher (more positive) P2 amplitudes. Additionally, the P2 amplitudes following the high-provocation condition were notably lower than those following the low-provocation condition. However, the difference between provocation and group was not significant (F_{(1, 80)} = 0.089, P = 0.766, η² = 0.001). For the P2 latencies, neither significant main effects nor interactions were observed.

Figure 3 Grand averaged amplitudes and topographical distribution of P2 of both schizophrenia and healthy control groups. A: Grand average event-related potential waveforms of P2 (time window 150-250 milliseconds) at Fz under different provocative conditions (high vs low); B: Topographic maps of P2 under different provocative conditions (high vs low). HC: Healthy control; SCZ: Schizophrenia; ms: Milliseconds.

As shown in Figure 4, for the N2 amplitudes, the main effect of provocation was significant (F_{(1, 80)} = 4.760, P = 0.032, η² = 0.056), and the main effect of group was significant (F_{(1, 80)} = 5.288, P = 0.024, η² = 0.062). The SCZ group exhibited markedly higher (more negative) N2 amplitudes compared with the HC group. Additionally, the N2 amplitudes following the high-provocation condition was notably lower than that following the low-provocation condition. However, the difference between provocation and group was not significant (F_{(1, 80)} = 0.003, P = 0.959, η² = 0.000). For the N2 latencies, the analysis revealed that the main effects were significant for provocation (F_{(1, 80)} = 6.370, P = 0.014, η² = 0.074), and the main effects were not significant for the group (F_{(1, 80)} = 0.377, P = 0.541, η² = 0.005). Additionally, the interaction of group × provocation was significant (F_{(1, 80)} = 8.256, P = 0.005, η² = 0.094). Simple effects analyses revealed that under the low-provocation condition, the N2 latencies of the HC group were longer than those of the SCZ group (P = 0.006); however, under the high-provocation condition, there was no significant difference between the two groups (P = 0.269). Additionally, the N2 latencies of the SCZ group under the high-provocation condition was significantly longer than that under the low-provocation condition (P = 0.000); however, this difference was not observed in the HC group (P = 0.803).

Figure 4 Grand averaged amplitudes and topographical distribution of N2 of both schizophrenia and healthy control groups. A: Grand average event-related potential waveforms of N2 (time window 200-300 milliseconds) at Fz under different provocative conditions (high vs low); B: Topographic maps of N2 under different provocative conditions (high vs low). HC: Healthy control; SCZ: Schizophrenia; ms: Milliseconds.

FRN and P3 components: Two-way mixed ANOVA was conducted on the FRN and P3 components, with outcome (win vs lose) being used as the within-subjects factor and group (HC vs SCZ) as the between-subjects factor.

As shown in Figure 5, for the FRN amplitudes, the main effect of the group was significant (F_{(1, 80)} = 4.619, P = 0.035, η² = 0.055), and the FRN amplitudes were lower (more negative) in the SCZ group than in the HC group. However, the difference between the outcome and group was not significant (F_{(1, 80)} = 0.000, P = 0.987, η² = 0.000). Additionally, the main effect of outcome was not significant (F_{(1, 80)} = 1.096, P = 0.298, η² = 0.014). The FRN latency demonstrated no significant main effect for either the outcome (F_{(1, 80)} = 0.313, P = 0.577, η² = 0.004) or the group × outcome interaction (F_{(1, 80)} = 0.429, P = 0.514, η² = 0.005). However, the main effect was significant for the group (F_{(1, 80)} = 15.296, P = 0.000, η² = 0.161). The FRN latency in the HC group was significantly longer than that in the SCZ group.

Figure 5 Grand averaged amplitudes and topographical distribution of feedback-related negativity of both schizophrenia and healthy control groups. A: Grand average event-related potential waveforms of feedback-related negativity (FRN; time window 250-280 milliseconds) at Fz with different outcomes (win vs lose); B: Topographic maps of FRN under different outcomes (win vs lose). HC: Healthy control; SCZ: Schizophrenia; ms: Milliseconds.

As shown in Figure 6, for the P3 amplitudes, the main effect of outcome was significant (F_{(1, 80)} = 6.057, P = 0.016, η² = 0.700), and the P3 amplitudes were larger (more positive) in the win trials compared with the lose trials. The main effect of the group was significant, and the P3 amplitudes were also larger for the HC group than for the SCZ group (F_{(1, 80)} = 4.651, P = 0.034, η² = 0.055). Furthermore, the group × outcome interaction was not significant (F_{(1, 80)} = 0.163, P = 0.688, η² = 0.002). For the P3 latencies, neither significant main effects nor interactions were observed.

Figure 6 Grand averaged amplitudes and topographical distribution of P3 of both schizophrenia and healthy control groups. A: Grand average event-related potential waveforms of P3 (time window 300-500 milliseconds) at Cz with different outcomes (win vs lose); B: Topographic maps of P3 under different outcomes (win vs lose). HC: Healthy control; SCZ: Schizophrenia; ms: Milliseconds.

Analysis of exploratory correlations between variables

For the HC group, no significant correlations across any domains were observed (all P > 0.05). As shown in Figure 7, in the SCZ group, the N2 amplitudes evoked by high-intensity provocation was positively related to PANSS-P scores (r = 0.524, P = 0.000) and the P2 amplitudes evoked by high-intensity provocation was positively related to PANSS-G scores (r = 0.428, P = 0.006), whereas the P3 amplitudes evoked in the winning trials showed a negative correlation with the PANSS-G scores (r = -0.325, P = 0.041) and the P3 amplitudes evoked in the losing trials showed a negative correlation with the PANSS-G scores (r = -0.367, P = 0.020).

Figure 7 Heatmap for correlations among variables in schizophrenia and healthy control patients. A: Correlation analysis of event-related potential (ERP) components’ amplitudes and scale scores for healthy control group; B: Correlation analysis of ERP components’ amplitudes and scale scores for schizophrenia group. Positive correlations are indicated in red, while negative correlations are marked in blue. ^aP < 0.05, ^bP < 0.01; HC: Healthy control; SCZ: Schizophrenia; RPQ-A: Reactive Proactive Aggression Questionnaire-active aggression; RPQ-R: Reactive Proactive Aggression Questionnaire-reactive aggression; RPQ-T: Reactive Proactive Aggression Questionnaire-total aggression; BIS-11: Barratt Impulsiveness Scale-11; FRN: Feedback-related negativity.

DISCUSSION

This is the first study to investigate the ERP characteristics of aggression with the CRTT in SCZ, which is also the first to clarify the neuroelectrophysiological mechanisms of aggression for SCZ patients. This study found that compared with that in HCs, the average punishment intensity of aggression in schizophrenic patients was higher during the CRTT. SCZ patients exhibited higher N2 amplitudes and P2 amplitudes during the decision phase of the CRTT. However, SCZ patients exhibited lower FRN amplitudes and lower P3 amplitudes during the outcome phase of the CRTT.

The CRTT is a widely used experimental paradigm designed to assess aggressive behavior in a controlled laboratory setting[41]. In the task, the intensity and duration of the selected punishment are commonly interpreted as behavioral indices of reactive aggression. In our study, SCZ patients have been found to exhibit significantly higher average punishment intensity. This elevated level of aggressive responding is typically interpreted as being an indicator of increased reactive aggression, particularly in response to perceived provocation during the task.

The heightened punishment intensity may reflect several underlying psychopathological and neurocognitive mechanisms. First, paranoid ideation and threat misattribution, which are prevalent in SCZ, may lead individuals to perceive the task environment as being more hostile or adversarial, thereby amplifying retaliatory behaviors. Second, impairments in impulse control and emotion regulation, which are often linked to dysfunction in prefrontal-limbic circuits[45-47], may hinder the modulation of aggressive impulses. Finally, cognitive deficits, such as impaired theory of mind or reduced social cue processing, could further exacerbate inappropriate or exaggerated aggressive responses, as SCZ patients may misinterpret the intentions or actions of their (fictitious) opponents[48,49]. Our findings underscore the complex interplay between psychotic symptomatology, social cognitive dysfunction, and affective dysregulation in contributing to aggression in SCZ.

In this study, ERP findings obtained from SCZ patients performing the CRTT revealed a distinct pattern of altered neural processing across task phases, reflecting abnormalities in conflict monitoring, salience detection, and feedback evaluation.

During the decision phase of the CRTT (i.e., when participants select the intensity of the punitive stimulus), SCZ patients exhibited increased N2 and P2 amplitudes compared with HCs. The N2 component, which is typically associated with conflict detection and cognitive control[50-52], suggests that patients may experience heightened neural conflict or increased effort in response selection. Moreover, the increased P2 amplitude, which is linked to early attentional allocation and salience detection[53,54], may indicate hyperresponsivity to socially provocative stimuli or aberrant attentional engagement during punishment decisions, thus possibly reflecting altered threat perception or exaggerated evaluation of social interactions.

In contrast, during the outcome phase (i.e., when the participants received feedback on the competition), SCZ patients demonstrated reduced FRN and lower P3 amplitudes. The FRN, which is typically elicited by negative or unexpected feedback, reflects early performance monitoring and reward prediction error processing[55,56]. The attenuation in SCZ may reflect blunted sensitivity to outcome valence, which is consistent with prior evidence of dysfunctional reward processing in this population[57,58]. Similarly, the P3 component, which is associated with the conscious evaluation of feedback and attentional resource allocation[59,60], was also diminished, suggesting deficits in integrating motivationally salient outcomes into goal-directed behavior.

Collectively, this ERP pattern indicates a dissociation between heightened early attentional or conflict-related processing and impaired feedback evaluation in SCZ patients during socially competitive interactions. These findings support the notion of dysregulated fronto-striatal and cingulo-parietal circuitries, and provide evidence of neurophysiological mechanisms for aggression in SCZ.

In summary, our study indicated that SCZ patients exhibited abnormal ERP characteristics evoked by the CRTT, which suggested the neural correlates of aggressive behavior in SCZ. These findings provide valuable insights into the understanding of the neural mechanisms of the impulsiveness and aggression and may guide targeted interventions in SCZ.

There are some limitations in this study. First, due to the small sample size, this study is preliminary in nature. In the future, large samples with the same parameters are needed to replicate these findings. Second, the current study involved only a cross-sectional design, thus making it difficult to distinguish causal mechanisms. Third, due to the low spatial resolution of the ERP equipment, future studies should combine high-spatial-resolution fMRI further to clarify the neurophysiological mechanism of aggression in SCZ. Moreover, the investigation of the specific impacts of different treatment modalities on ERP characteristics could yield a more comprehensive understanding of the neural mechanisms involved in this phenomenon.

CONCLUSION

In conclusion, our study demonstrated that patients with SCZ exhibited abnormal ERP characteristics evoked by the CRTT. This ERP pattern reflects a dissociation between heightened early attentional or conflict-related processing and impaired feedback evaluation in SCZ patients. These findings uncover the distinct neurodynamic characteristics of aggressive behavior in SCZ, offering a novel perspective for understanding its pathological mechanism and guiding the development of personalized therapeutic strategies. To elucidate the causal mechanisms, we recommend that future research adopt a longitudinal design and incorporate more comprehensive evidence regarding structural and functional aspects.

ACKNOWLEDGEMENTS

We thank the subjects and their families who participated in this study and we would like to acknowledge everyone who helped us in this project.