Neural mechanisms underlying the effects of open-skill and closed-skill sports on executive function

Abstract

Executive function is a higher-order cognitive capacity that integrates planning, organization, impulse control, and emotional regulation, thereby exerting a direct influence on decision-making and behavioral regulation in daily contexts. This study recruited open-skill athletes, closed-skill athletes, and non-athletes and employed event-related potentials (ERPs) to examine how different types of sport participation are associated with executive function. Participants completed the Stroop, task-switching, and N-back tasks to assess three core components of executive function: inhibitory control, cognitive flexibility, and working memory, thereby elucidating the underlying neural mechanisms. The behavioral results showed that open-skill athletes responded faster than non-athletes in both the task-switching and N-back tasks, while closed-skill athletes responded faster than non-athletes only in the N-back task. EEG results revealed that open-skill athletes exhibited smaller N2 amplitudes than non-athletes in the task-switching task. They also elicited larger P3 amplitudes than non-athletes in the Stroop task, and larger P3 amplitudes than both closed-skill athletes and non-athletes in the task-switching task. In the N-back task, both types of athletes elicited larger P3 amplitudes than non-athletes. Overall, athletes from different sport types exhibited task-specific behavioral and neuroelectric profiles during executive-function processing. Open-skill athletes showed faster responses in the task-switching and working-memory tasks, whereas closed-skill athletes showed a more selective pattern of faster responses in the working-memory task.

Introduction

Executive function is the capacity to monitor and regulate thought and behavior to support goal-directed control in complex situations [1–4]. It is commonly described by three core components: inhibitory control (suppressing prepotent responses), cognitive flexibility (shifting between task demands), and working memory (maintaining and manipulating information) [1, 4–9]. Executive functions are strongly associated with academic achievement and mental health, and impairments in executive functioning have been implicated in a range of psychiatric conditions [10–13]. In recent years, growing evidence has indicated that physical exercise positively influences cognitive abilities, particularly executive function [14–16]. For example, Chang et al. found that an acute bout of moderate-intensity aerobic exercise significantly shortened reaction times in the Stroop task among young adults, indicating enhanced inhibitory control [17]. Gothe et al. reported that eight weeks of Hatha yoga practice led to significant improvements in executive function in healthy older adults, particularly in cognitive flexibility and working memory efficiency [18].

To further explore the effects of physical exercise on executive function, researchers have categorized sports into open-skill and closed-skill sports based on the predictability of the sporting environment [19, 20]. Open-skill sports are characterized by dynamic and unpredictable contexts, requiring athletes to continuously adapt their actions and make rapid decisions in response to changing external demands [15, 20], such as in soccer. In contrast, closed-skill sports are performed in relatively stable and predictable environments, where movements are executed in a pre-planned manner with an emphasis on precision and consistency [15, 20], as exemplified by the 100-meter sprint.

Accumulating evidence have demonstrated that open-skill sports exert a more pronounced positive effect on executive function than closed-skill sports [21–26]. For instance, De Waelle et al. reported that children who frequently participated in team sports (open-skill) outperformed those in self-paced sports (closed-skill) on tasks assessing inhibitory control, cognitive flexibility, and working memory, indicating that cognitively demanding sports better support executive function development in children [27]. Koch et al. measured the differences in executive function between elite open-skill and closed-skill athletes and recorded the duration of their engagement in open- or closed-skill sports before the age of 18. The study found that open-skill athletes demonstrated greater cognitive flexibility and working memory capacity than closed-skill athletes [28]. However, the findings in this area are not entirely consistent. Similar to the study by De Waelle, Möhring et al. also examined the executive functions of children participating in open- and closed-skill sports. Their results revealed benefits of open-skill sports for cognitive flexibility but no cognitive advantage in inhibitory control or working memory [25]. The findings reported by Chang et al. appear more inconsistent with the studies described above, as their results showed that elite wushu athletes and marathon runners did not exhibit superior behavioral performance compared with non-athletes across a range of cognitive tasks(WAIS-III, Stroop, WCST, Tower of London Task), including measures of executive function [29].

Taken together, existing findings on the effects of sport type on executive function generally fall into three patterns: (1) open-skill sports are associated with greater improvements across multiple executive-function subcomponents compared with closed-skill sports [27, 28, 30]; (2) open-skill sports show more pronounced benefits for only certain executive-function subcomponents [25, 31, 32]; and (3) open-skill sports do not confer clear advantages for executive function relative to closed-skill sports [29]. Why might this pattern exist? One possibility is that some studies focus on only a subset of executive function components [33–35], which limits the ability to characterize how different executive function domains relate to one another. A second possibility is that studies adopting broader executive function batteries often rely exclusively on behavioral measures [35, 36]. Behavioral measures, however, do not directly index the underlying cognitive operations or the neural dynamics that support executive control. Neurocognitive evidence suggests that individuals can exhibit similar behavioral performance while relying on different neural control strategies or engaging distinct fatigue-related neural dynamics [37]. Consistent with this view, Wang et al. reported that after an acute bout of aerobic exercise, obese adolescents showed increased recruitment of attentional resources related to inhibitory control, even though behavioral indices did not improve significantly [38]. Together, these findings underscore the added value of neural measures beyond reaction time and accuracy alone. Accordingly, this study used event-related potentials (ERPs) and focused on three components of executive function to capture stage-specific executive processes underlying the effects of sports training on executive functioning.

ERPs technology, with millisecond-level temporal resolution, enables precise tracking of the neural dynamics of executive sub-processes, thereby compensating for the limitations of behavioral measures in cognitive research [39]. Consequently, an increasing number of scholars have turned to ERP to explore the intrinsic cognitive-neural mechanisms through which different types of exercise affect executive function [32, 34, 35, 40, 41]. EEG recordings capture brain activity in real time during task execution, offering valuable perspectives on the neural basis of executive processing. The ERP components P2, N2, and P3 have been found to reflect perceptual processing [42–44], response selection [32, 45, 46], and cognitive resource allocation [43, 47, 48], with distinct patterns observed across tasks targeting inhibitory control, working memory, and cognitive flexibility [31, 35, 40–42]. The P2 component, a fronto-central positive deflection typically appearing 170–200 ms after stimulus onset, is commonly interpreted as selective attention and early perceptual processing, forming the basis for subsequent cognitive operations [43, 49–52]. Previous studies have reported that physically active individuals tend to exhibit larger P2 amplitudes accompanied by faster reaction times across executive function tasks, suggesting more efficient early attentional processing.

Following P2, the N2 component appears as a fronto-central negative deflection within 280–320 ms post-stimulus and is primarily associated with conflict monitoring [45, 46, 53], response conflict [54], decision-making [55] and response inhibition [56]. Because both the Stroop and task-switching paradigms require the detection and resolution of competing stimulus-response mappings, N2 provides a theoretically relevant index of early conflict-processing operations in these tasks [42, 57]. In some inhibitory-control contexts, smaller N2 amplitudes have been observed alongside faster or more accurate behavioral responses and have been interpreted as reflecting reduced conflict-monitoring demands or more efficient early inhibitory processing. For example, Chen et al. reported that badminton athletes exhibited faster response times and smaller N2 amplitudes than non-athletes during an inhibitory-control task, suggesting that reduced N2 amplitudes may be associated with more efficient early-stage inhibitory processing [33]. Similarly, Drollette et al. found that 20 min of moderate-intensity aerobic exercise, children demonstrated improved accuracy and reduced N2 amplitudes in an inhibitory control task, indicating enhanced efficiency in conflict processing after acute exercise [58].

The P3 component—a parietal positive deflection occurring 300–600 ms post-stimulus—is associated with cognitive resource allocation and context updating, and has also been linked to decision-related processes such as evidence accumulation and decision confidence or uncertainty [32, 34, 35, 41, 47, 59–61]. Previous studies have suggested that larger P3 amplitudes are associated with greater allocation of task-relevant attentional resources and stronger engagement of cognitive processes required for task performance [31, 49, 62, 63]. In older adults, both aerobic exercise and martial arts have been associated with higher accuracy and faster response times in the Stroop task, alongside enhanced P3 amplitudes—suggesting increased attentional resource allocation during cognitive processing [31]. In adolescent populations, regular participation in physical activity, compared with irregular exercise, has also been associated with faster response times and higher P3 amplitudes, suggesting a greater allocation of attentional resources [49]. When examining exercise type, elderly participants involved in either open- or closed-skill sports responded more quickly than controls. However, reduced N2 amplitudes were observed only in those practicing open-skill sports, suggesting more effective conflict monitoring and inhibition of irrelevant information [35]. With respect to the P3 component, prior research has further demonstrated that athletes engaged in open-skill sports typically exhibit larger P3 amplitudes during inhibitory control and cognitive flexibility tasks. These effects are sometimes accompanied by sport-type-related differences in P3 scalp distribution (e.g., more pronounced central/parietal involvement) and, in some studies, reduced switching costs compared with closed-skill athletes or sedentary control groups [32, 34, 35, 40, 62].

Although numerous studies have examined the association between sport participation and executive function, the available evidence remains inconsistent. Two key questions therefore remain open: whether athletes engaged in open- versus closed-skill sports show systematic differences in inhibition, task switching, and working memory, and if such differences do exist, at which stages of information processing they emerge. To address these issues, the present study employed three tasks that respectively target these core domains of executive function (the Stroop task for inhibitory control, task-switching task for cognitive flexibility, and N-back task for working memory), thereby reducing the likelihood that any conclusions hinge on a single paradigm or on an incomplete sampling of executive processes. In parallel, ERPs were recorded to disentangle stage-specific neural processes that may not be apparent from reaction time and accuracy measures alone. This design permits a more mechanistic examination of how different types of sport are related to executive control at both behavioral and neural levels.

Based on previous research [23, 32, 34, 35, 40, 41, 64–66], we hypothesize that athletes will demonstrate faster reaction times and higher accuracy than non-athletes across the three cognitive tasks. Additionally, we expect open-skill athletes to outperform closed-skill athletes in behavioral performance. Regarding EEG results, previous research suggests that sport participation may be associated with differences in neurocognitive processing related to executive function; therefore, we predict that athletes will show more advantageous task-related neurocognitive processing [58, 67, 68]. Specifically, this may be reflected in larger P2 and P3 amplitudes and smaller N2 amplitudes in athletes than in non-athletes across the three tasks [35, 49, 52, 58, 66, 69–71]. Additionally, given the higher cognitive demands of open-skill sports, we hypothesize that open-skill athletes will exhibit larger P2 and P3 amplitudes and smaller N2 amplitudes relative to closed-skill athletes. Moreover, because our control group consists of university students—who are at a high cognitive baseline—differences between groups may be attenuated, making inter-group effects more difficult to detect [72].

Methods

Participants

The required sample size was pre-calculated using G*power 3.1 software [73]. For the two-way ANOVA, the parameters were set as follows: repeated measures ANOVA was selected, with effect size set at 0.25, significance level (α) at 0.05, power at 0.8, and degrees of freedom at 2. This calculation determined that 42 participants were needed. For the three-way ANOVA, the parameters were set as follows: repeated measures ANOVA was selected, with effect size set at 0.25, significance level (α) at 0.05, power at 0.8, and degrees of freedom at 6. This calculation determined that 24 participants were needed. Therefore, this study recruited 39 athletes and 20 healthy non-athlete volunteers as paid participants through advertisements. The athletes included 20 open-skill athletes (basketball, soccer) and 19 closed-skill athletes (track and field). All athletes were at least national level-2 athletes with over 5 years of training experience and maintaining sport-specific training (> 5 h/week) during college. All the non-athlete volunteers did not engage in regular structured training. The control group consisted of non-athletes matched for age and education level. All participants were healthy, right-handed, had normal or corrected-to-normal vision, and no history of psychiatric disorders. Due to excessive ERP artifacts, two participants were excluded, leaving 19 participants in each group (open-skill athletes: 18 males, aged 18–26 years, mean age 20.78 years; closed-skill athletes: 17 males, aged 18–25 years, mean age 21.17 years; control group: 16 males, aged 19–23 years, mean age 21.26 years). The data of these participants were included in the behavioral and EEG statistical analysis. All participants provided informed consent, and the study was approved by the Ethics Committee of Liaoning Normal University (approval number: LL2025100).

Materials and methods

Stroop task

The experiment consisted of 16 types of stimuli, composed of Chinese characters “红” (red), “绿” (green), “蓝” (blue), and “黄” (yellow) in four different colors (red, green, blue, yellow), with each stimulus being of the same size. There were four congruent stimulus conditions (color-word consistency) and twelve incongruent stimulus conditions (color-word inconsistency), with the background of the stimulus display being gray. The procedure for each trial is as follows: a fixation cross “+” is displayed at the center of the screen for 500 ms, followed by a random blank interval lasting between 500 and 800 ms. Then Chinese character were present for 1000 ms, followed by another random blank interval lasting between 1 and 1.5 s. Participants were required to ignore the meaning of the characters and respond only to the color of the characters by pressing the corresponding keys: the “D” key for red, “F” key for green, “J” key for blue, and “K” key for yellow. The keys were assigned such that the left hand controlled “D” and “F,” while the right hand controlled “J” and “K.” The mapping between the response keys (D, F, J, and K) and ink colours was held constant for all participants. The experiment consisted of a practice session and a formal session, with the formal session divided into two blocks and a break allowed between them. The experiment included a total of 240 trials, with 120 trials for the congruent condition and 120 trials for the incongruent condition; stimuli under each condition were presented in a random sequence (see Fig. 1).

Fig. 1.

Flowchart of the Stroop task

Task-switching task

The experimental materials consisted of red or green numbers from 1 to 9 (excluding 5) and two types of task instructions. When the instruction “Color Judgment Task” was presented, participants judged the color of the number, pressing the “F” key for green and the “J” key for red. When the instruction “Size Judgment Task” was given, participants judged the size of the number, pressing the “F” key for numbers smaller than 5 and the “J” key for numbers greater than 5. The experiment was divided into three blocks. The first and second blocks were non-switching tasks, and the third block was a switching task. The first block was a color judgment task, consisting of 40 trials. The procedure for each trial was as follows: a black fixation cross “+” appeared at the center of the screen for 500 ms, followed by a 1000 ms gray blank screen. Then, both a 1500 ms text prompt and a number stimulus were presented simultaneously, followed by a 1000 ms gray screen. The second block was a size judgment task, also consisting of 40 trials, and the procedure for each trial was the same as in the first block. The third block was a switching block, consisting of 180 trials. In each trial, participants had to perform either the color or size judgment task according to the text prompt. The specific procedure was the same as in the first two blocks (see Fig. 2).

Fig. 2.

Flowchart of the task-switching task

N-back task

The experimental materials consisted of six numbers (1–6), with each stimulus randomly presented in sequence. The experiment included two blocks: 1-back and 2-back, with each block consisting of 60 trials. In both blocks, the procedure for each trial was as follows: first, a gray screen was presented for 1000 ms, followed by a 300 ms number stimulus, and then another gray screen for 1500 ms. Participants were required to determine whether the currently presented number was the same as the previous one. If the numbers were the same, participants pressed the “F” key; if they were different, they pressed the “J” key. Finally, a randomly blank screen lasting 1000–1500 ms was presented to reduce temporal predictability and to maintain participants’ attention across trials. Therefore, the total stimulus onset asynchrony ranged from 2,800 to 3,300 ms. In the 2-back condition, participants compared the currently presented number with the one presented two trials earlier, with no responses made to the first two numbers (see Fig. 3).

Fig. 3.

Flowchart of the N-back task

Electroencephalographic recordings

EEG data were recorded using a 64-channel Ag/AgCl electrode cap arranged according to the extended international 10–20 system. The EEG signals were amplified through a Brain Amp amplifier (Brain Products GmbH, Germany) with a band-pass of 0.01 Hz to 100 Hz and sampled at 500 Hz. All electrode impedances were kept below 5 kΩ. FCz was used as the online reference during recording, and the data were re-referenced offline to the average of the bilateral mastoids. Offline analyses were conducted using the ERPlab v.13.5.4b (MathWorks, Natick, USA) plugin. After manually removing bad segments, the data were band-pass filtered from 0.01 to 30 Hz. Ocular artifact correction was performed through independent component analysis in EEGLAB [74]. Ocular components were identified based on their topographical distribution, time course, and power spectrum. Trials with amplitudes exceeding ± 80 µV after artifact correction were excluded. ERPs were analyzed within a 1000 ms window post-stimulus. Baseline correction was performed using the − 200 to 0 ms pre-stimulus interval, as this period precedes stimulus onset and provides a stable reference that minimizes slow drifts while avoiding contamination from stimulus-evoked activity. ERP waveforms were averaged across trials for each experimental condition, and then across conditions and participants to obtain a grand average waveform for each participant. The trial retention rate exceeded 75% across all groups and conditions.

Based on the scalp distribution and previous studies, the following components and time windows were selected for analysis: P2, N2, and P3 components at the Fz, Cz, and Pz electrode sites in the Stroop, task-switching, and N-back tasks. In the Stroop task, the P2 time window was 180–220 ms [49, 75], the N2 time window was 270–370 ms [49], and the P3 time window was 370–430 ms [31, 49]. In the task-switching task, the P2 component was analyzed within a time window of 180–240 ms [42, 52, 76], the N2 component within 280–360 ms [76], and the P3 component within 360–440 ms [52]. In the N-back task, the P2 component was analyzed within 140–180 ms [77, 78], the N2 component within 220–320 ms [78, 79], and the P3 component within 320–420 ms [77, 80].

Statistical analysis

Data were analyzed using SPSS 22.0 software. For the Stroop task, accuracy and correct trial response times were analyzed using a two-way ANOVA (Condition: congruent, incongruent × Group: open-skill athletes, closed-skill athletes, control group). The P2, N2, and P3 amplitudes in the Stroop task were analyzed separately using a three-way ANOVA (Condition: congruent, incongruent × Group: open-skill athletes, closed-skill athletes, control group × Electrode: Fz, Cz, Pz). For the Task-switching task, accuracy and correct trial response times were analyzed using a two-way ANOVA (Condition: switch, non-switch × Group: open-skill athletes, closed-skill athletes, control group). The P2, N2, and P3 amplitudes in the Task-switching task were analyzed separately using a three-way ANOVA (Conditions: switch, non-switch × Group: open-skill athletes, closed-skill athletes, control group × Electrode: Fz, Cz, Pz). For the N-back task, accuracy and correct trial response times were analyzed using a two-way ANOVA (Condition: 1-back, 2-back × Group: open-skill athletes, closed-skill athletes, control group). The P2, N2, and P3 amplitudes in the N-back task were analyzed separately using a three-way ANOVA (Condition: 1-back, 2-back × Group: open-skill athletes, closed-skill athletes, control group × Electrode: Fz, Cz, Pz). ERP analyses were restricted to pre-defined components and midline electrodes (Fz/Cz/Pz) based on prior literature. Greenhouse-Geisser corrections were applied when the sphericity assumption was violated, and Bonferroni adjustments were used for post hoc comparisons where appropriate.

Results

Behavioral data

Stroop task

For accuracy, the results showed a significant main effect of Condition, with accuracy in the congruent condition (92.28 ± 5.54) being significantly higher than in the incongruent condition (83.41 ± 9.2) (F_1,54=70.094, p < 0.001, η² = 0.798). The main effect of Group was not significant (F_2,54 = 2.310, p = 0.114, η² = 0.114), and the Group × Condition interaction was also not significant (F_2,54=0.917, p = 0.409, η²=0.048) (See Table 1).

Table 1.

Accuracy and reaction time (ms) of the participants in the Stroop task, Task-switching task, and N-back task

	Open-skill athletes	Closed-skill athletes	Control group
Stroop task
Congruent accuracy (%)	94.39 ± 3.18	91.14 ± 6.88	91.32 ± 5.59
Incongruent accuracy (%)	86.36 ± 6.36	83.33 ± 8.00	80.53 ± 12.09
Congruent reaction time (ms)	583.77 ± 49.40	612.52 ± 42.73	615.65 ± 44.17
Incongruent reaction time (ms)	648.37 ± 49.02	675.41 ± 49.96	695.63 ± 42.00
Task-switching task
Non-switch accuracy (%)	97.50 ± 1.77	96.25 ± 2.50	97.30 ± 2.05
Switch accuracy (%)	94.94 ± 1.98	94.82 ± 3.24	94.65 ± 2.73
Non-switch reaction time (ms)	610.25 ± 83.57	639.36 ± 74.34	681.81 ± 119.14
Switch reaction time (ms)	788.00 ± 85.68	858.98 ± 92.76	907.12 ± 100.79
N-back task
1-back accuracy (%)	89.04 ± 5.45	89.65 ± 6.56	93.60 ± 5.07
2-back accuracy (%)	89.74 ± 4.49	88.60 ± 7.46	88.95 ± 7.66
1-back reaction time (ms)	237.41 ± 75.51	238.55 ± 63.75	298.01 ± 119.60
2-back reaction time (ms)	317.63 ± 112.70	363.34 ± 117.82	494.93 ± 172.37

For reaction time, the main effect of Condition was significant, with reaction time in the congruent condition (603.98 ± 46.99 ms) being significantly faster than that in the incongruent condition (673.13 ± 50.23 ms) (F_1,54 = 406.847, p < 0.001, η² = 0.958). The main effect of Group was significant (F_2,54 = 3.881, p = 0.03, η² = 0.177). The Group × Condition interaction was not significant (F_2,54 = 2.615, p = 0.087, η² = 0.127). Although the omnibus Group effect was significant, corrected post hoc comparisons did not identify any significant pairwise differences among the three groups. Specifically, the difference between open-skill athletes and controls did not reach significance after correction (p = 0.056), and no significant differences were observed between closed-skill and open-skill athletes (p = 0.195) or between closed-skill athletes and controls (p = 1.000). Thus, open-skill athletes showed numerically shorter reaction times, but the corrected pairwise comparisons did not provide sufficient evidence for a reliable difference between any two groups (See Table 1).

Task-switching task

For accuracy, the main effect of Condition was significant (F_1,54 = 34.329, p๤0.001, η² = 0.656), with the accuracy in the switch condition (94.81 ± 2.63) being significantly lower than in the non-switch condition (97.02 ± 2.16). The main effect of Group was not significant (F_2,54 = 0.423, p = 0.658, η² = 0.023), neither the interaction between Condition and Group (F_2,54 = 2.130, p = 0.134, η² = 0.106). (See Table 1)

For reaction time, the main effect of Group was significant (F_2,54 = 6.181, p = 0.005, η² = 0.256), Multiple comparisons showed that the reaction time of the open-skill athletes was significantly faster than that of the control group (p = 0.014). However, reaction times did not differ significantly between the closed-skill athletes and either the open-skill or the control group (p = 0.122, p = 0.392). The main effect of Condition was significant (F_1,54 = 234.197, p๤0.001, η² = 0.929), with reaction times in the non-switch condition (643.81 ± 97.28 ms) being significantly faster than in the switch condition (851.37 ± 104.05 ms), and the Group × Condition interaction was not significant (F_2,54 = 2.003, p = 0.150, η² = 0.100). (See Table 1)

N-back task

For accuracy, the main effect of Condition was not significant (F_1,54 = 2.468, p = 0.134,η² = 0.121). The main effect of Group was not significant (F_2,54 = 1.263, p = 0.295, η² = 0.066), and the Group × Condition interaction was not significant (F_2,54 = 2.149, p = 0.131, η² = 0.107). (See Table 1)

For reaction time, the main effect of Condition was significant (F_1,54 = 54.586, p๤0.001, η² = 0.752), with reaction time in the 1-back condition (257.99 ± 92.48 ms) being significantly faster than in the 2-back condition (391.97 ± 154.42 ms). The main effect of Group was significant (F_2,54 = 7.618, p = 0.002, η² = 0.297), with the reaction time of both the open-skill and closed-skill athletes being significantly faster than that of the control group (p = 0.010, p = 0.038). The Group × Condition interaction was significant (F_2,54 = 4.920, p = 0.013, η² = 0.215). Simple effects analysis showed that, in the 1-back condition, the reaction time of the open-skill athletes was significantly faster than that of the control group (p = 0.045), while there were no significant differences between the closed-skill athletes and the other two groups (p = 0.960, p = 0.084). In the 2-back condition, both the open-skill and closed-skill athletes showed significantly faster reaction times than the control group (p = 0.003, p = 0.006), while there was no significant difference between these two groups (p = 0.257). These results indicate that both groups of athletes responded faster than the control group, and that the significant interaction reflects a stronger advantage at a specific load level; specifically, compared with the 1-back condition, athletes exhibited a more pronounced reaction time advantage over non-athletes in the 2-back condition (See Table 1).

EEG data

Stroop task

For the P2 amplitude, the main effect of Condition was significant (F_1,54 = 5.894, p = 0.026, η² = 0.247), with the P2 amplitude in the congruent condition (6.94 ± 4.51 µV) being significantly larger than in the incongruent condition (6.40 ± 4.49 µV). The main effect of Electrode was significant (F_2,108 = 37.70, p๤0.001, η² = 0.677), with the P2 amplitude at Pz (4.78 ± 4.04 µV) being significantly smaller than at Fz (7.56 ± 4.54 µV) and Cz (7.67 ± 4.35 µV) (p๤0.001,p๤0.001), and there was no significant difference between the P2 amplitude at Cz and Fz (p = 1.000). The the main effect of Group was not significant (F_2,54 = 0.090, p = 0.914, η² = 0.005); the Group × Condition interaction was not significant (F_2,54 = 0.589, p = 0.560, η² = 0.032);Group × Electrode × Condition interaction was not significant (F_4,108 = 1.748, p = 0.149, η² = 0.089).

For the N2 amplitude, the main effect of Electrode was significant (F_2,108 = 34.587, p๤0.001, η² = 0.658), with the N2 amplitude at Fz (0.06 ± 4.71 µV) being significantly larger than at Cz (2.42 ± 4.06 µV) and Pz (4.50 ± 3.30 µV) (p๤0.001, p๤0.001), and the N2 amplitude at Cz was significantly larger than at Pz (p = 0.002). The main effect of Condition was not significant (F_1,54 = 2.728, p = 0.116, η² = 0.132);the main effect of Group was not significant (F_2,54 = 0.835, p = 0.442,η² = 0.044); the Group × Condition interaction was not significant (F_2,54 = 0.047, p = 0.954, η² = 0.003);the Group × Electrode × Condition interaction was not significant (F_4,108 = 0.218, p = 0.928, η² = 0.012) (See Fig. 4).

Fig. 4.

The left panel shows the ERP waveforms at Fz, Cz, and Pz electrodes in the Stroop task, with the colored regions from left to right representing the P2, N2, and P3 components. The right panel shows the topographic maps of the P2, N2, and P3 components under congruent and incongruent conditions, as well as the difference topographic map (congruent minus incongruent)

For the P3 amplitude, the main effect of Group was significant (F_2,54 = 3.266, p = 0.050, η² = 0.154), with the open-skill athletes (6.30 ± 4.99 µV) showing a significantly larger P3 amplitude than the control group (3.76 ± 4.26 µV) (p = 0.038). The closed-skill athletes (3.80 ± 4.76 µV) did not differ significantly from either the open-skill athletes or the control group (p = 0.169, p = 1.000). The main effect of Condition was significant (F_1,54 = 18.953, p๤0.001, η² = 0.154), with the P3 amplitude in the congruent condition (5.01 ± 4.80 µV) being significantly larger than in the incongruent condition (4.23 ± 4.82µV) (p๤0.001). The Group × Condition interaction was not significant (F_2,54 = 0.581, p = 0.564, η² = 0.031).

Regarding electrode distribution, the main effect of Electrode was significant (F_2,108 = 23.642, p๤0.001, η² = 0.568), with the P3 amplitude at Pz (6.37 ± 3.67 µV) being significantly larger than at Cz (4.97 ± 4.60 µV) and Fz (2.52 ± 5.27 µV) (p๤0.001, p๤0.001), and the P3 amplitude at Cz was significantly larger than at Fz (p๤0.001). The Group × Electrode × Condition interaction was not significant (F_4,108 = 0.895, p = 0.471, η² = 0.047).

Overall, the Stroop task showed no reliable group differences in the early components (P2, N2). Group-related effects were mainly observed at the P3 stage, with open-skill athletes exhibiting larger P3 amplitudes than controls (See Fig. 4).

Task-switching task

For the P2 amplitude, the main effect of Condition was significant (F_1,54 = 13.030, p = 0.002, η² = 0.420), with the P2 amplitude in the non-switch condition (2.86 ± 3.43 µV) being significantly larger than in the switch condition (2.05 ± 3.31 µV). The main effect of Electrode was significant (F_2,108 = 9.792, p๤0.001, η² = 0.352), with the P2 amplitude at Pz (1.55 ± 2.79 µV) being significantly smaller than at Fz (3.03 ± 3.80 µV) and Cz (2.78 ± 3.35 µV) (p = 0.021, p = 0.004), and there was no significant difference between the P2 amplitude at Cz and Fz (p = 0.723). The main effect of Group was not significant (F_2,54 = 0.999, p = 0.378,η² = 0.053); the Group × Condition interaction was not significant (F_2,54 = 1.295, p = 0.286, η² = 0.067); the Group × Electrode × Condition interaction was not significant (F_4,108 = 1.690, p = 0.162, η² = 0.086)(See Fig. 5).

Fig. 5.

The left side is the waveform diagrams of the task-switching task at three electrode sites Fz, Cz, Pz, and the colored parts from left to right are the P2, N2, P3 components in sequence. The right side is the brain topographic maps of P2, N2, P3 components under non-switch and switch conditions, and the topographic maps of the difference between non-switch and switch conditions (non-switch minus switch)

For the N2 amplitude, the main effect of Condition was significant (F_1,54 = 20.076, p๤0.001, η² = 0.527), with the N2 amplitude in the switch condition (0.40 ± 4.06 µV) being significantly larger than in the non-switch condition (1.67 ± 3.82 µV). The main effect of Group was significant (F_2,54 = 7.460, p = 0.002, η² = 0.293), with the N2 amplitude in the open-skill athletes (2.79 ± 3.54 µV) being significantly smaller than in the control group (-0.56 ± 4.17 µV) (p = 0.005), while there were no significant differences between the closed-skill athletes (0.88 ± 3.52 µV) and the open-skill and control groups (p = 0.083, p = 0.387). In this context, smaller N2 amplitudes combined with faster responses may indicate reduced neural cost during conflict monitoring. The main effect of Electrode was significant (F_2,108 = 31.048, p๤0.001, η² = 0.633), with the N2 amplitude at Fz (-0.54 ± 4.02 µV) being significantly larger than at Cz (0.82 ± 3.84 µV) and Pz (2.83 ± 3.36 µV) (p๤0.001, p๤0.001), and the N2 amplitude at Cz was significantly larger than at Pz (p๤0.001). The Group × Condition interaction was not significant (F_2,54 = 0.179, p = 0.837, η² = 0.010); the Group × Electrode × Condition interaction was not significant (F_4,108 = 0.671, p = 0.615, η² = 0.036)(See Fig. 5).

For the P3 amplitude, the main effect of Condition was significant (F_1,54 = 44.239, p๤0.001, η² = 0.711), with the P3 amplitude in the non-switch condition (4.34 ± 4.44 µV) being significantly larger than in the switch condition (2.18 ± 4.66 µV). The main effect of Group was significant (F_2,54 = 4.699, p = 0.015, η² = 0.207), with the P3 amplitude in the open-skill athletes (5.08 ± 4.18 µV) being significantly larger than in the control group (1.85 ± 4.92 µV), whereas the closed-skill athletes (2.84 ± 4.34 µV) did not significantly differ from either the control group or the open-skill athletes (p = 1.000, p = 0.055) after correction. The main effect of Electrode was significant (F_2,108 = 62.270, p๤0.001, η² = 0.789), with the P3 amplitude at Pz (5.88 ± 4.01 µV) being significantly larger than at Fz (0.72 ± 4.18 µV) and Cz (3.17 ± 4.37 µV) (p๤0.001, p๤0.001), and the P3 amplitude at Cz was significantly larger than at Fz (p๤0.001). The Group × Electrode × Condition interaction was not significant (F_4,108 = 0.287, p = 0.886, η² = 0.016); the Group × Condition interaction was not significant (F_2,54 = 0.550, p = 0.581, η² = 0.030).

In the task switching task, P2 amplitudes did not differ significantly among groups. Open-skill athletes showed smaller N2 amplitudes and larger P3 amplitudes than controls (see Fig. 5).

N-back task

For the P2 amplitude, the main effect of Condition was significant (F_1,54 = 10.025, p = 0.005, η² = 0.358), with the P2 amplitude in the 1-back condition (3.68 ± 4.07 µV) being significantly smaller than that in the 2-back condition (4.52 ± 4.09 µV). The main effect of Electrode was significant (F_2,108 = 14.573, p๤0.001, η² = 0.447), with the P2 amplitude at Pz (3.04 ± 3.51 µV) being significantly smaller than that at Fz (4.70 ± 4.43 µV) and Cz (4.56 ± 4.12 µV) (p = 0.006, p๤0.001), and there was no significant difference in the P2 amplitude between Fz and Cz (p = 1.000). The main effect of Group was not significant (F_2,54 = 2.242, p = 0.121, η² = 0.111); Group × Condition interaction was not significant (F_2,54 = 0.768, p = 0.471, η² = 0.041); Group × Electrode × Condition interaction was not significant (F_4,108 = 0.382, p = 0.821, η² = 0.021) (See Fig. 6).

Fig. 6.

The left side is the waveform diagrams of the N-back task at three electrode sites Fz, Cz, Pz, and the colored parts from left to right are the P2, N2, P3 components in sequence. The right side is the brain topographic maps of the P2, N2, P3 components under the 1-back and 2-back conditions, as well as the topographic maps of the difference between 1-back and 2-back conditions (1-back minus 2-back)

For the N2 amplitude, the main effect of Condition was significant (F_1,54 = 7.930, p = 0.011, η² = 0.306), with the N2 amplitude in the 1-back condition (1.41 ± 4.92 µV) being significantly larger than that in the 2-back condition (2.29 ± 4.21 µV). The main effect of Group was significant (F_2,54 = 3.821, p = 0.031, η² = 0.175). However, corrected post hoc comparisons did not identify a significant difference between open-skill athletes and the control group (p = 0.055), although the mean N2 amplitude was numerically smaller in the open-skill group. No significant differences were found between closed-skill athletes and open-skill athletes or between closed-skill athletes and controls (p = 0.672, p = 0.416). The main effect of Electrode was significant (F_2,36 = 49.970, p๤0.001, η² = 0.735), with the N2 amplitude at Fz (-0.04 ± 4.57 µV) and Cz (1.64 ± 4.70 µV) being significantly larger than that at Pz (3.95 ± 3.53 µV) (p๤0.001, p๤0.001), and the N2 amplitude at Fz was significantly larger than that at Cz (p๤0.001). The Group × Condition interaction was not significant (F_2,36 = 0.251, p = 0.779, η² = 0.014);the Electrode × Group interaction was not significant (F_4,72 = 0.921, p = 0.457, η² = 0.049); the Group × Electrode × Condition interaction was not significant (F_4,72 = 0.950, p = 0.440, η² = 0.050) (See Fig. 6).

For the P3 amplitude, the main effect of Group was significant (F_2,36 = 8.865, p = 0.001,η² = 0.330), with the P3 amplitude in the open-skill athletes (6.47 ± 4.41 µV) and the closed-skill athletes (4.36 ± 4.00 µV) being significantly larger than that in the control group (1.60 ± 4.29 µV) (p = 0.007, p = 0.028), while there was no significant difference in the P3 amplitude between the open-skill athletes and the closed-skill athletes (p = 0.212). The main effect of Electrode was significant (F_2,36 = 49.970, p๤0.001, η² = 0.735), with the P3 amplitude at Pz (6.05 ± 3.93 µV) being significantly larger than that at Fz (2.26 ± 4.59 µV) and Cz (4.11 ± 4.70 µV) (p๤0.001, p๤0.001), and the P3 amplitude at Cz was significantly larger than that at Fz (p๤0.001). The main effect of Condition was not significant (F_1,55 = 3.308, p = 0.086, η² = 0.155); the Group × Condition interaction was not significant (F_2,36 = 0.777, p = 0.467, η² = 0.041); the Group × Electrode × Condition interaction was not significant (F_4,72 = 0.277, p = 0.892, η² = 0.015) .

In the N-back task, P2 showed no reliable group differences. For N2, although the omnibus Group main effect was significant, corrected post hoc comparisons did not identify any significant pairwise differences, including the open-skill versus control comparison. By contrast, P3 amplitudes were larger in both athlete groups than in controls, indicating greater task-related attentional resource allocation during working-memory processing (See Fig. 6).

Discussion

The purpose of this study was to examine the associations of open-skill and closed-skill sport participation with executive function and their underlying neural correlates. We employed the Stroop task, task-switching task, and N-back task, combined with ERP, to assess behavioral and neuroelectric correlates of executive function among athletes of different sport types and non-athletes. Behaviorally, open-skill athletes showed faster responses than controls in the task-switching and N-back tasks, whereas closed-skill athletes showed faster responses mainly under higher working-memory load. EEG results indicated no group differences in P2 amplitude across tasks. In the task-switching task, open-skill athletes showed significantly smaller N2 amplitudes than controls. In contrast, P3 amplitudes showed group differences across all three tasks: in the Stroop and task-switching tasks, open-skill athletes showed larger P3 amplitudes than controls; and in the N-back task, both athlete groups elicited larger P3 amplitudes than controls. We discuss these findings in terms of the three executive subcomponents in the following sections.

Inhibitory control (Stroop task)

The Stroop task was used to assess inhibitory control. Behavioral results showed significantly faster response times and lower error rates under the congruent condition than under the incongruent condition, reflecting the classic Stroop effect. This indicates that participants needed to suppress automatic, incorrect responses under the incongruent condition, leading to slower responses and higher error rates [32, 49, 81, 82]. No significant group difference was observed in Stroop reaction time or accuracy. Thus, the behavioral data did not provide evidence for a sport-type-related advantage in inhibitory control.

At the electrophysiological level, we observed no group differences in P2 or N2 during the Stroop task. By contrast, a between-group effect emerged for P3: open-skill athletes showed larger P3 amplitudes than controls. This pattern suggests that, in the Stroop task, differences among sport types and non-athletes may not primarily arise during early perceptual processing or subsequent conflict monitoring, but rather during later-stage recruitment of attentional resources required to implement task goals under interference conditions [49, 83–89]. This result is consistent with previous research showing that older adults who engage in open-skill sports exhibit larger P3 amplitudes than their non-exercising peers [35]. The P3 is related to attentional resource allocation [87], and the P3 amplitude reflects the amount of attentional resources allocated when individuals process target stimuli [47, 88]. Thus, the larger P3 amplitudes in open-skill athletes may reflect greater involvement of later-stage task-relevant processing during the Stroop task. For example, basketball players must inhibit task-irrelevant distractions and rapidly prioritize goal-relevant cues [90–92]; such repeated cognitive demands may be associated with enhanced attentional resource recruitment.

Cognitive flexibility (Task-switching task)

In the task-switching task, no significant accuracy differences were found among groups, but open-skill athletes responded significantly faster than controls, whereas closed-skill athletes did not differ significantly from either of the other two groups. This finding is consistent with previous research [52, 90], suggesting that open-skill athletes may show better task-switching performance, reflected in their ability to adapt to changing rules, shift flexibly between task sets, and maintain efficient performance under changing task demands [1, 93]. Open-skill sports involve rapid decision-making under dynamic conditions. For example, basketball players must constantly track teammates and opponents to adjust strategies and motor actions [89, 90]. In contrast, closed-skill sports such as swimming take place in relatively stable environments where such adaptive shifts are less frequent [91]. Thus, long-term involvement in open-skill sports may be linked to faster performance on the task-switching paradigm, possibly because these sports repeatedly require real-time cognitive adjustments.

In this task, no group differences were observed for the P2 component, indicating that during task switching, athletes from different sport types did not differ at the early stage of cognitive processing [42, 52, 62, 92]. However, N2 amplitudes were significantly smaller in open-skill athletes than in controls. Although larger N2 amplitudes are often associated with stronger cognitive control [67, 76]. In the present study, open-skill athletes showed smaller N2 amplitudes accompanied by preserved accuracy and faster responses, a pattern that may reflect lower conflict-monitoring demands during task switching. This association between reduced N2 and enhanced task performance has been documented in multiple previous studies. For example, Li et al. reported that older adults engaged in open-skill exercise exhibited better Stroop performance and smaller N2 amplitudes than sedentary peers, implying reduced cognitive effort and enhanced conflict-monitoring efficiency [35]. Similarly, adolescents with higher physical fitness have been shown to exhibit smaller N2 amplitudes than their lower-fit peers, and this reduction was not accompanied by slower reaction times or higher error rates, suggesting that they exerted less effort in response monitoring [70]. Studies on acute exercise interventions have also found that reduced N2 amplitudes are associated with enhanced conflict processing and faster classification speed [58]. Therefore, considering that the open-skill athletes in our study achieved comparable accuracy but significantly faster responses in the task-switching paradigm, their smaller N2 amplitudes may reflect reduced neural cost during conflict monitoring and more efficient task preparation [57, 70]. Because the present study is cross-sectional, this interpretation should be regarded as an association rather than evidence of a causal training effect.

Regarding P3, open-skill athletes showed larger amplitudes than controls, suggesting greater attentional resource allocation during task-set updating [60, 92]. This finding aligns with previous evidence that open-skill exercisers show larger P3 amplitudes and better cognitive flexibility than irregular exercisers [94]. Together, these results suggest that open-skill sport experience may be associated with neurocognitive profiles related to flexible cognitive control, although causal conclusions cannot be drawn from the present cross-sectional design.

Working memory (N-back task)

In the N-back task, accuracy did not differ among groups. However, closed-skill athletes showed their clearest behavioral advantage under the more demanding 2-back load, whereas open-skill athletes exhibited faster responses at both 1-back and 2-back loads. These findings indicate that open-skill athletes achieved comparable accuracy with faster responses across both difficulty levels, whereas closed-skill athletes demonstrated faster response speeds only under higher working-memory demands. The 1-back condition represents a low-load working-memory task primarily dependent on short-term retention and stimulus recognition, with minimal requirements for information updating. This task mainly reflects processing speed and attentional focus [95]. Frequent engagement in variable and dynamic sporting environments may be associated with open-skill athletes’ ability to process information rapidly and respond efficiently in low-complexity tasks [24]. Both athletic groups responded faster than controls in the more demanding 2-back condition. These findings are consistent with the view that long-term engagement in organized sport is associated with more efficient task execution under increased working-memory load. The 2-back task requires continuous updating of stimulus information and matching over greater temporal distances, thereby imposing higher cognitive demands on working-memory maintenance and executive control [96, 97]. Consistent with this, long-term, structured sports participation has been associated with processing speed, executive regulation, and attentional control, which are key components supporting efficient performance in high-load working-memory tasks [98, 99].

In the N-back task, no significant group differences were observed in P2 amplitude. As P2 is generally associated with early attentional engagement and contextual processing [100, 101], these results do not provide evidence for sport-type effects at this early stage of working-memory processing. For the N2 component, the main effect of Group reached statistical significance, but post hoc comparisons did not confirm a significant difference between open-skill athletes and controls after correction. The N2 in the N-back task is thought to index conflict monitoring between incoming stimuli and representations maintained in working memory [102]. Because the key post hoc comparison did not reach significance, the N2 findings should be interpreted cautiously and should not be taken as clear evidence of a sport-type-related difference in working-memory conflict monitoring.

Regarding the P3 component, both open-skill and closed-skill athletes demonstrated significantly larger P3 amplitudes than controls. The P3 in the N-back paradigm reflects higher-order cognitive control processes, including goal-directed response selection [103, 104] and attentional engagement supporting the integration of new stimuli with stored representations [88]. These results within the tasks used in this study suggest that athletes, irrespective of sport type, allocated greater attentional resources and exhibited stronger top-down attentional control than non-athletes. This interpretation is supported by previous findings. Chueh et al. reported that both open- and closed-skill athletes showed superior behavioral performance and larger P3 amplitudes in spatial memory tasks, suggesting enhanced neural resource allocation for visuospatial processing [105]. Another study on executive function in elderly individuals also found that after a 6-month exercise intervention, elderly individuals engaging in both open-skill and closed-skill sports showed enhanced P3 amplitudes and behavioral performance in the N-back task [94]. The present findings further suggest that long-term participation in physical exercise, whether in open-skill or closed-skill sports, may be associated with enhanced resource mobilization during working-memory updating.

Open-skill athletes (e.g., basketball and soccer players) must rapidly identify cues, update goals, and execute adaptive responses within constantly changing contexts—processes closely aligned with the dynamic updating mechanisms required in the N-back task. In contrast, closed-skill athletes (e.g., track and swimming athletes) operate in more stable environments but rely heavily on self-monitoring, rhythm maintenance, and sustained goal focus, which may foster attentional stability and goal maintenance. Together, these findings suggest that athletes in both sport types may allocate greater task-related attentional resources, although the associated neurocognitive patterns appear to differ between open-skill and closed-skill athletes.

However, this study has some limitations, which may restrict our interpretation of the results. This cross-sectional study cannot establish causal effects of exercise on executive function. The modest sample size and the high number of statistical tests, even with corrections, may limit the detection of smaller effects and should be taken into account when interpreting the findings. Additionally, the predominantly male sample limits generalizability, and potential interactions between gender and sport type warrant further exploration. Uncontrolled variables such as general physical activity, intelligence, and stress levels may have influenced the results and should be better controlled in future work. Although controls were matched on age and education and did not engage in organized sport training, variability in general physical activity could have influenced executive-function performance and/or ERP indices.

Conclusion

The present study suggests that athletes from different sport types exhibit task-specific behavioral and neuroelectric differences during executive-function tasks. Open-skill athletes showed clearer behavioral advantages in task switching and the N-back task, whereas group differences in inhibitory-control processing were more evident in P3 amplitude than in overt behavioral performance. Closed-skill athletes showed more selective advantages in the N-back task. Overall, these findings indicate that open-skill and closed-skill athletes are associated with distinct behavioral and neuroelectric profiles of executive processing. Future longitudinal and intervention studies directly comparing sport modalities are warranted to determine whether season-to-season changes in training exposure are accompanied by measurable changes in executive performance and associated neural markers.

Authors’ contributions

Yongliang Wu: Conceptualization, Data curation, Formal analysis, Writing - original draft, Writing -review & editing. Yong Jiang: Conceptualization, Writing - review & editing. Weijun Li: Conceptualization, Data curation, Writing - review & editing.

Funding

This research was supported by Ministry of Education in China Project of Humanities and Social Sciences [25YJA190006].

Data availability

Data will be made available on request.

Declarations

Ethics approval and consent to participate

The study was approved by the Ethics Committee of Liaoning Normal University (approval number: LL2025100). All participants signed the informed consent and assented to participate. The work described has been carried out in accordance with the ethical standards of the institutional research committee and/or the national research committee, and with the 1964 Helsinki declaration for experiments involving human participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.