The studies involving humans were approved by Ethics Committee of the Faculty of Physical Education and Sport, Charles University (UK FTVS), under reference number 248/2023. The studies were conducted in accordance with the local legislation and institutional requirements. The studies were conducted in accordance with the Helsinki Declaration requirements. The participants provided their written informed consent to participate in this study.

This exploratory cross-sectional study was conducted in the VR laboratory at SENSE Arena, Faculty of Physical Education and Sport, Charles University. The aim was to assess cognitive functions in professional and semi-professional BJJ athletes under standardized conditions. Given the small sample (n = 10), findings are indicative rather than generalizable.

The determination of the sample size was guided by an a priori power analysis performed using GPower 3.1 software. Assuming a medium effect size (Cohen's d = 0.5) for differences in cognitive performance between professional and semi-professional athletes, a two-tailed Mann–Whitney U test with an alpha level of 0.05 and statistical power of 0.80 would require a minimum of 21 participants per group. An a priori power analysis indicated that at least 21 participants per group would be required to detect medium effects with adequate statistical power (α = .05, 1 − β = .80). Because recruiting professional BJJ athletes matching the inclusion criteria was highly constrained, the study proceeded with a convenience sample of 5 athletes per group. As a result, the study is severely underpowered, and any statistically significant findings must be viewed as unstable and potentially sample-specific.

Ten male athletes aged 20–35 (x˜ = 26.20 ± 5.03) and with average training experience 8.20 ± 3.19 were deliberately divided into two groups by quantifiable competitive criteria. Professionals (n = 5; average age was 27.20 ± 5.63 years): ≥10 years of BJJ experience (x˜ = 11.00 ± 1.41), brown (n = 3)/black (n = 2) belt, recent international medal (IBJJF). Semi-professionals (n = 5; 25.20 ± 4.76 years): 4–7 years of experience (x˜ = 5.40 ± 1.14), blue (n = 3)/purple (n = 2) belt, regional/national-level activity. All were right-handed, with comparable body types and autonomous motor skill execution. Informed consent was obtained. Information on chronological age, training experience, activity level, and competitive history was gathered using a diagnostic survey method, implemented through direct interviews with the athletes and their coaching staff. The study was conducted during the preparatory phase of the training cycle. The inclusion criteria for the study were as follows: no history of serious injuries, a minimum of four years of training experience, right-handedness and active participation in competitions. The exclusion criteria included violation of any of the above conditions, current injuries or medical conditions that could affect participation in the study, lack of informed consent, absence of competitive experience, previous VR training experience and the use of performance-enhancing substances.

All participants were tested individually in a controlled laboratory environment between 14:00 and 16:00 to minimize diurnal variation. Task order was fixed across participants due to software constraints, and all athletes were instructed to refrain from strenuous training for 24 h prior to testing to minimize fatigue-related variability. No participant reported prior experience with VR training.

Participants were familiarized with the VR interface before completing a battery of standardized cognitive tests via SenseArena software (SenseArena s.r.o., Prague, Czech Republic) using the HTC Vive Pro 2 headset. Although the SenseArena tests were not designed specifically for BJJ, they were selected because they target general cognitive components—such as stimulus discrimination, divided attention, and rapid decision-making—that are theoretically relevant to open-skill combat scenarios. Importantly, no peer-reviewed data are available regarding the reliability, validity, or psychometric properties of these tasks in combat sports or BJJ populations, and the platform does not provide grappling-specific stimuli (e.g., positional cues, opponent pressure, movement patterns). As a result, the ecological validity and measurement precision of these tests for BJJ remain limited. Therefore, the present results should be interpreted strictly as indicators of general cognitive tendencies rather than sport-specific cognitive performance, and any extrapolation to real-world BJJ contexts should be considered preliminary.

All participants were tested individually in a controlled laboratory environment between 14:00 and 16:00 to minimize diurnal variation. Task order was fixed across participants due to software constraints, and all athletes were instructed to refrain from strenuous training for 24 h prior to testing to minimize fatigue-related variability. No participant reported prior experience with VR training.

Participants were familiarized with the VR interface before completing a battery of standardized cognitive tests via SenseArena software (SenseArena s.r.o., Prague, Czech Republic) using the HTC Vive Pro 2 headset. Although the SenseArena tests were not designed specifically for BJJ, they were selected because they target general cognitive components—such as stimulus discrimination, divided attention, and rapid decision-making—that are theoretically relevant to open-skill combat scenarios. Importantly, no peer-reviewed data are available regarding the reliability, validity, or psychometric properties of these tasks in combat sports or BJJ populations, and the platform does not provide grappling-specific stimuli (e.g., positional cues, opponent pressure, movement patterns). As a result, the ecological validity and measurement precision of these tests for BJJ remain limited. Therefore, the present results should be interpreted strictly as indicators of general cognitive tendencies rather than sport-specific cognitive performance, and any extrapolation to real-world BJJ contexts should be considered preliminary.

All tests originated from the SenseArena platform, which has internal validation but lacks peer-reviewed BJJ-specific reliability data. This remains a key limitation.

Testing employed HTC Vive Pro 2 headsets (HTC Corporation, Taoyuan City, Tajwan) and SenseArena software (Sense Arena s.r.o., Prague, Czech Republic). Although the VR tasks include general perceptual-motor elements common to combat sports, they do not simulate grappling-specific interactions such as positional transitions, grip sequences, or opponent pressure. This lack of contextual specificity reduces the ecological validity of the assessment and limits the degree to which performance in these VR tasks can be extrapolated to real BJJ decision-making or tactical behavior. Future protocols should integrate sport-specific stimuli to improve ecological validity.

The overall study procedure is summarized in Figure 1. This flow diagram provides a structured overview of all methodological steps, from participant screening and group allocation through virtual reality familiarization, cognitive testing, and statistical analysis. Presenting these stages in a visual format enhances clarity and allows readers to follow the study design more intuitively.

Descriptive statistics were calculated for all variables. Group comparisons were conducted using non-parametric Mann–Whitney U tests, as the small sample size (n = 5 per group) precluded assumptions of normality. For each comparison, we report the U statistic, standardized Z value, exact p-value, and corresponding effect size (r), in line with recommended practice for small-sample non-parametric testing.

To enable exploratory comparisons across heterogeneous VR task metrics, raw values were also standardized using Z-scores computed across the entire sample (N = 10). These standardized values serve purely descriptive purposes and should be interpreted with caution, as Z-scores are highly sensitive to outliers and may exaggerate apparent between-group differences.

Cohen's d was additionally computed for descriptive effect size estimation, but given the small sample and non-parametric distributions, r-values from the Mann–Whitney tests are considered more informative. Standard deviations (SDs) were used to characterize within-group variability.

Because of the small sample size, raw medians and interquartile ranges (IQRs) for all cognitive variables are included in Supplementary Table S1 to allow transparent evaluation of group distributions. All analyses were conducted using SPSS (IBM SPSS Statistics, Version 29.0.2.0) and Microsoft Excel (Version 2511). The statistical significance level was set at p < 0.05, but given the exploratory nature of the study, p-values should not be interpreted as confirmatory evidence.

Given the exploratory nature of this study, no correction for multiple comparisons was applied. As a result, the probability of Type I error is increased, and statistically significant findings should be interpreted with caution, particularly in the context of the very small sample size.

This test evaluated Reaction Time and Average Time to Correct Decision.

Reaction Time (RT): Professionals had a longer mean MT (M = 2,801.2 ms; SD = 1,690.24) than semi-professionals (M = 1,691.48 ms; SD = 253.96), with a statistically significant difference (p = 0.047; d = 0.92). Given the extremely small sample size (n = 5 per group), these observed differences—including large effect sizes—must be interpreted with caution, as they may reflect sample-specific variability rather than systematic expertise-related effects. In addition, this comparison is statistically underpowered and no correction for multiple comparisons was applied, increasing the likelihood of a Type I error. Importantly, no inference can be made regarding underlying cognitive strategies (e.g., delayed responding or “strategic processing”), as the study did not include any measures capable of assessing such mechanisms. Alternative explanations such as individual variability or age differences remain equally plausible.

This test included Caught Balls (%) and Release Time (ms).

Caught Balls (%): Professionals showed a higher success rate (M = 49.2%; SD = 6.42) than semi-professionals (M = 38.4%; SD = 54.86), with statistical significance (p = 0.0432) but only a small effect size (d = 0.28). However, the extreme variability in the semi-professional group, combined with the very small sample size (n = 5 per group), indicates substantial measurement instability. In addition, this comparison is statistically underpowered and no correction for multiple comparisons was applied, increasing the risk of a Type I error. As such, the apparent group difference may be partly or entirely attributable to noise, and the effect size is likely inflated. Given also that this VR task has limited ecological validity for BJJ, the practical meaningfulness of this result remains highly uncertain.

Four variables were analyzed: Covered Game Field (%), Median Motor Time (ms), Median Reaction Time (ms), and Correct Attempts (%). Although several effect sizes appeared moderate to large, their interpretability is limited by the high inter-individual variability and the extremely small sample size, both of which can artificially inflate standardized effect estimates. However, all interpretations must be considered preliminary given the limited sample size and potential influence of uncontrolled variables such as fatigue, prior exposure to VR, or inter-individual learning rates

Covered Game Field (%): Professionals covered a marginally larger field area (M = 29.8%) than semi-professionals (M = 28.69%), with a small effect (Cohen's d = 0.21). This minimal difference is unlikely to reflect a systematic group effect and may instead be attributable to inter-individual variability or measurement noise. Because the VR task does not assess constructs such as spatial awareness, no psychological interpretation can be drawn from this outcome.

Three variables were examined: Median Reaction Time (ms), Successful Tasks (%), and Median Motor Time (ms). None of the group differences reached statistical significance (p > 0.05), though effect sizes suggested minor trends.

Median Reaction Time (ms): Semi-professional fighters demonstrated slightly faster median RT (M = 396.0 ms) compared to professionals (M = 418.0 ms), although this difference occurred alongside notably higher variability in the semi-professional group (SD = 39.3 vs. 26.0). The effect size was moderate (Cohen's d = 0.66), but given the very small sample size (n = 5 per group), this value should be regarded as descriptive rather than indicative of a stable performance trend. The Mann–Whitney test (p = 0.347) did not yield statistical significance, and with such limited statistical power, non-significant findings cannot be interpreted as evidence for or against group differences. Therefore, this numerical pattern should be considered inconclusive rather than reflective of genuine differences in reaction capabilities.

This test measured Efficiency, Concentration, and Reaction Time. While only concentration differences were statistically significant, effect sizes suggested practical trends. Due to the small sample and lack of controls (e.g., cognitive baseline, psychological state), findings remain tentative.

Efficiency (%): Although professionals obtained higher efficiency scores (91.20% vs. 79.20%), this difference was not statistically significant (p = 0.236), and the very large effect size (Cohen's d = 1.22) is likely inflated due to high within-group variability and the extremely small sample size (n = 5 per group). As such, this numerical difference should be interpreted as a descriptive pattern rather than as evidence of superior task execution or performance efficiency. Moreover, the VR task does not assess constructs such as execution quality or cognitive efficiency, and non-significant findings in an underpowered design cannot be taken as evidence for or against group differences.

Following the presentation of all outcome-specific results, the complete set of statistical outputs for every test battery included in this study is summarized below. Table 2 reports the U statistic, Z-value (calculated without continuity correction), exact p-values, asymptotic p-values, and both effect size indices (Cohen's r and Cohen's d) for all measured outcomes.

An overview of all outcomes and their corresponding test groups is presented in Table 2.

Table 3 presents an exploratory comparison of all five test batteries using standardized Z-scores. Given the small sample size (N = 10) and variability in several tasks, these Z-scores should be interpreted cautiously, as their magnitude may reflect scaling artefacts rather than stable performance differences. A statistically significant difference emerged only in the Multiple Objects Test (p = 0.036), where professionals obtained higher scores in this sample. Although professionals obtained higher composite Z-scores, this pattern cannot be interpreted as evidence of superior multitasking or decision-making. Z-scores calculated on such a small sample (N = 10) are highly sensitive to outliers and variability, and may inflate apparent between-group differences. With such a small sample (n = 10), Z-scores are highly sensitive to outliers and measurement variability, and the differences may largely reflect scaling artefacts rather than true performance distinctions. The non-sport-specific nature of the VR tasks further limits the interpretability of these standardized comparisons.

Although other tests did not reach statistical significance, several trends are notable. Easy Moving Target: semi-professionals scored slightly higher (Z = 0.325 vs. 0.246), possibly due to faster—but less controlled—responses. However, high variability limits interpretability. Reflex with Action: both groups achieved similar results, suggesting comparable performance in integrated sensorimotor tasks. Minor differences may reflect noise rather than systematic effects. Synchro Reflex: high and closely matched scores (professionals Z = 0.89; semi-professionals Z = 0.812) suggest this test may reflect individual motor coordination predispositions rather than training level. And the Coach's Drop: professionals scored higher (Z = 0.682 vs. 0.579), though without statistical significance. This mirrors the numerical pattern observed earlier in the concentration variable, but cannot be interpreted as evidence of attentional differences.

Summary: While only one test showed statistical significance, professionals generally obtained numerically higher scores in several cognitive metrics, particularly in attention and decision-making. Nevertheless, due to small sample size, homogeneity, and the non-BJJ-specific nature of the tasks, these findings should be interpreted as exploratory. Future research with larger, stratified samples and sport-specific stimuli is needed to validate these preliminary trends.

This study is limited by its small sample size, which constrained statistical power relative to an a priori calculation in G*Power 3.1 indicating 21 participants per group. The use of only five participants per group underscores the exploratory nature of the findings and highlights the need for replication in larger samples. Furthermore, the scarcity of professional male BJJ athletes in the Czech Republic restricted recruitment and limits generalizability. Furthermore, the validity of the VR tests for BJJ remains unestablished. While the tasks capture generic cognitive domains, their lack of sport-specific cues means that high task performance may not correspond to grappling decision-making in realistic contexts. Therefore, the present findings should not be used to infer sport-specific cognitive advantages without further validation.

Because Z-scores were calculated across all participants in a very small sample, their stability is limited and they may exaggerate relative differences between groups. Therefore, the standardized scores presented here should be interpreted purely as exploratory visualizations.

The results of this study provide preliminary insights that may inform future research and long-term methodological development in BJJ and other combat sports. However, given the exploratory design, small sample size, and limited ecological validity of the VR tasks, the present findings should not be used to guide athlete selection, talent identification, or individualized training prescriptions.

First, although professionals obtained numerically higher scores in some VR-based cognitive metrics, these differences were inconsistent across tasks and cannot be interpreted as cognitive advantages or performance determinants. At best, they highlight domains—such as reaction time and decision accuracy—that may warrant further investigation in more robust, sport-specific studies.

Second, while VR technology offers a controlled and low-risk environment for cognitive-motor assessment, the tasks used in this study do not simulate combat pressure or measure psychological constructs such as resilience, anticipation, or adaptive decision-making. Therefore, any application of VR to training design remains speculative and requires validation with larger samples and sport-specific task paradigms.

Third, although the exploratory numerical patterns observed in this study may encourage further investigation into how cognitive–motor performance relates to athlete development, the present findings do not provide evidence for talent identification, detection of stable cognitive characteristics, or prediction of competitive outcomes. Any such applications would require longitudinal research, validated sport-specific cognitive measures, and substantially larger samples to establish reliability and practical relevance.

Finally, the integration of cognitive and motor assessments represents a promising direction for interdisciplinary research. Future work combining VR-based performance metrics with validated psychophysiological measures may help clarify how cognitive processes interact with technical and tactical skill execution, but these approaches remain beyond the scope of the present study.