Effects of a 4-week combined breathing control and self-talk intervention on neural activity and coping skills in athletes: a randomized control trial Authors: Kittichai Tharawadeepimuk, Ekarat Onnom, Ampika Nanbancha Published: 2026 ABSTRACT AbstractBackgroundEffective coping skills are essential for athletes aiming to achieve peak performance. Breathing control and self-talk are recognized as commonly used psychological interventions to support coping. However, their combined effects on neurophysiological and psychological markers have not been examined.ObjectiveTo evaluate the effects of a 4-week combined breathing control and self-talk intervention on electroencephalogram (EEG) activity and coping skills in elite national-level athletes.MethodTwenty-six athletes (11 swimmers and 15 gymnasts) were randomized to an experimental or control group. The intervention consisted of 16 sessions over four weeks (four days per week). Resting-state EEG and Athletic Coping Skills Inventory-28 (ACSI-28) scores were collected pre- and post-intervention. EEG power spectral density (PSD) across frequency bands was analyzed using generalized linear mixed models, and ACSI-28 subscales were assessed using repeated measures ANOVA.ResultSignificant group differences were found in theta (p < .001) and beta (p = .048) power. The experimental group showed increased theta power in the prefrontal area (p < .001) and decreased beta power in the frontal area (p = .003). Improvements were also observed in ACSI-28 subscales, including coping with adversity (p = .018), concentration (p = .027), confidence and achievement motivation (p = .039), goal setting/mental preparation (p = .009), and peaking under pressure (p = .043).ConclusionThis combined intervention of breathing control and self-talk was associated with changes in neurophysiological markers and psychological coping skills. EEG findings suggest potential changes in neural processes potentially related to anxiety regulation and attention. However, these should be interpreted as indirect neural correlates rather than direct markers of psychological states. Overall, these findings provide preliminary evidence that the combined intervention may represent a promising approach for supporting pre-competition psychological readiness in athletes.Trial registrationClinicalTrials.gov NCT05042518 (09/01/2021). Prospectively registered.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40359-026-04814-w.Keywords: Breathing control, Self-talk, Electroencephalogram (EEG), Athletic Coping Skills Inventory-28 (ACSI-28), Cognitive control Introduction The most critical and meaningful aspect for athletes in competition is achieving peak performance during the event. Optimal performance is influenced by several key factors, including physical readiness (e.g., overcoming physical fatigue), competitive experience, and cognitive processes such as decision-making and anxiety regulation, as well as coping with psychological pressure, unpleasant emotions, and mental fatigue [1, 2]. Athletes are frequently confronted with numerous stressful situations throughout their training and competitive endeavors. During competitions, they may experience adverse feelings, including fear of failure, self-doubt, and anxiety related to interactions with their coach, while training sessions can induce exhaustion, fatigue, and concerns about injury. Therefore, the capability to cope with these stressors is a crucial factor in preventing diminished performance and potential failure [3]. Lazarus and Folkman (1984) defined coping as “the cognitive and behavioral efforts” used to manage specific external and internal demands [4]. Given the importance of psychological coping in athletic performance, cognitive processes are often an overlooked component of athletic preparation. Insufficient emphasis on cognitive readiness can hinder an athlete’s ability to manage competitive situations and achieve peak performance [5–7]. Optimizing cognitive processes and psychological factors requires systematic psychological training [8]. Psychological skill or mental training can influence athletes’ performances, particularly their attention and focus [9]. Furthermore, various mental training techniques and strategies consistently demonstrate a positive influence on performance and mental skills [10–13]. One example of these is self-talk, a multidimensional psychological strategy. Self-talk refers to the use of positive or negative open or closed statements that individuals direct toward themselves. These statements serve both instructional and motivational purposes [14]. The efficacy of this strategy is well-documented, with research demonstrating its ability to improve motor performance [12], self-efficacy [15], self-confidence [16], and motivation [14]. Additionally, self-talk has been found to reduce anxiety [17] and help manage attention and focus [18]. Consequently, when faced with challenges in competition, athletes use self-talk as a self-regulation strategy to cope with the mental demands of the sport [19]. This strategy can be used either independently or in combination with other psychological skill training [20]. Breathing control training has also been recognized as a fundamental psychological intervention in pre-competition and warm-up routines, playing a crucial role in optimizing athletic performance. This method helps athletes cope with stress, reduce anxiety, maintain cognitive clarity, and sustain attentional focus under competitive pressure [10, 21]. Breathing techniques modulate autonomic nervous system activity, thereby facilitating physiological and psychological readiness. This physiological readiness, in turn, can improve motor performance, reaction time, and decision-making [22–24]. Integrating breathing techniques with therapeutic interventions significantly enhances heart rate variability, suggesting that breathing exercises are valuable tools for improving physiological regulation in individuals with high anxiety sensitivity [25]. Beyond its psychological effects, breathing has also been shown to modulate neural activity. For instance, slow-symmetric breathing with breath-hold phases leads to meaningful alterations in electroencephalography (EEG) parameters, such as beta band power [26]. This finding is consistent with several studies suggesting that EEG provides a more comprehensive and objective evaluation of psychological interventions by offering neural correlates, thereby allowing for an understanding of their impact [27–30]. Although breathing control and self-talk have typically been studied independently, they target complementary mechanisms of self-regulation. Self-talk primarily influences cognitive and emotional domains, including attentional focus [18], motivation [14], self-efficacy [15], self-confidence [16], and anxiety regulation [17]. In contrast, breathing control directly modulates physiological arousal, supporting emotional stability and cognitive clarity through autonomic regulation [10, 21–24]. Controlled breathing also enhances physiological readiness [25], such as heart rate variability, and influences neural activity [26] measured by EEG. Integrating these techniques may therefore provide a more comprehensive self-regulation strategy with self-talk facilitating top-down cognitive control and breathing control providing bottom-up physiological regulation. This combination may have a potentiate effect, whereby increased physiological calmness enhances the effectiveness of cognitive strategies, and improved cognitive framing strengthens the benefits of breathing control. Despite this conceptual rationale, empirical evidence examining this combined technique with EEG and psychological measures in competitive setting remains limited. Crucially, while both techniques independently demonstrate psychological and physiological benefits, research examining their combined application using EEG and the Athletic Coping Skills Inventory-28 (ACSI-28) in competitive athletes remain scarce. Understanding how this integrated intervention influences neural activity and psychological coping mechanisms may contribute to advancing sport psychology and the design of multidimensional mental training programs. The aim of this study was to examine the effects of a combined breathing control and self-talk intervention on EEG parameters and the ACSI-28 in elite national-level athletes. We hypothesized that the combined intervention would produce measurable changes in EEG markers associated with psychological regulation (e.g., attention and anxiety modulation) and improvements in coping skills, which may contribute to enhanced psychological readiness for competition. Specifically, we anticipated alterations in theta [31] and beta [32] power. EEG signals were analyzed across six brain areas using corresponding electrodes, reflecting functional and anatomical brain organization [33, 34]. Materials and methods Participants The primary interaction of interest (Condition x EEG response) in the previous EEG research [35] demonstrated a partial eta-squared = 0.110. This effect was converted to Cohen’s \:f using /(1- ), yielding \:f = 0.350. For a two-group comparison, this effect-size corresponds to approximately Cohen’s d = 0.700 (. Using these effect-size estimates, an a priori power analysis was performed in G*power (α = 0.05, 1 – β = 0.80). For an independent two-sample test approximately 33 participants per group (N = 66) would be required. The present study recruited 13 participants per group (N = 26). Therefore, statistical power for detecting an effect of this magnitude is limited. Observed effect sizes (Cohen’s d) and 95% confidence intervals are reported to support transparent interpretation. Twenty-six elite national-level athletes (7 male, 19 females; mean±SD age: 20.9 ± 3.2 years; height: 166.4 ± 9.2 cm; weight: 58.0 ± 11.3 kg) volunteered for the study, consisting of eleven swimmers and fifteen gymnasts. Exclusion criteria included acute injury, a history of head injury, or neurological disease. Twenty-two participants were right-side dominant and four were left-side dominant. All participants were members of the Thai National Swimming and Gymnastic Teams and were preparing for the Southeast Asian Games, an international regional competition. These athletes train year-round in high-performance programs and have an average of 3.7 years (range 3–9 years) of competitive experience. Athletes were provided with detailed information regarding the study’s objectives, procedures, and potential risks, before providing written informed consent. Participants were randomized to either the experimental or control group. The CONSORT study flow diagram is presented in Fig. 1. Stratified random sampling by sex and sport type was first applied to ensure comparable proportions in both groups, followed by simple random sampling within each stratum. This experiment was approved and performed in accordance with the research ethics committee of Mahidol University - Central Institutional Review Board (COA No. MU-CIRB 2021/147.2206). It was also conducted in accordance with the Declaration of Helsinki. Participant recruitment flow chart Procedure and data collection This study investigated the effects of a combined breathing control and self-talk intervention on neural activity and coping skills in elite national-level athletes. Data were collected at two time points: baseline (pre-intervention) and post-intervention (one week prior to competition). The measurements included electroencephalogram (EEG) data and scores from the Athletic Coping Skills Inventory-28 (ACSI-28). At each measurement point, participants first completed the demographic questionnaire and ACSI-28, followed by having an EEG recording. Resting-state EEG assessments were conducted in the morning for both pre- and post-intervention sessions to reduce circadian influences. Participants received standardized instructions the previous day regarding required sleep duration, restricted medication use, and rest level. Before each session, participants completed a brief readiness check covering sleep quality, fatigue, and overall physical state; testing was rescheduled if the participants were not adequately rested. To minimize variation in training load, the research team coordinated with coaching staff to ensure that training intensity and timing before pre- and post-intervention sessions were comparable. For female participants, menstrual cycle information was collected to avoid testing during periods when hormonal fluctuations may influence the EEG signal. These procedures were implemented to reduce potential confounding influences on the resting-state EEG recording. Data collection was conducted in a dedicated, environmentally-controlled (humidity and temperature) and quiet room at the athletes’ training camp. The same room was used for pre- and post-intervention measurements across both groups. EEG data were acquired while participants were seated in a relaxed, upright posture with their hands resting on their lap to minimize movement artifacts. Participants were explicitly instructed to maintain immobility and open eyes throughout the five-minute recording. To minimize the influence of cognitive processing and ensure a baseline resting-state measure, participants were directed to maintain a neutral mental state, characterized by passive observation without engaging in directed thought or task performance. A demonstration and explanation of the EEG equipment and procedures were provided to ensure participant comfort and familiarity before data collection commenced. All EEG recording and artifact evaluations were performed by a single experienced and trained examiner using a standardized protocol, thereby eliminating inter-rater variability. Consistency across sessions was ensured through fixed recording parameters, identical electrode placement procedures, impedance thresholds (< 20 kΩ), and controlled environmental conditions. Automated artifact rejection based on amplitude thresholds (± 150 µV) further reduced subjective variability. Following the baseline data collection, the sport psychologist explained the intervention and provided practice instructions to participants in the experimental group. Adherence and fidelity were monitored throughout the 4-week program. The sport psychologist documented detail of each session, including attendance and duration. All participants in the experimental group completed 16/16 sessions (100% adherence), and all sessions followed the structured intervention protocol. In the control group, consultations were only provided when requested. One gymnast requested four sessions, and two swimmers requested one session each, resulting in a total of six sessions across the 13 control participants (mean 0.46 sessions per participant). These consultations were unstructured, involved goal-setting techniques, and covered general topics such as quality of life, training, and the weather. Blinding and analyst masking Only the statistical analyst was blinded to group allocations and conducted all data analyses using anonymized datasets coded as group A and group B. Due to the nature of the intervention, it was not possible to blind the sport psychologist or participants to group allocation. Participants were aware of the type of sessions they received (combined self-talk and breathing control intervention vs. routine consultation), and therefore true participant blinding was highly difficult. Combined breathing control and self-talk intervention protocol The intervention was designed to improve the athletes’ ability to cope with anxiety and develop relaxation skills. The combined intervention consisted of psychological techniques, consisting of breathing control and self-talk. The intervention structure adhered to the recommendation that psychological skill training should be performed for 15–30 min per session, 3–4 times per week [36]. Furthermore, slow-paced based breathing was incorporated due to its documented relaxing and anti-anxiety effects [37, 38]. For the experimental group, each intervention session was 15 min in duration, consisting of three 5-minute cycles of the combined breathing control and self-talk intervention. These sessions were performed four times per week (on Mondays, Wednesdays, Fridays, and Sundays) for a period of one month. A total of 16 sessions were administered for the intervention. The steps for combined breathing control and self-talk intervention were as follows: The participants began by sitting in a chair with a backrest, maintaining a comfortable and relaxed posture for one minute. The participants began by sitting in a chair with a backrest, maintaining a comfortable and relaxed posture for one minute. Next, they focused their attention on their breath, becoming aware of the sensation of air entering and exiting through their nose for one minute. Next, they focused their attention on their breath, becoming aware of the sensation of air entering and exiting through their nose for one minute. This was followed by paced breathing, where they inhaled for a 3-second count and exhaled for a 4-second count. This step was performed for 2 min. This was followed by paced breathing, where they inhaled for a 3-second count and exhaled for a 4-second count. This step was performed for 2 min. Then, participants were instructed to say the word “relax” during the exhalation phase for three cycles (30 s). Then, participants were instructed to say the word “relax” during the exhalation phase for three cycles (30 s). Finally, they were instructed to say the word “calm” during exhalation phase for three cycles (30 s). Finally, they were instructed to say the word “calm” during exhalation phase for three cycles (30 s). EEG and ACSI-28 measurements The EEG instrument (eego™mylab, ANT Neuro, USA) consisted of 32 recording channels. These channels were grouped into six anatomical brain areas: Prefrontal (Fp1, Fpz, Fp2), Frontal (F7, F3, Fz, F4, F8, FC5, FC1, FC2, FC6), Temporal (T7, and T8), Central (C3, Cz, C4, CP5, CP1, CP2, CP6), Parietal (P3, P7, Pz, P4, P8, POz), Occipital (O1, Oz, O2). CPz served as the reference electrode, and GND as the ground electrode. Electrode impedances were maintained below 20 kΩ, following established recommendations [39]. All preprocessing steps and spectral analyses were performed using Advanced Source Analysis software (ASA, ANT Neuro, version 4.10.1). EEG signals were recorded at a sampling rate of 512 Hz and preprocessed following widely adopted EEG signal processing procedures [40–43]. Continuous data were band-pass filtered between 0.3 and 30 Hz using a zero-phase Butterworth filter to preserve the temporal integrity of the signal, followed by a 50 Hz notch filter to suppress power-line interference. Artifact detection was performed using an amplitude-based rejection criterion (± 150 µV), which was selected to remove gross artifacts, particularly ocular artifacts such as eye blinks that typically produce large voltage deflections in resting-state EEG recordings [42, 44]. Although more stringent thresholds are often used in event-related potential (ERP) studies (± 75–100 µV) [43], a slightly more liberal threshold is commonly adopted in resting-state spectral analyses to preserve sufficient data for reliable power estimation [41]. The same amplitude-based criterion was applied to exclude other non-physiological artifacts, including movement-related disturbances and high-amplitude EMG contamination. Additionally, slow drifts and high-frequency noise were attenuated via band-pass filtering (0.3–30 Hz), ensuring overall signal quality prior to spectral analysis [41, 44]. Although high-pass filtering attenuates DC components, an additional DC offset correction was applied to eliminate any residual baseline shifts and to ensure zero-mean signals prior to Fast Fourier Transformation (FFT) based spectral analysis, thereby improving numerical stability. Following artifact removal, the cleaned EEG signals were segmented into 2-second non-overlapping epochs for spectral analysis. Power spectral density (PSD) was computed using FFT implemented in ASA software. A Hanning window was applied to each epoch to reduce spectral leakage, resulting in a frequency resolution of approximately 0.5 Hz. PSD values were extracted for the standard frequency bands: delta (0.3–4 Hz), theta (4.5–8 Hz), alpha (8.5–13 Hz), and beta (13.5–30 Hz). No channels met the criteria for bad-channel detection; therefore, no channel interpolation procedures were required. Because the preprocessing pipeline employed automatic amplitude-based artifact rejection, the software does not provide a direct summary of the percentage of rejected epochs. Nevertheless, the rejection threshold and standardized preprocessing ensured consistent treatment of artifacts across participants. The Thai version of the Athletic Coping Skills Inventory-28 (ACSI-28) was used to assess athletes’ coping skills. This version was previously validated in a large sample of Thai youth athletes, demonstrating acceptable internal consistency (overall α = 0.750) and strong construct validity (x2/df = 1.12, RMSEA = 0.018, CFI = 0.990, GFI = 0.930, AGFI = 0.920) while retaining the original seven-subscale structure [45]. The questionnaire consisted of 28 items grouped into seven subscales (four items each): (1) coping with adversity, (2) peaking under pressure, (3) goal setting/mental preparation, (4) concentration, (5) freedom from worry, (6) confidence, and achievement motivation (7) coachability. Participants rated each item on a 4-point Likert scale: 0 (almost never), 1 (sometimes), 2 (often), and 3 (almost always). Cronbach’s alpha coefficients were computed for the current sample to assess internal consistency. Statistical analysis This study analyzed EEG data alongside the ACSI-28. Data normality was assessed using the Shapiro-Wilk test, and a subset of variables exhibited deviation from a normal distribution. Generalized linear mixed models (GLMM) were utilized to examine the linear relationships between groups, EEG across time points, and EEG across different brain areas. Although the raw variables exhibited deviation from normality, model diagnostics indicated that the residuals derived from the GLMM successfully approximated a normal distribution. This outcome, which validates the GLMM approach [46, 47], was verified through both visual inspection and formal testing (histograms, Q-Q plots, and Shapiro-Wilk p; see Supplementary File E). Consequently, a GLMM employing a gamma family distribution with a log link function was utilized for the analysis of the EEG data. Repeated measures ANOVA were conducted to compare groups across the seven subscales of the ACSI-28. As the repeated measures factors possessed only two levels, sphericity was mathematically guaranteed (ε = 1), and Mauchly’s test was not applicable. The use of parametric models is supported by extensive evidence demonstrating that GLMM residual-based assumptions and repeated measures ANOVA are robust to moderate violations of normality. This robustness is particularly evident under balanced experimental designs and when appropriate corrections for sphericity or non-normality are implemented [48, 49]. When significant main effects were identified, post hoc analysis with Bonferroni correction was applied. Effect sizes (d) are reported to determine the magnitude of differences between sex groups and assessed against Cohen’s d definitions of: <0.2, 0.3, 0.5, and 0.8 for trivial, small, moderate, and large, respectively [50]. Finally, correlation analyses were performed to investigate the relationships between the PSD of specific EEG frequency bands and scores from the ACSI-28 questionnaire within each group. Spearman’s rank correlation coefficients were interpreted using the following thresholds: 0 to < 0.3 as negligible, 0.3 to 0.49 as low, 0.5 to 0.69 as moderate, and 0.7 to 0.89 as high, 0.9 to 1.0 as very high [51]. A Comprehensive set of 168 correlations was computed, derived from the cross-analysis of EEG band-power variables and ACSI-28 questionnaire (4 frequency bands x 6 brain regions x 7 ACSI-28 subscales). To mitigate the risk of Type I error inflation resulting from these multiple comparisons, p-values were adjusted using the false discovery rate (FDR) correction, specifically the Benjamini-Hochberg procedure. Statistical significance was uniformly defined as an FDR-adjusted p < .05. Statistical analyses were executed using the Jamovi software suite, supplemented by RStudio for specific computations. Data visualization was subsequently conducted using both RStudio and MATLAB. The results presented are restricted to findings demonstrating statistical significance (p < .05) or identifying critical interactions. Results Baseline demographic, EEG, and pre-test ACSI- 28 data confirmed that the experimental and control groups were equivalent prior to the intervention (see Supplementary File A). The effects of the combined breathing control and self-talk intervention were subsequently examined by comparing the EEG and ACSI-28 data outcomes. EEG data in frequency bands The EEG analysis revealed a significant three-way interaction among group, time, and brain area in the theta (p < .001) and beta (p = .048) power, as illustrated in Table 1. As presented in Table 1, the experimental group exhibited a significant increase in theta power within the prefrontal area compared to the control group at the post-intervention assessment (p < .001, mean difference = 5.04, d = -2.33, 95% CI: 3.62 to 6.46). Concurrently, beta power within the frontal area demonstrated a significant decrease in the experimental group relative to the control group at the same time point (p = .003, mean difference = -5.06, d = 3.46, 95% CI: -7.26 to -2.86). The estimated marginal means, including the mean and 95% confidence intervals (95% CI), are visually represented in Figs. 2 and 3, respectively. Furthermore, grand average EEG topographic maps illustrating power spectral density (PSD) in the theta and beta frequency bands at pre- and post-intervention periods are displayed in Fig. 4. Table 1Generalized linear mixed model (GLMM) of EEG in group, time, and brain areaModel results(Fixed EffectOmnibus tests)Delta powerTheta powerAlpha powerBeta powerDiff d 95%CI p Diff d 95%CI p Diff d 95%CI p Diff d 95%CI p LBUBLBUBLBUBLBUBGroup*Time*Brain area.982<.001*1.00.048*Post Hoc TestsE*Pre*PF – C*Pre*PF10.0-.156-19.339.51.00-.901.501-2.32.5161.00-2.36.253-9.965.221.003.02-.966.8165.221.00E*Post*PF– C*Post*PF-18.3.215-47.611.21.00 5.04 -2.33 3.62 6.46 <.001*-3.07.343-10.74.521.00-2.251.18-4.45-.0461.00E*Pre*PF – E*Post*PF3.10-.033-26.332.51.00 -5.73 2.91 -7.15 -4.31 <.001*.763-.097-6.838.351.00 5.05 -2.37 2.85 7.25 .003* C*Pre*PF – C*Post*PF-25.2.463-54.64.181.00.212-.105-1.211.631.00.062-.010-7.537.651.00-.226.075-2.421.981.00E*Pre*F – C*Pre*F11.0-.411-18.440.41.00-.686.441-2.100.741.00-2.52.291-10.15.071.001.34-.630-.8643.541.00E*Post*F – C*Post*F-14.8.439-44.214.61.001.83-.980.4243.261.00-1.48.179-9.076.111.00 -5.06 3.46 -7.26 -2.86 .003* E*Pre*F – E*Post*F8.22-.242-21.237.61.00-1.36.818-2.78.0561.00-.542.069-8.137.051.00 5.86 -3.23 3.66 8.06 <.001* C*Pre*F – C*Post*F-17.6.665-47.011.81.001.16-.655-.2562.581.00.495-.053-7.098.091.00-.543.284-2.741.661.00E*Pre*T – C*Pre*T19.0-.620-10.548.31.00-.020.020-1.441.401.00-1.17.304-8.766.421.001.43-.278-.7743.631.00E*Post*T – C*Post*T-11.4.399-40.818.01.00-.267.156-1.691.151.00-.684.194-8.276.911.00-1.52.405-3.72.6841.00E*Pre*T – E*Post*T17.1-.496-12.346.51.00-.148.092-1.571.271.00-.127.037-7.727.461.002.81-.617.6165.021.00C*Pre*T – C*Post*T-13.2.558-42.616.21.00-.396.344-1.821.021.00.360-.093-7.237.951.00-.129.029-2.332.071.00E*Pre*C – C*Pre*C9.59-.495-19.839.01.00.739-.507-.6762.161.00-.939.088-8.536.651.00-.401-.184-1.802.601.00E*Post*C – C*Post*C-6.17.239-35.523.31.00-.214.088-1.641.201.00.029-.011-7.567.621.00-.713.315-2.911.491.00E*Pre*C – E*Post*C6.47-.261-23.035.81.00.515-.214-.8961.941.00-.690.067-8.286.901.00.255-.108-1.952.451.00C*Pre*C – C*Post*C-9.29.449-38.720.11.00-.438.281-1.86.9761.00.278-.033-7.317.871.00-.860.407-3.061.341.00E*Pre*P – C*Pre*P8.22-.415-21.237.61.001.07-.829-.3362.501.001.62-.121-5.979.211.00.734-.314-1.472.931.00E*Post*P – C*Post*P-4.12.181-33.525.31.00.015-.004-1.411.431.00.899-.081-6.698.491.00-.826.359-3.031.371.00E*Pre*P – E*Post*P7.24-.304-22.236.61.00.546-.246-.8661.971.00-.346.021-7.947.241.00.920-.384-1.283.121.00C*Pre*P – C*Post*P-5.10.275-34.524.31.00-.511.391-1.94.8961.00-1.07.100-8.666.521.00-.640.287-2.841.561.00E*Pre*O – C*Pre*O15.4-.756-14.044.81.001.62-1.10.1943.031.002.18-.144-5.419.771.00.871-.275-1.333.071.00E*Post*O – C*Post*O1.80-.079-27.631.21.00.912-.386-.4962.341.001.09-.111-6.508.681.00.292-.101-1.912.491.00E*Pre*O – E*Post*O6.13-.222-23.335.51.00.058-.02-1.371.471.001.57-.095-6.029.161.00.679-.192-1.522.881.00C*Pre*O – C*Post*O-7.50.557-36.921.91.00-.646.525-2.06.7761.00.487-.045-7.118.071.00.101-.042-2.102.301.00Abbreviations: E Experimental group, C Control group, Pre Recording data at 3 months (pre-intervention), Post Recording data at 1 week (post-intervention), 95%CI LB 95% Confidence interval lower bound, 95%CI UB 95% Confidence interval upper bound, d Effect size (dCohen), Diff Mean difference, PF Pre-frontal area, F Frontal area, T Temporal area, C Central area, P Parietal area, O Occipital area* Significance level seat at p < 0.05 Generalized linear mixed model (GLMM) of EEG in group, time, and brain area Abbreviations: E Experimental group, C Control group, Pre Recording data at 3 months (pre-intervention), Post Recording data at 1 week (post-intervention), 95%CI LB 95% Confidence interval lower bound, 95%CI UB 95% Confidence interval upper bound, d Effect size (dCohen), Diff Mean difference, PF Pre-frontal area, F Frontal area, T Temporal area, C Central area, P Parietal area, O Occipital area * Significance level seat at p < 0.05 Estimated marginal means and 95% confidence interval of PSD for EEG theta power. Abbreviations: Exp = Experimental group, Con = Control group, d = Effect size (dCohen). * Significance level set at p < .05 Estimated marginal means and 95% confidence interval of PSD for EEG beta power. Abbreviations: Exp = Experimental group, Con = Control group, d = Effect size (dCohen). * Significance level set at p < .05 Grand average EEG topographic maps of PSD (µ) in the theta and beta power for the experimental and control groups Subscales of the ACSI-28 questionnaire data In the current sample (N = 26), pre-intervention alphas were exceptionally high (α = 1.00), reflecting highly restricted variability in baseline responses among elite athletes. Post-intervention alphas ranged from 0.581 to 0.840 across subscales, with good reliability for the total scale (α = 0.826) as shown in Table 2. The relatively lower reliability of coachability (α = 0.432) is acknowledged as a limitation Table 2Internal consistency (Cronbach’s alpha) for ACSI-28 at pre- and post-interventionACSI-28 subscalePre-intervention αPost-intervention αCoping with adversity1.000.840Coachability1.000.432Concentration1.000.583Confidence and achievement motivation1.000.581Goal setting/Mental preparation1.000.689Peaking under pressure1.000.835Freedom from worry1.000.602Total ACSI-28 subscale 1.00 0.826 Abbreviations: ACSI-28 Athletic Coping Skills Inventory-28 Internal consistency (Cronbach’s alpha) for ACSI-28 at pre- and post-intervention Abbreviations: ACSI-28 Athletic Coping Skills Inventory-28 As presented in Table 3, the repeated measures ANOVA revealed significant group-by-time interaction effects across several subscales. Significant differences between the experimental and control groups were observed in coping with adversity (p = .018, = 0.212), concentration (p = .027, = 0.188), confidence and achievement motivation (p = .039, = 0.166), goal setting/mental preparation (p = .009, = 0.251), and peaking under pressure (p = .043, = 0.160). Post hoc analysis conducted at the post-intervention time point (one week prior to the competition) demonstrated that the experimental group achieved significantly higher scores than the control group in coping with adversity (p < .001, mean difference = 3.69, d = -2.03, 95% CI: 2.40 to 4.98), concentration (p < .001, mean difference = 3.00, d = -1.86, 95% CI: 1.76 to 4.24), confidence and achievement motivation (p = .006, mean difference = 2.92, d = -1.50, 95% CI = 1.40 to 4.44), goal setting/mental preparation (p = .003, mean difference = 3.08, d = -1.56, 95% CI: 1.57 to 4.59), and peaking under pressure (p < .001, mean difference = 3.54, d = -2.00, 95% CI: 2.17 to 4.89). The estimated marginal means of the ACSI-28 data, including the mean and 95% confidence intervals (95% CI), are visually presented in Fig. 5 Table 3Repeated measures ANOVA of ACSI-28 subscales in group and timeSubscalesGroup*TimePost Hoc Tests Mean square p Comparisons - Group*Time Diff d 95%CI p bonferroni LB UB Coping with adversity 15.1 0.212 0.018* E*Pre – C*Pre1.54− 0.699− 0.1543.230.526E*Post – C*Post 3.69 -2.03 2.40 4.98 < 0.001* E*Pre – E*Post -2.54 1.29 -4.04 -1.04 0.002* C*Pre – C*Post− 0.3850.215-1.891.111.00Coachability0.0190.0010.908E*Pre – C*Pre2.15-1.140.7023.620.032E*Post – C*Post2.23-1.511.093.370.005E*Pre – E*Post− 0.3080.172-1.611.011.00C*Pre – C*Post− 0.2310.139-1.541.081.00Concentration 16.2 0.188 0.027* E*Pre – C*Pre0.769− 0.378− 0.7962.341.00E*Post – C*Post 3.00 -1.86 1.76 4.24 < 0.001* E*Pre – E*Post -2.54 1.20 -3.95 -1.13 0.005* C*Pre – C*Post− 0.3080.208-1.721.101.00Confidence and achievement motivation 14.0 0.166 0.039* E*Pre – C*Pre0.846− 0.506− 0.4422.141.00E*Post – C*Post 2.92 -1.50 1.40 4.44 0.006* E*Pre – E*Post -2.31 1.60 -3.71 − 0.889 0.013* C*Pre – C*Post− 0.2310.107-1.641.181.00Goal setting/Mental preparation 20.9 0.251 0.009* E*Pre – C*Pre0.538− 0.287− 0.9071.991.00E*Post – C*Post 3.08 -1.56 1.57 4.59 0.003* E*Pre – E*Post -2.39 1.32 -3.87 − 0.912 0.006* C*Pre – C*Post0.154− 0.074-1.331.631.00Peaking under pressure 13.0 0.160 0.043* E*Pre – C*Pre1.54− 0.627− 0.3453.420.738E*Post – C*Post 3.54 -2.00 2.17 4.89 < 0.001* E*Pre – E*Post -2.46 1.02 -4.10 − 0.817 0.006* C*Pre – C*Post− 0.4620.256-2.111.171.00Freedom from worry4.920.0680.199E*Pre – C*Pre− 0.3850.305-1.370.5941.00E*Post – C*Post0.846− 0.369− 0.9122.591.00E*Pre – E*Post-2.461.21-3.88-1.040.006C*Pre – C*Post-1.230.750-2.650.1910.445Abbreviations: E Experimental group, C Control group, Pre Recording data at 3 months (pre-intervention), Post Recording data at 1 week (post-intervention), 95%CI LB 95% confidence interval lower bound, 95%CI UB 95% confidence interval upper bound, d Effect size (dCohen), Diff Mean difference* Significance level seat at p < .05 Repeated measures ANOVA of ACSI-28 subscales in group and time Abbreviations: E Experimental group, C Control group, Pre Recording data at 3 months (pre-intervention), Post Recording data at 1 week (post-intervention), 95%CI LB 95% confidence interval lower bound, 95%CI UB 95% confidence interval upper bound, d Effect size (dCohen), Diff Mean difference * Significance level seat at p < .05 Estimated marginal means and 95% confidence interval of ACSI-28 subscales. Abbreviations: Exp = Experimental group, Con = Control group, d = Effect size (dCohen). * Significance level set at p < 0.05 Correlation between EEG power spectral density and ACSI-28 The correlation analyses investigated the association between the EEG power spectral density across frequency bands and the ACSI-28 subscale scores in both the experimental and control groups. In the experimental group, a significant negative correlation was observed only within the delta frequency band. Delta power in the prefrontal area was significantly associated with goal setting/mental preparation (Spearman’s rho = − 0.702, FDR-adjusted p = .042). Significant correlations were also found in the control group, spanning delta and theta frequency bands. Specifically, delta power in three areas comprising prefrontal (Spearman’s rho = − 0.851, FDR-adjusted p = .006), frontal (Spearman’s rho = − 0.669, FDR-adjusted p = .024), and central (Spearman’s rho = − 0.700, FDR-adjusted p = .024) were significantly negatively associated with coping with adversity. Furthermore, theta activity in the frontal area showed a significant positive correlation with confidence and achievement motivation (Spearman’s rho = 0.718, FDR-adjusted p = .036). The magnitude of these significant correlations ranged from moderate to high (0.669 to 0.851) in both the experimental and control groups. Discussion The present study examined the effects of a combined breathing control and self-talk intervention on elite national-level athletes. We specifically examined changes in self-assessment of coping skills and electroencephalogram (EEG) by comparing an experimental group to a control group. Our findings suggest changes in both neural activity and self-assessed psychological skills following the combined intervention, indicating that this combined intervention may influence neurophysiological markers related to processes such as anxiety regulation and attention, as well as self-reported coping skills. Consistent with our hypothesis, the combined intervention was associated with changes in EEG activity. We observed an increase in theta power in the prefrontal cortex and a decrease in beta power in the frontal cortex when compared to the control group. These neural findings occurred alongside potential improvements in self-assessed psychological skills. However, the relationship between these neural and subjective changes remains unclear. Specifically, the experimental group showed an increase in several ACSI-28 subscales, including coping with adversity, concentration, confidence and achievement motivation, goal setting/mental preparation and peaking under pressure. We interpreted our findings based on the power spectral density (PSD) values, which are typically expressed in microvolts squared (), representing the signal power contained within each respective frequency band [52]. The observed magnitude of these differences falls within the neurophysiologically meaningful range reported in previous EEG studies examining cognitive and psychophysiological training in athletes [53, 54], suggesting that the measured changes may reflect a functionally relevant modulation of cortical activity [55]. However, EEG measures are inherently variable and may be influenced by individual and contextual factors. The observed decrease in beta power in the frontal cortex in our experimental group may be interpreted as being consistent with changes in attention and sensory processing, a finding that aligns with previous research [56]. Beta power has been associated with neurocognitive functions and network dynamics, including attentional orienting [57], cognitive effort [58], and working memory [59]. Therefore, the changes we observed in beta power could correspond to alterations in the activity of cognitive networks that operate across a variety of tasks [32]. Furthermore, the observed decrease in beta power may be associated with reduced arousal or anxiety-related processes [60], which are often linked to the relaxation effects of breathing interventions [61]. However, such interpretations should be made cautiously, given the indirect nature of resting-state EEG measures. Furthermore, the combined intervention produced an increase in prefrontal theta power in the experimental group relative to the control group. Theta oscillations in the prefrontal cortex have been linked to several processes relevant to anxiety and cognitive control, including threat detection, conflict monitoring, and top-down emotion regulation [62–66]. The literature on theta-anxiety relationships is not uniform; while some studies report that increases in theta power emerge during anxious or uncertain states [64], whereas others associate increased theta power with a sense of safety [65]. These divergent findings suggest that prefrontal theta oscillations are context-dependent and likely reflect a combination of vigilance, regulatory engagement, and cognitive control processes. In our study, the observed theta power increase may be consistent with increased engagement of regulatory processes, rather than necessarily reflecting heightened anxiety. This interpretation is supported by the fact that theta power is also associated with control and response inhibition [66], functions that are typically engaged during regulatory strategies such as controlled breathing control and self-talk. When considered alongside the observed reduction in frontal beta power, the overall pattern is more consistent with improved regulatory efficiency and attentional control rather than elevated threat processing. Thus, while theta power changes alone cannot be interpreted as a direct marker of reduced or increased anxiety, the combined pattern of increased theta and decreased beta power is consistent with a pattern that may reflect more efficient top-down regulatory engagement. Breathing-related entrainment is known to occur in cortical neural networks, which are critical for psychological factors including emotion, cognition and memory [67, 68]. Interventions involving breathing have been shown to produce consistent anti-arousal changes in affective states [69] and to reduce feelings of stress and anxiety during exercise [70]. Our findings are consistent with the previous study by Cheng et al., which demonstrated that 0.1 Hz breathing can alter EEG [71]. Specifically, it can increase frontal theta power, and this change is interpreted as being associated with enhanced attentional engagement [72]. Additionally, self-talk is a well-established technique that enhances top-down control, a process synonymous with self-control. It has been shown to improve awareness and problem-solving skills, and increased resilience [11, 73]. These benefits are also reflected in applied settings, such as improved sport performance, and also enhance cognitive functions [74]. Our findings suggest that the intentional self-talk may contribute to changes in neural patterns that are often associated with more regulated or focused mental states, rather than directly indicating a transition between specific psychological states. This is evidenced by the balanced mix of beta and theta power observed in our EEG data. These changes in neural activity may be related to the observed improvements in psychological coping skills following the combined breathing control and self-talk intervention. However, this relationship remains inferential and is not directly demonstrated in the present design. Regarding the Athletic Coping Skills Inventory-28 (ACSI-28) questionnaire results, we found a post-intervention increase in five subscales for the experimental group compared to the control group: coping with adversity, concentration, confidence and achievement motivation, goal setting/mental preparation, and peaking under pressure. These findings suggest that the 4-week combined intervention provides preliminary evidence that the intervention may contribute to improvements in these psychological skills. The observed increase in the goal setting/mental preparation subscale may be linked to the self-talk component of the intervention. The ‘relax and calm’ self-talk used in the study may serve as a form of mental rehearsal, allowing athletes to internalize and focus the situation [75]. This type of preparatory self-talk could help athletes to redirect their attention away from negative or anxiety-provoking thoughts. The mechanism by which these psychological scores were enhanced may be due to the influence of self-talk on emotional and cognitive control [76], which is further supported by the observed increase in the concentration and coping with adversity subscale scores. Furthermore, the breathing control component of the intervention may also contribute to the increased concentration subscale score, as controlled breathing is a well-established technique used to improve focus and maintain attention in athletes. This aligns with the proposed mechanism that breathing control can alter cognitive and emotional states [69] and may contribute to improved concentration. The increased scores in the confidence and achievement motivation and peaking under pressure subscale may be an indirect effect from the combined intervention. This might be mediated by changes in psychological processes such as anxiety regulation. However, EEG findings should be interpreted as indirect neural correlates rather than direct evidence of reduced anxiety. When athletes are better able to manage their anxiety, they typically experience an increase in confidence and a greater ability to perform effectively during high-pressure situations [17]. Overall, these findings collectively suggest that combined intervention may contribute to improvements in athletes’ coping skills. However, because participants knew the general nature of the sessions they received, expectancy effects or demand characteristics may have influenced self-reported measures such as ACSI-28. These factors are inherent challenges in psychological intervention research and should be considered when interpreting the findings. The examination of associations between EEG activity and psychological skills revealed that only a selective set of correlations remained significant after applying false discovery rate (FDR) correction, indicating that the most robust EEG-ACSI-28 relationships were focused within specific frequency bands and frontal brain areas. In the experimental group, the only significant association was a negative relationship between prefrontal delta power and goal setting/mental preparation. Reduced delta activity in frontal areas is commonly interpreted as reflecting greater cortical activation and regulatory engagement [77, 78]. Thus, athletes who demonstrated strong cognitive preparation skills also exhibited a profile characterized by reduced delta activity (an activated prefrontal profile) during rest. This finding aligns with the intervention’s aim to enhance psychological readiness and self-regulatory control. A broader pattern emerged in the control group, where significant negative correlations linked delta power across prefrontal, frontal, and central areas with coping with adversity. These results suggest that athletes with superior adversity-coping abilities tend to show reduced resting state delta power, pointing to a more aroused or engaged cortical state. This aligns with recent EEG research linking low-frequency power suppression to effective emotion regulation and stress resilience [79]. A significant positive association between frontal theta power and confidence/achievement motivation was also observed. Frontal theta power is repeatedly linked to motivational drive, cognitive effort and allocation of attentional resources [66, 80]. Collectively, these findings highlight the potential role of frontal delta and theta activity in profiling psychological skills related to planning, coping, and motivation. The selective nature of the results after correction underscore the importance of interpreting EEG-behavior correlations cautiously and reinforces the idea that only the strongest and most theoretically meaningful neural-psychological associations should be emphasized [81]. Notably, the limited number of significant associations after multiple-comparison correction further suggest that neural and self-reported measures may operate partially independently. As such, any apparent convergence between EEG and psychological findings in the present study should be considered inferential rather than directly demonstrated. Despite some incongruence between neural and self-assessment findings, our results suggest that the 4-week combined intervention was associated with improvements in athletes’ psychological skills at both neural and subjective levels. Specifically, EEG findings suggest potential changes in neural processes. However, these neural patterns do not directly correspond to self-reported psychological measures and may reflect distinct aspects of adaptation. These findings should be interpreted cautiously, particularly when linking neural activity to psychological processes. Overall, these findings indicate that the combined intervention may represent a promising approach for enhancing athletes’ psychological skills, with potential benefits for anxiety management and coping abilities in high-pressure situations. Importantly, given the use of resting-state EEG and the absence of objective behavioral performance measures, the present study findings should be interpreted as reflecting indirect neural correlates rather than specific markers of psychological states. The relationship between EEG frequency bands and constructs such as anxiety and attention remains complex and context-dependent, and therefore requires cautious interpretation. Given the relatively small sample size and study design, the present findings should be considered preliminary and exploratory in nature. Our study has several limitations related to the research design and logistical constraints. First, because we were unable to include single-component control groups (breathing-only or self-talk-only), the design does not allow the unique contribution of each technique to be isolated. Therefore, the present study design does not allow causal inference regarding the independent contributions of breathing control and self-talk components of the combined intervention, and observed effects should be interpreted as reflecting the combined intervention rather than specific mechanisms. Furthermore, while control participants received only unstructured, on-request consultations, these interactions may still have produced non-specific benefits or placebo-like effects, as attention-control activities have been shown to influence outcomes in previous behavioral research [82], which limits the ability to attribute observed effects specifically to the intervention. However, because participants knew the general nature of the sessions they received, expectancy effects or demand characteristics may have influenced self-reported outcomes. Additionally, participants were not blinded to group allocation, which may have introduced expectancy or performance-related biases. A second limitation is the restricted statistical power due to small sample size (N = 26). The relatively large effect sizes should be interpreted with caution, as they may reflect sample-specific characteristics or potential overestimation due to small sample size. Replication in larger samples is warranted. Although a priori calculations indicated that a larger sample would have been ideal, recruitment was constrained by the logistical demands of working with elite athletes during their competition preparation period. As a result, small-to-moderate effects may not have been reliably detected; however, reporting observed effect sizes and confidence intervals provides useful preliminary evidence to guide future research. This constraint also affected ACSI-28 subscales which demonstrated low internal consistency, likely due to the small sample size and limited score variability in this elite national-level athlete cohort. Therefore, the subscale-level findings offer useful preliminary insight that can guide subsequent psychometric validation efforts in elite athlete cohorts. Third, data collection could not be conducted during actual competitions due to coaching and scheduling constraints, and no task or simulation replicating high-pressure competitive environments was used. While our resting-state EEG measurements reveal changes in neural markers associated with psychological states, these cannot be directly equated with improvements in objective performance ability. Confirming objective performance enhancement would require additional measures, such as task-based EEG, sport-specific motor performance tests, or induced pressure scenarios. Incorporating these conditions would have provided more ecologically valid insight into athletes’ neural and psychological response to stress and performance. Finally, generalizability is limited as the sample consisted exclusively of elite Thai national team athletes preparing for an international event. Therefore, the findings may not generalize to recreational athletes, developmental athletes, or competitors from other sports, age groups, or performance levels. Future studies should consider incorporating component-specific control groups, larger and more diverse samples, and ecologically valid performance simulations (e.g., virtual reality) to strengthen generalizability and interpretability. Conclusion This study suggested that a combined breathing control and self-talk intervention may be an effective approach for enhancing psychological coping skills in elite national-level athletes. EEG analysis showed increased theta power in the prefrontal cortex and decreased beta power in the frontal cortex in the experimental group relative to the control group, patterns that may be associated with neural processes related to attentional and stress regulation. In addition, ACSI-28 scores increased across several subscales, including coping with adversity, concentration, confidence and achievement motivation, goal setting/mental preparation, and peaking under pressure. Taken together, these findings suggest that the combined intervention may contribute to improvements in aspects of mental resilience and performance under stress in this athletic population. However, these conclusions should be interpreted within the context of the present sample, and broader applications beyond the domain of sports warrant further investigation in larger and more diverse populations. Supplementary Information Supplementary Material 1. Acknowledgements This work was supported by College of Sports Science and Technology, Mahidol University, Thailand. We thank Prof. Mark Willems (University of Chichester, UK) for comments on the manuscript. Abbreviations Athletic Coping Skills Inventory-28 Electroencephalogram Fast Fourier Transformation False Discovery Rate Generalized Linear Mixed Modelling Ground Electrode Power spectral density Authors’ contributions KT, AN, and EO conceptualized and designed the study, performed data curation, and recruited participants. Formal analysis was conducted by three authors. KT and AN drafted the original manuscript. All authors (KT, AN, and EO) reviewed and edited the manuscript, have read and approved the final version, and agree with the order of presentation of the authors. Funding None. Data availability The datasets generated and analyzed during this study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate All participants provided written informed consent prior to participation. This study was conducted following the approval of the Mahidol University Central Institutional Review Board. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets generated and analyzed during this study are available from the corresponding author on reasonable request.