Improvements in executive functions by domain-specific cognitive training in youth elite soccer players

The study investigates whether executive functions (EFs)—inhibition, working memory, and cognitive flexibility—can be improved in youth elite soccer players through a domain-specific, sport-focused cognitive training (CT) program. EFs are critical for goal-directed behavior in dynamic environments like team sports. Prior literature shows mixed evidence for links between EFs and sports performance and questions the transfer of cognitive training from lab tasks to sport contexts (near vs far transfer). The authors aimed to test if an 8-week soccer-specific CT would enhance EF task performance compared to an active control group and hypothesized that the intervention would lead to improvements in inhibitory control, working memory, and cognitive flexibility in the intervention group relative to controls.

The paper reviews core EFs (inhibition, working memory, cognitive flexibility) and their importance in team sports such as soccer, framed by the unity-diversity model of EFs. Two approaches are outlined: expert-performance (domain-specific) and cognitive skills component (domain-generic). A debate exists on ecological validity of standard EF tasks for sports, with emerging sport-specific EF assessments. Evidence for cognitive training in athletes is mixed. Domain-generic CT often shows near-transfer to trained tasks but limited far-transfer to sport performance; commercial brain training products show inconsistent efficacy. Studies using 3D multiple object tracking (NeuroTracker) demonstrate improvements in MOT performance (near transfer) and mixed results on EFs and sport outcomes. Far transfer from CT to sports performance is contested, with age effects likely mediating EF-performance associations. Domain-specific CT targeting sport contexts may be more promising but is under-researched. Ceiling effects in high-performing youth athletes and limited ecological validity of lab EF tasks may attenuate observed benefits.

Design: Pre-post, two-group design with an intervention group (IG) and an active control group (CG). Allocation was based on organizational reasons (non-random). Assessments occurred at pretest (first week of the intervention; October 2022) and posttest (eighth week; November 2022). Consent was obtained; ethics approved (HU-KSBF-EK_2022_0018). Participants: 31 male youth soccer academy players (highest league for age; age 13–15 years, M=14.30, SD=0.59); IG n=13 (boarding school players), CG n=18. Three players discontinued the intervention (left club or declined). Active control: matched-duration technical skills training. Cognitive Training Intervention: 8 weeks, 2 sessions/week, 40–50 min per session (2×20–25 min). Exercises adapted from Memmert (2019) and Nowak & Vestberg (2019) targeted soccer-specific scenarios to train inhibition, working memory, and cognitive flexibility. Details provided in supplemental material (https://osf.io/aucdf/?view_only=2ec940fb40554f6da7c3545c348a3336). EF Measurements: Conducted with Inquisit Lab 6 on a 17-inch screen and QWERTZ keyboard; tasks administered in randomized order between 10:00–16:00, approximately 1 hour before training; total testing ~30 min. - Inhibition: Flanker task (5 arrows; respond to central arrow using E/I keys). 4 practice trials, 72 test trials (24 incongruent, 48 congruent). Outcomes: response times (RT) for correct trials (congruent/incongruent), accuracy (%), Flanker effect (incongruent minus congruent RT). - Working memory: 3-back task with neutral images; indicate if current image matches the one 3 trials prior (A-key). 23 practice trials, 46 target trials. Outcomes: RT for correct target trials, accuracy including hit rate minus false alarm rate; missed target trials also noted. - Cognitive flexibility: Number-letter (alternating runs) task; 2×2 matrix with number-letter pairs moving clockwise. Top boxes: letter vowel/consonant (I/E keys); bottom boxes: number even/odd (I/E keys). Practice: 24 trials per condition; test: 64 target trials (32 switch, 32 no-switch). Outcomes: RT and accuracy for switch/non-switch trials; switch costs (RT and accuracy differences). Data Preparation: Flanker task—removed incorrect trials (pre 2.64%; post 3.23%), RTs <200 ms or >1750 ms (0%), and ±3 SD outliers (pre 1.34%; post 1.38%); one player excluded for incomplete dataset. 3-back—excluded missed targets (pre 36.43%; post 41.14%), responses to nontargets (false alarms; pre 17.88%; post 15.71%) from RT analyses; ±3 SD outliers (0%); two players excluded for incomplete datasets. Number-letter—removed incorrect trials (pre 15.82%; post 9.17%), applied additional filters (pre 7.43% and 0.42%; post 3.65% and 1.11%); no datasets excluded. Statistical Analysis: MANOVA/ANOVA approach robust to non-normality; modified Bartlett test used to check heteroscedasticity. Inhibition: 2×2 MANOVA on RT parameters; separate 2×2 ANOVAs for accuracy. Working memory: separate 2×2 ANOVAs for RT and accuracy. Cognitive flexibility: 2×2 MANOVA on RT and accuracy parameters, followed by Bonferroni-corrected post hoc tests (univariate ANOVA and t-tests). SPSS 28 used; alpha p<.05.

Baseline EF measures did not differ between groups. Inhibition (Flanker task): No significant main effects for group (F[3,31]=0.56, p=.460, ηp²=0.020), time (F[3,31]=0.53, p=.473, ηp²=0.019), or interaction (F[3,31]=0.43, p=.516, ηp²=0.015) for RT; accuracy also showed no significant effects (group F[3,31]=0.26, p=.616, ηp²=0.009; time F[3,31]=2.449, p=.080, ηp²=0.257; interaction F[3,31]=1.213, p=.280, ηp²=0.042). Working Memory (3-back): No significant main effects for RT (group F[3,31]=0.27, p=.607, ηp²=0.010; time F[3,31]=0.02, p=.872, ηp²=0.001; interaction F[3,31]=2.924, p=.099, ηp²=0.101) or accuracy (group F[3,31]=0.30, p=.589, ηp²=0.011; time F[3,31]=1.125, p=.299, ηp²=0.041; interaction F[3,31]=0.00, p=.992, ηp²=0.000). Cognitive Flexibility (Number-letter): Significant main effect of time on RT (F[3,31]=8.49, p=.007, ηp²=0.226) with improvements in both groups; confirmed for no-switch RT (F[29]=631.14, p<.001, ηp²=0.956) and switch RT (F[29]=8.49, p=.007, ηp²=0.226), but not for switch cost RT (F[29]=1.04, p=.316, ηp²=0.035). Accuracy showed a significant time effect (F[3,31]=9.91, p=.004, ηp²=0.255), confirmed for no-switch accuracy (F[29]=13.563, p<.001, ηp²=0.319) and switch accuracy (F[29]=7.818, p=.009, ηp²=0.212), but not for accuracy switch cost (F[29]=0.85, p=.363, ηp²=0.029). No significant group or time×group interactions were observed for cognitive flexibility. Overall, the 8-week soccer-specific CT did not produce superior EF improvements versus active control; cognitive flexibility improved over time in both groups, potentially reflecting general cognitive development or practice effects.

Contrary to the hypothesis, the domain-specific soccer CT did not improve EF task performance (inhibition, working memory, cognitive flexibility) beyond the active control. The observed gains in cognitive flexibility across both groups likely stem from general developmental changes or task practice effects rather than the intervention. The findings align with literature questioning broad (far) transfer from cognitive training to sport performance and suggesting that CT effects are often narrow and task-specific (near transfer). Potential reasons for the null effects include ceiling effects among highly skilled youth athletes, insufficient training duration/intensity, lack of precise alignment between training content and measured EF components, and limited ecological validity of laboratory EF tasks relative to real soccer contexts. Age is a critical factor in EF-performance relations, and high baseline cognitive function may reduce room for improvement. The results underscore the need for ecologically valid assessment tools and interventions closely matched to sport-specific cognitive demands, potentially leveraging virtual reality to enhance specificity and measurement sensitivity.

An 8-week, soccer-specific cognitive training program did not enhance executive functions in youth elite soccer players compared to an active control. While cognitive flexibility improved over time in both groups, no differential intervention effects were detected. The study highlights that CT is only one component within multifactorial sports performance and that transfer from CT to EF tasks or sport performance may be limited without high specificity, adequate duration, and ecologically valid assessment. Future research should develop and employ sport-specific, ecologically valid EF tasks, consider longer and more intensive training protocols, leverage technologies such as virtual reality for standardization and tracking, and account for individual differences and age-related development when designing CT programs.

- Small sample size and a homogeneous, high-performance cohort may limit variance and detectable effects. - Non-random group allocation (organizational reasons) may introduce bias. - Training quality and adherence within the CT sessions were not objectively monitored; improvements within CT were not tracked. - EF assessments used domain-generic stimuli and responses (keyboard-based lab tasks), reducing ecological validity and potentially limiting detection of sport-relevant changes. - Testing occurred under neutral resting conditions, which may not reflect sport-specific EF demands under psychological and physiological stress. - The 8-week duration may be insufficient to elicit measurable EF changes in highly skilled youth athletes, who may exhibit ceiling effects.