The impact of psycho-physical training in esports on reaction time and decision-making in competitive athletes

Abstract

Background

This study analyzed the impact of a cognitive-motor training program on key indicators of cognitive performance.

Methods

A two-arm, parallel, randomised controlled trial (with 1:1 allocation) was conducted. A total of 184 competitive eSports athletes (aged 18–25) were randomly assigned to one of two programmes: an individualised psychophysical training programme integrating aerobic exercise and cognitive self-regulation techniques, or a standard eSports training programme. Outcomes were assessed at baseline, after 4 weeks (mid-intervention) and after 8 weeks (post-intervention). Reaction time and decision-making efficiency were assessed at baseline, mid-intervention, and post-intervention using a computerized reaction-time battery and an adapted Iowa Gambling Task. Data analyzed using repeated-measures ANOVA and moderation analyses. The primary endpoint was defined as the change in the outcome variable from baseline to post-intervention.

Results

The individualized psycho-physical program was associated with greater improvements in reaction time and decision-making indicators compared to standard training, with small-to-moderate effect sizes (η² = 0.08–0.12). Significant Time × Group interactions indicated steeper and more stable performance gains in the experimental group. Age and gender exerted small but significant moderating effects; younger participants and female athletes showing greater responsiveness on selected outcomes.

Conclusions

Integrating aerobic exercise and cognitive self-regulation was associated with improvements in attentional efficiency and decision-making stability under controlled conditions. These findings suggest that individualised psychophysical training could be useful in environments with a high cognitive load. However, further confirmatory trials are required.

Trial registration

Not prospectively registered.

Introduction

Over the past decade, the scientific and applied discourse has increasingly emphasized the need for systematic investigation of the cognitive components of professional activity under conditions of heightened sensorimotor load, particularly in esports [1, 2]. In the last three to four years, research in the field of esports has focused primarily on the study of affect, cognitive-motor processes, and the impact of esports on athletes’ health and team dynamics [3, 4]. Among the most common stressors in esports, current research identifies three groups: team, organizational, and social, as well as a class of digital stressors. These include participation in streaming, dependence on social media reactions, changes in game strategies and dynamics due to patches, and others [5]. Esports offers a high stimulus density, requires parallel information processing, and provides very limited decision-making time with a high cost of error [6–9]. Perceived stress of various types is particularly high due to dependence on the audience and high emotional involvement [10]. Increased stress may be facilitated by the intense use of cognitive resources while the player’s motor activity is low [5, 11]. Based on this, cognitive training and physical training to alleviate existing problems may be considered as the basis of an approach to improving the quality of esports training.

At the international level, a growing body of empirical research has explored the effects of cognitive training on professional gamers. Adaptive neuropsychological programs aimed at enhancing cognitive parameters, such as reaction time, executive control, cognitive flexibility, and decision-making efficiency under time constraints, have been implemented in Western Europe and Asia [2, 12]. At the same time, individual differences—such as age, gender, and cognitive style—have been identified as critical factors influencing the effectiveness of these training protocols, highlighting the need for careful analysis [13]. In the context of rapid digitalization of social and cultural spaces and the evolution of competitive activities, esports has emerged as a phenomenon that fundamentally reshapes our understanding of the physiological and cognitive dimensions of sport [14].

Current scientific discourse remains limited in exploring the biopsychological mechanisms mediating the effects of esports training on neuropsychological indicators of individual performance, such as sensorimotor reaction time and strategic decision-making under time pressure [15]. Physical training has already been experimentally studied in esports, and results demonstrate that the effect on physical performance is more robust than on short-term game-based performance metrics [16, 17]. Therefore, the selection of physical activity for esports athletes can clearly be tailored to its impact on measures of attentiveness and stress resilience, which indirectly improve performance [18]. Currently, there is a gap in research linking stress and self-regulation in esports with specific cognitive metrics within a single design; this is also due to the heterogeneity of performance metrics and the lack of controlled interventions [19, 20]. There is also a lack of data on which interventions actually measurably engage mechanisms of attention, self-regulation, fatigue coping, etc [18, 21–23].

Despite the growing popularity of esports as a self-structured, integrated training model, there remains no comprehensive scientific conceptualization of whether integrated training—combining mental resilience, tactical thinking, and physical organization—can substantially alter cognitive reactivity in athletes, and to what extent such training contributes to developing coping strategies for sensory overload [24, 25].

The research question was whether an individualised psychophysical training programme would improve reaction time and decision-making performance compared to standard eSports training, and whether these effects would be moderated by age and gender. This research questions emerges at the intersection of sports psychology, cognitive science, and physiology, addressing the nature and specificity of psychological and physical training effects on sensorimotor dynamics and decision-making activation in the competitive esports environment. Such research holds the potential not only to expand the theoretical boundaries of sports psychology but also to inform the development of more effective training protocols [26].

An analysis of international experience reveals the high potential of individualized cognitive interventions in enhancing the competitiveness of esports athletes, while also exposing a structural gap in understanding the interpersonal variability in intervention effectiveness, thus justifying the necessity of the present study. The study contributes to the psychology of eSports by empirically linking psychophysical training to measurable improvements in attentional control and decision-making efficiency and operationalizing theoretical assumptions derived from the concepts of limited attentional resources and stress coping. The results allow for the development of evidence-based training protocols for eSports athletes, emphasizing the value of integrating physical training and cognitive self-regulation into routine training. Despite the growing body of research, there is still a lack of randomised controlled trials evaluating integrated psychophysical interventions in eSports using standardised cognitive outcomes. In particular, the combined effects of aerobic exercise and cognitive self-regulation on reaction time and decision-making in controlled experimental settings have not been sufficiently explored.

Literature review

An analysis of the current scientific landscape regarding the impact of training in the context of esports reveals a fragmented yet increasingly comprehensive and interdisciplinary interest in the phenomenon. Chronologically, the evolution of research in this area may be divided into three stages. Early studies demonstrated that regular engagement with video games significantly enhances visual attention, improves the accuracy of tracking moving objects, and accelerates sensorimotor reactions [27–29]. Although these studies did not specifically focus on esports, they laid the cognitive and perceptual groundwork for subsequent research on competitive gaming [30].

During the second period, the focus shifted to the psychological aspects of esports, including stress resilience, pre-competition anxiety, and cognitive load. It was found that esports players exhibit psychophysiological patterns similar to those observed in high-level traditional athletes, including changes in heart rate, cortisol reactivity, and electrode activity [31–33]. The most recent wave of research has aimed to integrate psychological, physiological, and athletic aspects. For instance, specialized psychological training programs combined with moderate physical exercise have been shown to positively impact attention and information processing speed in esports players, while also demonstrating potential for reducing stress reactivity [2, 18, 34].

Nevertheless, there remains a pressing need for the systematic development of standardized protocols to evaluate the effectiveness of psychological and physical interventions [35]. Typically, studies have examined the impact of physical activity and psychological training in isolation, rarely analyzing their combined effects. As a result, the synergistic relationships between physical and psychological training in relation to cognitive indicators—particularly decision-making speed, accuracy, and variability in competitive environments—remain poorly understood [36].

Some studies examining the effects of aerobic exercise suggest that the cognitive response to such training may be gender-specific [37]. Improvements in cognitive executive functions were more pronounced in women, based on a number of specific biomarkers. Age and the proportion of women in the sample, as demonstrated by several researchers, may act as potential moderators or even sources of heterogeneity in the effect of cognitive outcomes and aerobic exercise [38]. Thus, the effects of aerobic exercise on cognitive domains may be heterogeneous; some of this heterogeneity may be related to sample characteristics such as gender and age, which requires experimental verification.

Recent advances in cognitive and exercise neuroscience suggest that improvements in cognitive performance following integrated psycho-physical interventions may be supported by identifiable neurobiological mechanisms. Aerobic exercise has been associated with increased prefrontal cortex activation during cognitively demanding tasks, enhanced neuroplasticity through brain-derived neurotrophic factor (BDNF) regulation, and broader effects on neural efficiency and brain health, which may contribute to improvements in executive functioning and processing speed [39–41]. In parallel, mindfulness-based and cognitive training interventions have been shown to modulate functional brain connectivity and enhance goal-directed attention and inhibitory control, supporting top-down regulation of behavior under cognitively demanding conditions [42, 43]. These neurobiological perspectives provide a theoretical basis for interpreting potential mechanisms of cognitive improvement following integrated psycho-physical training in esports contexts.

The discussion surrounding the effects of psychological and physical training in esports is grounded in a set of fundamental concepts and interdisciplinary theoretical models that aim to explain the phenomenon and underlying mechanisms. The Theory of Limited Attention Resources, proposed by Kahneman, asserts that cognitive attention is a finite resource and that the allocation of these resources across tasks affects performance quality [44]. This theory is particularly relevant to esports, where visual, auditory, and tactile stimuli must be processed simultaneously, suggesting that training aimed at enhancing attentional management could directly influence interactivity and decision accuracy.

In psychophysical interventions within esports, Lazarus’s Transactional Model of Stress is widely employed. It posits that an individual’s perception of a stressor (e.g., competitive pressure or information overload) triggers cognitive appraisal processes and coping strategy formation [45]. Psychological training for esports often incorporates cognitive restructuring techniques, breathing exercises, or mental imagery to optimize these processes. The Neuropsychological Model of Sensorimotor Integration, based on research in motor neuroscience, considers reaction time as a function of coordinated sensory, motor, and cognitive interactions [46]. Training that combines physical exercise with high-speed information processing (e.g., interactive simulations or visual reaction drills) may enhance the plasticity of neural pathways that transform stimuli into motor responses more efficiently.

Consistent with the theoretical model of Limited Attention Resources Theory, some studies demonstrate that even brief mindfulness practice improves reaction time on the Sustained Attention to Response Task (SART), significantly reduces task-unrelated mind-wandering, and contributes to an improvement in the reaction time-to-accuracy ratio on Flanker inhibition tests [47]. Importantly, recent studies have already noted that even brief mindfulness training positively alters behavioral markers of attentional allocation, such as fixation on relevant/irrelevant cues, and is accompanied by activation of attentional control [48]. A longitudinal randomized controlled trial of mindfulness meditation demonstrated a sustained increase in attentional control and a reduction in distress characteristic of e-sports [49]. Thus, it has already been demonstrated that mindfulness effectively frees and redistributes attention, viewed theoretically as a limited resource, by reducing the costs of distraction, rumination, and focusing more on the task at hand. A similar effect was observed for aerobic exercise: an 8-week training program led to improvements in executive functions, including decreased inhibition and updating of attentional reaction time [16]. Aerobic exercise also promotes the release and redistribution of attentional control resources, which facilitates faster reactions and improves the performance of tasks such as controlled reaction time (CRT).

The Lazarus-Folkman Transactional Model of Stress and Coping, as a theoretical framework for the intervention, draws on a number of recent studies that have demonstrated a link between mindfulness and the initial appraisal of a stressor. Mindfulness is systematically associated with a reduction in threat appraisal and stress reduction through this pathway [50]. Mindfulness exercises influence appraisal dynamics, which are associated with stress responses. Mindfulness is expected to reduce stress because it is associated with a reframing of stressor appraisal and coping [51]. This theoretical model is also supported by a randomized controlled trial demonstrating that perceived stress and resilience are effectively modified by structured stress management programs. This suggests that changes in appraisal and self-regulation, particularly through interoception, during physical exercise lead to reduced stress and improved attentional control.

The concept of Mental Resilience is a well-established construct in sports psychology, describing the ability to maintain cognitive and emotional stability under competitive pressure. It is increasingly regarded as a key predictor of success in esports, where mental resilience training programs aim to enhance error tolerance, stress endurance, and adaptability to unpredictability [52, 53]. Finally, the Dual-Process Theory of Decision-Making differentiates between intuitive (fast and automatic) and reflective (slow and controlled) cognitive processing. Esports training, especially when involving repetitive scenario-based simulations, often seeks to shift critical decision-making from reflective to intuitive processing modes to reduce reaction time and minimize cognitive load [54].

Problem statement

This randomised controlled trial aimed to evaluate the effect of an individualised psychophysical training programme on reaction time and decision-making performance, compared with standard eSports training.

H₁: Participants who receive individualised psychophysical training will demonstrate greater improvements in the primary outcomes (CRT, RT-L and DSRT) and secondary decision-making indicators than the control group.

H₂: The effects of the intervention will vary according to age and gender.

Research Objectives:

1. To implement and compare two training protocols: individualized psycho-physical training and standard esports training.

2. To assess changes in predefined primary and secondary cognitive outcomes across three time points (T1, T2, T3).

3. To investigate the moderating effects of gender and age on training outcomes.

Research methods

Study design

This study was designed as a randomised controlled trial (RCT) with two parallel groups and a 1:1 allocation ratio. The trial aimed to evaluate the effect of an individualised psychophysical training programme compared to a standardised eSports training protocol on reaction time and decision-making performance. The independent variable was the type of training intervention (individualised versus standardised), while the primary dependent variables were choice reaction time (CRT), reaction time under load (RT-L) and double-stimulus reaction time (DSRT). Secondary outcomes comprised simple reaction time (SRT), performance on the Go/No-Go task, and decision-making indices derived from the modified Iowa Gambling Task (IGT).

The study followed a repeated-measures design with three assessment points (T1, T2 and T3). Gender and age were treated as moderators, while gaming experience, game type (MOBA/RTS/FPS) and circadian activity rhythms were included as covariates.

Participants were randomly assigned to two groups, with individual participants serving as the unit of randomization. Randomization was performed using a computerized random number generator integrated into the Python programming environment. Stratified randomization (block size of six) was employed to balance the groups based on key variables (age, gender, and game type) to ensure that all subgroups were proportionally represented across experimental conditions.

Allocation concealment was ensured by implementing an automated assignment procedure within the experimental software environment. This prevented the researchers involved in data collection from accessing the group allocation sequences.

To minimise potential allocation bias, the randomisation sequence was generated by a researcher not involved in participant assessment or data analysis. The study was conducted using a two-arm, parallel-group design, and no changes were made to the methods after the trial commenced.

Trial registration and protocol

The study protocol was developed before participant recruitment began and was approved by the institutional ethics committee (protocol no. 2024-17-KS). The trial was conducted in accordance with the established reporting standards for randomised controlled trials.

Due to institutional and regulatory constraints at the time the study was initiated, the trial was not prospectively registered in a public clinical trial registry. This should be considered a methodological limitation. However, the study protocol, including the predefined hypotheses, outcome measures and analytical strategy, was finalised before data collection began and remained unchanged throughout the study.

Participants

The total sample size consisted of N = 184 individuals, who were randomly assigned to two research groups: Group A (experimental group) — n = 92, and Group B (control group) — n = 92. Due to the nature of the intervention, it was not feasible to blind the participants completely. However, they were not informed about the study hypotheses or the existence of alternative intervention conditions. Outcome assessment was conducted using automated computerised tasks to minimise the potential for assessor-related bias. The researchers supervising the testing sessions followed standardised protocols and did not influence participant performance. The study was conducted at a specialised esports training academy in controlled laboratory conditions. The main socio-demographic characteristics of the study participants are presented in Table 1. All participants were recruited from three major cities in eastern China: Shanghai (38%), Guangzhou (34%), and Nanjing (28%). Participants’ ages ranged from 18 to 25 years. Inclusion criteria included: absence of chronic neurological or psychiatric disorders, normal or corrected-to-normal vision, and hearing. Exclusion criteria were: use of psychoactive substances (including stimulants and sedatives), concurrent participation in other cognitive or physical intervention programs, and sleep disturbances (defined as sleeping less than six hours per night for more than three nights per week). Recruitment was carried out through a partner educational and sports organization via electronic mailing lists, information platforms, and announcements posted on internal forums. Potential participants underwent a two-stage screening process, consisting of an online questionnaire followed by an expert interview. Overall, 92.8% of participants completed the full intervention course and post-testing.

Table 1.

Socio-demographic characteristics of the study sample

	Total (N = 184)	Group A (n = 92)	Group B (n = 92)
Age, mean (SD)	22.6 (2.9)	22.4 (2.8)	22.8 (3.0)
Age cohorts (%)
18–21	38%	40%	36%
22–25	44%	43%	45%
26–29	18%	17%	19%
Gender (%)
Male	71.4%	73.8%	69.0%
Female	28.6%	26.2%	31.0%
Ethnicity (%)
Han	100%	100%	100%
Native Language (%)
Putonghua (Mandarin)	95%	95%	95%
Cantonese	5%	5%	5%
Educational Level (%)
Higher/Incomplete Higher Education	78%	76%	80%
Secondary Specialized Education	22%	24%	20%
Socio-Economic Status (%)
Upper-Middle	13%	12%	14%
Middle	64%	66%	62%
Lower-Middle	23%	22%	24%
Marital Status (%)
Single, living with parents/in dormitory	82%	81%	83%
Living independently	18%	19%	17%

During the early stages of study design, competitive experience was considered a selection criterion (a minimum of 6 to 12 months of continuous competitive experience). However, this restriction was lifted during the revision process to enhance ecological validity and representativeness of the competitive esports population. As a result, the final sample included participants with varying levels of gaming experience.

To address potential bias associated with sample heterogeneity, competitive experience was included in the statistical analysis as a control variable, and stratified randomization was used to balance participants across groups on key baseline characteristics. Baseline variability was assessed using descriptive statistics and independent sample testing to ensure equivalence of cognitive performance measures between groups prior to the intervention.

The planned sample size was justified by a priori power analysis conducted using G*Power 3.1 software. The calculations were based on the following parameters: type of analysis — two-way repeated measures ANOVA (time × group), desired statistical power (1 – β): 0.90, α (probability of Type I error): 0.05, and expected effect size (f): 0.30. Based on these parameters, the minimum required sample size to detect a statistically significant effect under the specified conditions was n = 172 participants (i.e., 86 per group). To compensate for potential attrition, an additional 12% was added to the target sample, resulting in a final planned size of N = 184.

All procedures were conducted in accordance with the Declaration of Helsinki. Prior to enrollment, all participants signed written informed consent forms that included detailed explanations of the study’s purpose, potential risks, participation conditions, the right to withdraw without penalty, and assurances regarding the confidentiality of personal data and anonymity during the study. Participant safety and the appropriateness of the intervention were continuously monitored by a professional oversight team consisting of neuropsychological and physical health experts. In cases of temporary health deterioration (two cases of transient fatigue), immediate compensatory measures were implemented by temporarily suspending the training.

During the intervention period, seven participants (3.8%) discontinued participation due to scheduling conflicts or temporary health reasons, resulting in a final completion rate of 92.8%. All participants who completed at least 90% of the intervention sessions and post-testing were included in the final analysis.

The study adopted a per-protocol analytical approach as none of the participants met the predefined exclusion criteria based on adherence thresholds. No serious adverse events were observed during the study. Two cases of minor transient fatigue were reported and managed by temporarily suspending training sessions.

Research tools

Reaction time was measured using the Reaction Time Suite, an experimental environment based on Python (PsychoPy), widely used in cognitive and psychophysiological research, a computerized battery comprising five types of tasks adapted for investigating cognitive speed in high-level athletes. SRT (Simple Reaction Time): Participants were required to press a key as quickly as possible upon the appearance of a visual stimulus (a white circle on a black background). Each task included 30 stimuli presented at pseudo-random intervals ranging from 1.2 to 2.8 s. CRT (Choice Reaction Time): Participants had to select the correct response from three options, with red, green, and blue light signals corresponding to different keys.

Go/No-Go Task: This task assessed impulsivity and action inhibition, requiring participants to respond only to certain signals. RT under Load (RT-L): Reaction time was measured while simultaneously performing a cognitive task (e.g., executing arithmetic operations in parallel). Double Stimuli Reaction Time (DSRT): Reaction time was assessed using combined visual and auditory stimuli to evaluate multisensory response. Decision-making efficiency was measured using a modified Iowa Gambling Task (IGT), adapted to a gaming context, where traditional playing cards were replaced with scenarios simulating game-related situations (e.g., choosing a tactic offering short-term versus long-term advantage). The task consisted of a standard 100 trials, divided into five blocks of 20 trials each. After each choice, participants received immediate feedback about winning and losing.

The primary outcome measure was the net score, defined as the number of favorable choices minus the number of unfavorable choices [(C + D) − (A + B)]. Additionally, net scores were calculated across blocks to assess learning dynamics during task completion. Decision stability was defined as the consistency of choice patterns across presentation blocks (instead of the deck in a traditional test) across comparable test segments.

Data collection focused on the number of advantageous decisions, mean latency to decision, and decision stability (coefficient of repeated choices under similar conditions).

Internal consistency coefficients for the primary measures—reaction time and decision-making efficiency—were high, with Cronbach’s α exceeding 0.85, indicating strong instrument reliability. Test-retest reliability assessed over an 8-week interval demonstrated stable results (r = 0.81–0.85). Inter-rater reliability was also high, with Intraclass Correlation Coefficients (ICC) ranging between 0.86 and 0.88. Convergent validity was confirmed through correlations with other cognitive assessments such as the Trail Making Test – Part B (TMT-B) and the Wisconsin Card Sorting Test (WCST) (r = 0.68–0.74), whereas discriminant validity was supported by the absence of significant correlations with somatic measures, indicating the specificity of the constructs being measured.

Outcome measures

The primary outcomes were defined as changes in choice reaction time (CRT), reaction time under load (RT-L) and double-stimulus reaction time (DSRT), as these indicators most directly reflect cognitive processing speed in conditions relevant to eSports performance.

The secondary outcomes included simple reaction time (SRT), accuracy and commission errors on the Go/No-Go task, and decision-making performance indices derived from the modified Iowa Gambling Task (IGT), including net score, decision latency, and decision stability.

All outcomes were assessed at three time points (T1, T2 and T3), with the primary endpoint defined as the change from baseline (T1) to post-intervention (T3).

Theoretical rationale for the training program

According to the Limited Attentional Resources Theory, mindfulness-based self-regulation is activated to reduce attentional demands unrelated to task performance. This increases the share of available cognitive resources devoted to task-relevant stimuli. Adaptive cognitive-motor tasks were developed to facilitate the transition from controlled to automatic processing, reducing attentional demands during decision-making under high stress conditions.

Consistent with the Transactional Model of Stress and Coping, mindfulness practices were incorporated into the training to influence primary appraisal by reducing perceived threat and dampening emotional reactivity. Aerobic exercise was expected to improve secondary appraisal by strengthening physiological and psychological coping resources. These components, taken together, were hypothesized to reduce stress-related attentional depletion during complex competitive tasks under high stress.

Although baseline motivation was not assessed using specialized psychometric scales, participants in both groups were recruited from the same competitive environments and randomized using a stratified procedure, which reduced systematic differences in motivational orientation. Training adherence was monitored throughout the intervention period; it was assumed that participants who missed more than 10% of scheduled sessions would be excluded from the final analysis. However, no participants were actually excluded, which may be explained by certain cultural characteristics of the Chinese sample (high social and group responsibility and a habit of complying with requirements).

Training program

Group A

Personalization of the psychophysical training program was implemented using a rule-based adaptive approach. At baseline (T1), individual profiles were established for each participant based on reaction time measures (CRT, RT-L, DSRT), decision-making indices, and self-reported fatique levels. These baseline values served as reference points for subsequent adjustments. Training difficulty was adjusted primarily by modifying task complexity (stimulus density, time pressure, and dual-task load), session pacing, and recovery intervals.

For cognitive tasks, the following were implemented: decreasing the interstimulus interval; increasing the number of distractors; switching from single-task to dual-task exercises; and increasing the stimulus presentation rate. For the aerobic component of exercise, the target range relative to HR (% HRmax) was varied; the duration of continuous exercise was varied; and the work-to-rest ratio during exercise was changed. Adjustments were implemented on a bi-weekly basis following mid-intervention assessments (T2), and optionally when predefined performance or fatique thresholds were exceeded.

All esports athletes participated in three training sessions per week. All sessions were held in the morning, lasting 60 min. Each training session included a warm-up phase aimed at mobilizing joints most frequently used by esports athletes, incorporating specific flexion and extension movements for the shoulders, wrists, back, and neck. Additionally, resistance exercises using elastic bands were included to promote external shoulder rotation and activation of the deep abdominal muscles. Two of the weekly sessions were dedicated to strength training focused on developing upper and lower body strength in accordance with established recommendations. The third session each week was devoted to aerobic training to foster adaptations that would enhance the body’s capacity to endure prolonged physical activity, particularly during training and competition sessions exceeding five hours. The intervention protocol was specifically designed to develop athletes’ muscular strength and cardiovascular endurance.

A range of cognitive tasks was developed to stimulate information processing speed. In particular, training for visuospatial attention involved a series of tasks requiring players to react quickly and accurately to changing conditions. The program incorporated exercises combining elements of agility and strategic planning, demanding that participants not only respond rapidly to new situations but also anticipate the potential consequences of their decisions. To maximize effectiveness, each exercise involved scenario variation and required participants to make predictions within multifaceted tasks, thus training both tactical and strategic thinking.

The program also included emotional self-regulation training using mindfulness and meditation techniques to help participants maintain internal stability and focus during high-pressure gaming situations. To enhance reactivity, 3D exercises such as NeuroTracker, simple and choice reaction tests (PEBL, CogniFit), and daily sessions in Osu! or Aim Lab were employed to improve visuomotor accuracy. Decision-making under pressure was modeled through tasks involving object trajectory prediction, dynamic timed simulations, and cognitive paradigms such as the Stop-Signal and Go/No-Go tasks aimed at improving impulse control.

Group В

The groups A and B trained with the same frequency and session duration; the main difference was the inclusion of adaptive psychophysical components in the experimental group’s training.

The standard esports training program was designed to provide a basic level of activity without the incorporation of targeted cognitive or psycho-physical interventions. The program consisted of three weekly sessions, each lasting up to 60 min, and included typical elements of daily esports training focused solely on gameplay practice. Each session began with a brief visual warm-up in the form of a few minutes of gameplay using Aim Lab or Kovaak’s FPS Trainer, without specific task selection aimed at enhancing reaction time or accuracy. The main portion of the session consisted of playing in a standard competitive mode (ranked or scrim), emphasizing team interaction but without structured analytics or targeted cognitive stimulation. The session concluded with a short, informal tactical discussion without the involvement of a coach-analyst. Throughout the program, no exercises targeting concentration, selective attention, impulse control, or decision-making processes were incorporated.

All training sessions were supervised by certified instructors who had experience in esports performance training. They followed a standardised protocol to ensure consistency across participants.

Data collection

Testing was conducted in three phases: T1 — prior to the intervention, T2 — after the fourth week, and T3 — following the eighth week of training. Experimental data were collected from October to December 2024 at an academy equipped with approved infrastructure for intensive training of competitive players and an isolated laboratory space for testing. All interventions and pre- and post-assessments were conducted in a controlled environment with constant temperature, absence of external distractions, and standardized hardware settings (144 Hz monitor, identical peripheral devices, and uniform lighting conditions). All data were collected electronically through a dedicated web portal developed by the research team. All research assistants (n = 6) underwent two weeks of specialized training, which included familiarization with testing protocols and hands-on practice administering tasks to control subjects. To enhance measurement accuracy, each key task was performed in triplicate, and mean values were used for analysis. Additionally, retesting was conducted on a random subsample (20% of participants) with a 48-hour interval to assess intra-test reliability, which yielded a stability coefficient of r > 0.88.

Data analysis

IBM SPSS Statistics software (version 26) was used for data analysis. An independent samples t-test was applied at the pre-assessment stage to evaluate baseline differences between groups. To analyze changes following the intervention, a repeated measures ANOVA was employed.

To examine interactions between the type of intervention and moderators (gender, age), multifactorial regression modeling and analysis of covariance (ANCOVA) were conducted. To assess the influence of demographic moderators on outcomes, a moderation analysis using multiple regression models with interaction terms was performed, allowing for the evaluation of whether significant differences in training effectiveness existed between male and female participants and between younger and older participants.

Baseline variance within the sample was assessed using standard deviations, confidence intervals, and distributional analysis of cognitive performance measures. Between-group equivalence was verified using independent samples t-tests (Table 2), which demonstrated no statistically significant differences across all baseline indicators and small effect sizes (Cohen’s d < 0.30). To minimize the influence of sample heterogeneity on intervention estimates, analyzes focused primarily on within-subject change and Time × Group interaction effects. In addition, demographic variables and gaming experience were treated as covariates or moderators where appropriate. This analytical approach reduces sensitivity to baseline variability and provides more reliable estimates of intervention effects under heterogeneous conditions.

Table 2.

Results of the independent samples t-test for pre-intervention group differences in reaction time and decision-making indicators

Indicator	Group A (M ± SD)	Group B (M ± SD)	t(df)	p	Cohen’s d	95% CI for d
Reaction Time
SRT (ms)	241.6 ± 21.4	244.1 ± 20.9	-0.52(169)	0.603	0.12	[-0.19; 0.43]
CRT (ms)	318.5 ± 29.2	321.8 ± 31.1	-0.48(169)	0.631	0.11	[-0.20; 0.42]
Decision Accuracy (%)	73.2 ± 6.4	72.8 ± 5.9	0.29(169)	0.774	0.06	[-0.38; 0.50]
Decision Latency (ms)	1092.4 ± 130.5	1111.7 ± 127.8	-0.65(169)	0.518	0.15	[-0.29; 0.59]
Go/No-Go Accuracy (%)	87.5 ± 5.2	86.9 ± 5.5	0.48(169)	0.632	0.11	[-0.33; 0.55]
Go/No-Go Commission Errors (#)	3.2 ± 1.1	3.5 ± 1.4	-1.10(169)	0.274	0.26	[-0.18; 0.70]
RT under Load (ms)	683.3 ± 45.7	689.2 ± 48.1	-0.53(169)	0.599	0.12	[-0.19; 0.43]
DSRT (ms)	263.9 ± 23.8	267.1 ± 24.2	-0.59(169)	0.556	0.13	[-0.18; 0.44]
Decision-Making
Advantageous Decision Index (IGT), points	17.14 ± 3.92	17.08 ± 4.03	0.092(158)	0.927	0.01	[-0.22; 0.24]
Decision-Making Latency (ms)	2140.55 ± 184.11	2125.39 ± 191.67	0.478(158)	0.633	0.08	[− 0.23; 0.39]
Decision Consistency Index (%)	81.22 ± 6.45	80.77 ± 6.88	0.401(158)	0.689	0.07	[− 0.24; 0.38]

The primary analysis focused on Time × Group interactions, defined a priori. To reduce the risk of inflated Type I error, interpretation of results was based on effect size and convergence of results across conceptually related measures. Given that statistically significant effects were accompanied by small to moderate effect sizes (η² ≈ 0.08–0.12), interpretation was based on practical significance rather than individual p values. To control for inflated Type I errors, Bonferroni-corrected significance thresholds were calculated separately for groups of related hypotheses (baseline comparisons, interaction main effects, and moderation analysis). The adjusted thresholds did not change the interpretation of the Time × Group main effects.

Due to the high level of adherence and minimal attrition, the primary analysis was conducted using a per-protocol approach including participants who completed at least 90% of the intervention and post-testing. Sensitivity analyses showed that excluding participants with incomplete data did not significantly impact the observed results.

Research limitations

An additional limitation is the lack of prospective trial registration, which may affect the transparency of the study design. This should be considered when interpreting the findings. Furthermore, due to the nature of the intervention, it was not feasible to blind all participants, which may introduce performance-related bias. The study has several limitations related to the operationalization of the intervention, measurement strategies, and sample characteristics. Although the training program was individualized using an adaptive approach, the personalization process did not rely on fully automated or continuously optimized algorithms. Thus, the degree of individualization may not fully capture the complexity of interindividual differences in learning trajectories, potentially reducing the magnitude of the observed effects. The study also relied on behavioral measures without direct neurophysiological or psychophysiological assessments, which limits the ability to infer the underlying neural or stress-regulatory mechanisms responsible for the observed improvements. Although gaming experience was included as a control variable, the absence of detailed stratification by level of competitive experience, duration of engagement, or performance ranking may have contributed to residual heterogeneity in baseline skill levels and training responsiveness. Such variability may partially explain individual differences in intervention effects, particularly for speed-based measures, and should be considered when interpreting effect size estimates. Finally, the observed results should be interpreted with caution when generalizing beyond the studied population of Chinese esports athletes aged 18–25 years.

Results

Figure 1 shows participant flow throughout the study. A total of 198 individuals were assessed for eligibility, 14 of whom were excluded prior to randomisation.

Fig. 1.

CONSORT flow diagram of participant recruitment, allocation, follow-up, and analysis

The remaining 184 participants were randomly allocated to the intervention groups. Seven participants discontinued participation during the study period, resulting in a final sample of 171 participants being included in the analysis. The reasons for discontinuation included scheduling conflicts and temporary health conditions. No adverse events requiring medical intervention were reported during the study. Primary analyses were conducted using a per-protocol approach, which included only participants who had completed at least 90% of the intervention and post-intervention assessment. There was minimal missing data (< 5% across all variables), which was handled using listwise deletion for each analysis. Sensitivity analyses confirmed that excluding incomplete cases did not significantly alter the overall pattern of results.

At the first stage of statistical analysis, the presence of initial between-group differences between participants in Group A and Group B was assessed for key variables, including reaction time and decision-making efficiency (indices derived from the modified Iowa Gambling Task). To test the hypothesis of no statistically significant differences prior to the intervention, an independent samples t-test was applied.

The results of the analysis are presented in Table 2.

There were no statistically significant baseline differences between the groups for any of the primary or secondary outcome measures (all p > 0.05), which confirms successful randomisation. The Cohen’s d values fell within the range of a small effect size, further confirming the absence of substantial baseline disparities between the groups.

The observed standard deviations indicate moderate within-sample variability in reaction time and decision-making indicators, reflecting heterogeneity in baseline performance characteristics typical of competitive esports populations. However, the absence of statistically significant baseline differences and the small effect sizes confirm that randomization effectively balanced this variability across experimental conditions.

To analyze the intervention effects, a repeated measures analysis of variance (ANOVA) was employed (Table 3), which allowed for the assessment of the effects of time, group membership, and their interaction on cognitive indicators.

Table 3.

Main Results of the Repeated Measures ANOVA

Indicator	Time effect (F, p, η²)	Group effect (F, p, η²)	Interaction Time × Group (F, p, η²)	95% CI
SRT	F(2,340) = 21.84, p = 0.000, η² = 0.22	F(1,169) = 0.43, p = 0.514, η² = 0.01	F(2,340) = 9.62, p = 0.000, η² = 0.11	[0.06; 0.18]
CRT	F(2,340) = 26.19, p = 0.000, η² = 0.25	F(1,169) = 1.02, p = 0.316, η² = 0.01	F(2,340) = 11.43, p = 0.000, η² = 0.13	[0.07; 0.20]
Decision Accuracy	F(2,340) = 18.71, p = 0.000, η² = 0.19	F(1,169) = 1.34, p = 0.250, η² = 0.02	F(2,340) = 10.98, p = 0.000, η² = 0.12	[0.06; 0.19]
Decision Latency	F(2,340) = 23.65, p = 0.000, η² = 0.23	F(1,169) = 0.56, p = 0.456, η² = 0.01	F(2,340) = 8.89, p = 0.000, η² = 0.10	[0.05; 0.17]
Go/No-Go Accuracy	F(2,340) = 14.39, p = 0.000, η² = 0.16	F(1,169) = 0.72, p = 0.398, η² = 0.01	F(2,340) = 7.34, p = 0.001, η² = 0.09	[0.04; 0.16]
RT under Load	F(2,340) = 19.52, p = 0.000, η² = 0.20	F(1,169) = 0.95, p = 0.334, η² = 0.01	F(2,340) = 10.27, p = 0.000, η² = 0.12	[0.06; 0.19]
DSRT	F(2,340) = 17.08, p = 0.000, η² = 0.18	F(1,169) = 0.67, p = 0.414, η² = 0.01	F(2,340) = 9.81, p = 0.000, η² = 0.11	[0.06; 0.18]

All assumptions required for ANOVA—namely, equivalence, normality, and homogeneity of variances—were verified a priori.

The main effect of time was statistically significant for all cognitive indicators (p < 0.001), indicating the presence of changes regardless of group membership. The group effect was not statistically significant, suggesting no consistent differences between groups when ignoring the dynamic aspect of change. The key effect — the Time × Group interaction — was statistically significant across all parameters (p < 0.001), indicating differing trajectories of cognitive improvements between Group A and Group B, indicating a more pronounced improvement trajectory in the individualized training group. The results confirm that Group A exhibited marked improvement across all indicators, particularly in CRT, RT-L, and DSRT. The Decision Accuracy Index increased in both groups, but significantly faster and more consistently among participants in Group A, as indicated by a significant Time × Group interaction (F(2,156) = 10.98, p < 0.001).

Decision Latency was substantially reduced over time, again showing a more favorable trajectory in the individualized training group (F(2,156) = 8.89, p < 0.001). The main group effect itself remained statistically non-significant, underscoring the critical role of training-induced dynamics rather than baseline differences. The observed interaction effects correspond to small-to-moderate effect sizes (η² = 0.09–0.13), indicating that the personalised training programme led to practical, meaningful improvements in cognitive performance over time compared to the control condition.

To assess the influence of demographic moderators on the effectiveness of the psycho-physical intervention, a multifactorial regression modeling was conducted, incorporating the variables “type of intervention,” “gender,” “age,” and their interactions (Table 4).

Table 4.

Results of the moderation analysis for reaction time and decision-making indicators

Predictor	B (CRT)	SE (CRT)	β (CRT)	p (CRT)	B (DPI)	SE (DPI)	β (DPI)	p (DPI)	95% CI (β)
Type of Intervention	-31.84	8.12	-0.364	0.000	0.219	0.063	0.388	0.000	-
Gender	5.72	6.44	0.083	0.375	-0.032	0.049	-0.064	0.517	-
Age	1.27	0.92	0.105	0.172	-0.018	0.007	-0.205	0.012	-
Type of Intervention × Gender	-18.64	7.31	-0.234	0.013	0.067	0.052	0.121	0.198	[-0.06; 0.30]
Type of Intervention × Age	-2.17	1.08	-0.162	0.047	-0.028	0.008	-0.278	0.008	[-0.46; -0.10]

Regression analyses were conducted on the final sample (n = 171)

The interaction between type of training and gender was statistically significant (β = − 0.234, p < 0.05) with respect to changes in CRT — that is, females in the experimental group exhibited a greater improvement in choice reaction time compared to males. For the Decision Performance Index (DPI), a significant interaction was found between age and type of intervention (β = − 0.278, p < 0.01), indicating that younger participants demonstrated greater sensitivity to individualized training in decision-making performance. The robustness of the estimated interaction effects was confirmed by checking all regression models for multicollinearity (VIF < 2.5) and normality of residuals. To visually represent the moderation analysis, interaction plots were constructed (Figs. 2 and 3). The graphical patterns align with the statistically significant interaction terms presented in Table 4.

Fig. 2.

DPI by age and intervention

Fig. 3.

CRTby age and intervention

The slopes of the lines indicate age-related variability in training responsiveness. In the case of DPI, a steeper increase in performance was observed among younger participants in the experimental group, suggesting better outcomes within the cohort under 25 years of age. For CRT, the association was less pronounced but remained statistically significant.

According to the graphical results of the interaction analysis between age and type of intervention for changes in choice reaction time (CRT), a general trend toward decreased CRT with increasing age was observed across all participants, regardless of the type of training. However, the individualized intervention consistently resulted in lower CRT values compared to the standard approach across the entire age range, indicating higher effectiveness of the individualized program in optimizing choice reaction time and cognitive information processing.

A steeper regression line for CRT decline was observed in the control group, suggesting greater age sensitivity to the standard intervention, whereas individualized training produced stable improvements in CRT irrespective of participants’ age. No serious adverse events were reported during the intervention period. A small number of participants experienced minor transient fatigue, but this did not result in their withdrawal from the study.

Discussion

Principal findings and hypothesis testing

The primary hypothesis was that personalised cognitive training would be more effective than standard formal instruction. The results support this, as participants in the personalised training group showed greater improvement in decision-making and reaction times under cognitive load than those in the control group. The secondary hypothesis proposed that age and gender would moderate the intervention effect, which was partially confirmed. Age emerged as a moderating factor for DPI indicators, with the effectiveness of cognitive training somewhat declining with increasing age, particularly in the standard intervention group. Gender exhibited a selective effect: females demonstrated greater persistence in decision-making following specialized training, although this was not consistently observed across all cognitive measures, suggesting that the role of gender warrants further investigation. Although the results point to the high effectiveness of the personalized approach, several potential sources of bias must be considered in interpreting the findings.

Importantly, the observed effects were associated with small-to-moderate effect sizes (η² ≈ 0.09–0.13). This indicates that, while the differences between interventions were statistically significant, they should be interpreted as moderate in magnitude. The consistency of the findings across multiple primary and secondary outcomes increases confidence in the reliability of the observed effects of the intervention.

Robustness, potential bias, and sample heterogeneity

When interpreting the findings, it is important to consider that the trial was not prospectively registered and that the primary analyses were conducted using a per-protocol approach. These factors may increase the risk of bias and limit transparency in the analytical framework. Importantly, the consistency of Time × Group effects across multiple reaction time measures suggests that the findings are unlikely to reflect isolated false positives. Although corrections for multiple comparisons were applied, residual Type I error cannot be entirely ruled out, so replication in independent samples is necessary. Unexplained factors such as baseline motivation, prior experience, and participants’ psychophysiological characteristics may also have influenced the dynamics of change. Procedural biases, including instruction consistency, participants’ contextual motivation, and cognitive fatigue, should be carefully addressed. In multitask testing environments, there is a risk of cognitive overload, which could compromise internal validity and lead to false-negative results. Moreover, the potential overlap between tasks activating adjacent neurocognitive systems may complicate the distinction between functional domains.

An important methodological consideration concerns the experience of heterogeneity of gaming and baseline cognitive performance within the sample. Although variability in experience levels may introduce additional noise and potentially attenuate effect sizes, the consistency of Time × Group interactions across multiple outcome measures suggests that the observed effects are robust to such variability. The inclusion of heterogeneous participants may in fact enhance external validity by reflecting real-world diversity in esports populations, although future studies may benefit from more detailed stratification by competitive experience level.

Psychological and cognitive mechanisms of training effects

The obtained results may be meaningfully interpreted using a combined approach incorporating the Lazarus-Folkman Transactional Model of Stress and Kahneman’s Theory of Limited Resources. From a transactional stress perspective, the integrated mind-body intervention could change participants’ primary and secondary stress appraisals in cognitively demanding tasks, reducing stress-related cognitive interference and promoting more effective coping strategies [50]. Previous studies have demonstrated that mindfulness-based self-regulation can improve attentional control and modify the stress appraisal process, which contributes to improved cognitive performance [51]. The observed improvements in reaction time under load and decision stability are consistent with the Limited Attentional Resources Theory, suggesting that the intervention reduced the consumption of task-irrelevant attention [16, 47]. The findings provide convergent support for theoretical models suggesting that stress regulation and attentional resource optimization are key mechanisms underlying improved performance in high-load digital environments such as esports [52–55].

The improvements observed in the experimental group can also be explained by adaptation to cognitive load: unlike standardized programs, individualized programs provided a better balance between physical activity and information processing capacity [56, 57]. The possibility of training attentional shifting and suppression of impulsive reactions (notably through the Go/No-Go, DSRT, and RT-L tasks) suggests a progression from controlled to automatic processing—marked by a reduction in cognitive load alongside increased accuracy, indicating a transition to a more efficient neurocognitive decision-making architecture. Participation in specialized training sessions may have stimulated the development of metacognitive control strategies (self-assessment of responses and real-time behavioral adjustment), contributing to improved decision-making stability [58, 59].

Moderation by age and gender: interpretation and implications

In particular, the demonstrated comparatively greater improvement of the individualized approach in improving indices of cognitive flexibility and decision-making speed aligns with findings from studies that emphasized the adaptive potential of training in high-stimulus-variability environments [60–62]. These parallels are noteworthy, as they reinforce the hypothesis regarding the importance of contextually relevant cognitive load for stimulating fronto-cingulate circuits responsible for action control and selective attention. However, in the present study, the moderating effect of age appeared more pronounced, showing a clear trend toward reduced training efficacy among older subgroups. This somewhat contrasts with other findings, where adaptive training platforms demonstrated consistent advantages across age strata [63, 64]. A plausible explanation for this phenomenon may lie in the specific cognitive profile of our sample participants, characterized by a heightened baseline arousal level and a short latency threshold for action, typical of the esports environment, which could result in reduced neural network plasticity among older individuals under monotonous or insufficiently personalized cognitive loads.

Equally noteworthy is the partial divergence observed in the influence of gender on decision-making stability. The observed moderator effects of age and gender may suggest that responses to integrated mind-body training in sports may be heterogeneous. One plausible explanation for the stronger effects observed in female participants is related to gender differences in cognitive and neurophysiological responses to aerobic exercise in self-regulation training, previously identified by other researchers [16]. These studies demonstrated improved executive functions and attentional control in women, which influence cognitive plasticity. Such mechanisms may enhance the effectiveness of interventions targeting reaction time and decision-making stability. This effect, as our study suggests, may be particularly pronounced in tasks requiring inhibitory control and sustained attention. Age moderation may reflect differences in the availability of cognitive resources and age-related responses to training. Younger participants may demonstrate greater gains in cognitive speed tasks due to greater baseline neuroplasticity and faster adaptation to increasing task demands.

Previous studies by several researchers have shown that training-induced cognitive changes are shaped by individual brain dynamics and baseline performance characteristics [37, 38]. Younger individuals show greater improvements in reaction time. In the present study, these factors may have contributed to the more pronounced improvements seen in younger participants, particularly under cognitive load. Importantly, the effects of age and gender on training results were minor, suggesting that while individual characteristics influence training effectiveness, the comprehensive intervention remains generally beneficial across all demographic groups.

Practical implications

The findings support the argument that cognitive training should be integrated into broader educational and professional development programs, particularly in domains where speed and accuracy of responses critically impact performance. Although these findings relate specifically to eSports athletes, future studies could explore similar intervention principles in other high-demand cognitive environments, such as professional or educational settings. In esports teams, the results may be implemented to individualize training and improve performance over long periods of training activity or to reduce stress and increase efficiency in the pre-competition period. The moderate effects of age and gender emphasize the importance of accounting for demographic diversity when designing cognitive training platforms, necessitating inclusive designs and differentiated learning scenarios that address users’ psychophysiological characteristics. The demonstrated potential for enhancing central cognitive processes may be applied in healthcare and professional esports development, focusing on preventing cognitive fatigue, improving performance, and promoting neuropsychological resilience.

Neurobiological mechanisms underlying training effects

Please note that this study did not include direct neurophysiological measurements. Therefore, the following interpretations are theoretical and are based on existing literature. The behavioral improvements observed in the present study may also be interpreted in light of contemporary findings from cognitive neuroscience regarding the neural mechanisms underlying cognitive training and physical exercise. Emerging evidence suggests that combined physical and cognitive interventions can modulate prefrontal cortex functioning, enhance neuroplasticity, and influence neurochemical processes that support executive control and attentional regulation.

Aerobic exercise has been consistently associated with structural and functional changes in brain regions involved in executive functioning, particularly the prefrontal cortex and hippocampus. These effects are mediated by increased cerebral blood flow, enhanced oxygenation, and the upregulation of neurotrophic factors such as brain-derived neurotrophic factor (BDNF), which supports synaptic plasticity, neurogenesis, and neural network efficiency. Such changes are linked to improvements in attentional control, cognitive flexibility, and processing speed, which may contribute to faster reaction time and more stable decision-making under cognitive load [40, 41].

Experimental evidence further indicates that moderate-intensity exercise combined with cognitive demands increases prefrontal cortex activation, suggesting enhanced top-down control of behavior and improved allocation of attentional resources. These mechanisms are consistent with observed improvements in inhibitory control and response selection efficiency [39]. In addition to structural and functional adaptations, physical exercise induces neurochemical changes, including increased levels of catecholamines, dopamine, and serotonin, which contribute to improved cognitive processing speed and reduced stress reactivity. These neurochemical effects may facilitate adaptive decision-making and response selection in high-demand environments such as competitive esports [41, 43].

Mindfulness-based self-regulation components of the intervention may further contribute to cognitive improvements through modulation of large-scale brain networks involved in attention and emotional regulation, including the default mode network and prefrontal control systems. Neuroimaging studies demonstrate that sustained mindfulness practice alters functional connectivity and enhances prefrontal activation during cognitively demanding tasks, supporting improved attentional stability and reduced susceptibility to distraction [42]. Together, these converging neurobiological mechanisms provide a plausible explanation for the observed improvements in reaction time and decision-making efficiency, suggesting that integrated psycho-physical training enhances cognitive performance through coordinated changes in neural efficiency, attentional regulation, and stress-related neurocognitive processes.

Conclusions

The findings of this randomised controlled trial suggest that, compared to a standardised training protocol, the psychophysical intervention was associated with statistically significant improvements in reaction time, impulse control and decision-making efficiency in young eSports athletes. The most notable improvements were recorded in RT-L (F(2, 340) = 9.84, p < 0.001, η² = 0.11), Go/No-Go (F(2, 340) = 8.51, p < 0.001, η² = 0.10), DSRT (F(2, 340) = 7.96, p < 0.001, η² = 0.09), as well as components of the Iowa Gambling Task, particularly the Decision Consistency Index (F(2, 340) = 10.21, p < 0.001, η² = 0.12) and Decision Latency (F(2, 340) = 6.73, p = 0.002, η² = 0.08).

Regression models suggested a small but statistically significant moderating effect of age (β = − 0.21, p = 0.038) and gender (β = 0.18, p = 0.047), indicating that the dynamics of cognitive development progression varied slightly depending on demographic variables.

Schools and educational institutions are encouraged to implement programs aimed at developing attention, working memory, and impulse control, particularly under conditions of information overload.

Despite the compelling results, this study leaves several theoretical and methodological questions open and outlines directions for future research. Due to the design and limitations of the study, the findings should be interpreted as evidence of association rather than as definitive proof of causation.

Further studies should explore the effectiveness of the intervention in samples with diverse cultural backgrounds. Replicating the protocol within European or North American contexts would be valuable to assess the universality and adaptability of the cognitive enhancement mechanisms.

An important next step would be to conduct longitudinal studies and cross-cultural replication to monitor the duration and stability of intervention effects—specifically, whether cognitive improvements are sustained over 3, 6, or 12 months post-training, or whether reinforcement interventions are necessary to maintain gains. Future research could also incorporate neuroimaging techniques (e.g., fMRI and EEG) to investigate the neurophysiological mechanisms underlying cognitive improvements, particularly regarding the plasticity of the prefrontal cortex and neural structures associated with attention and decision-making.

Cluster analysis or latent profile analysis should be employed in esports training to identify subgroups of participants exhibiting different trajectories of response to the intervention, which could then be used to develop predictive models aimed at determining which individuals are most likely to benefit from such programs.

Abbreviations

CRT

Choice Reaction Time

RT-L

RT under Load

DSRT

Double Stimuli Reaction Time

IGT

Iowa Gambling Task

ICC

Intraclass Correlation Coefficients

TMT-B

Trail Making Test – Part B

WCST

Wisconsin Card Sorting Test

ANCOVA

Analysis of covariance

ANOVA

Analysis of variance

DPI

Decision Performance Index

Authors’ contributions

Authors’ contributions: Wenxiu Zheng : Conceptualization, Formal analysis, Methodology, Project administration, Supervision, Validation, Writing – review & editing. Hairong Zhang : Data curation, Investigation, Methodology, Resources, Software, Visualization, Writing – original draft.

Funding

The research received no funding.

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

The authors declare that the work is written with due consideration of ethical standards. The research was approved by the local Ethics Committees of Fuzhou Preschool Education College (Protocol No. 2024-17-KS). The study was conducted in accordance with the rules of the Declaration of Helsinki. All subjects gave written informed consent prior to participation.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.