Long-term basketball training shapes cerebellar-frontal integration for enhanced cognitive control

Abstract

Background

In competitive sports, elite athletes demonstrate exceptional proficiency in resolving sensorimotor conflicts, exemplified by the basketball head-fake phenomenon. Whether long-term basketball training leads to adaptive cognitive control in athletes and the underlying neural mechanisms is still unclear.

Methods

Using a spatial conflict task called Swimmy and functional magnetic resonance imaging, this study investigated the brain function of 50 basketball athletes and 55 gender- and age-matched healthy controls during the Swimmy tasks.

Results

Our findings revealed that basketball athletes showed distinct neurocognitive profiles between basketball athletes and non-athletes during the Swimmy task performance. Behaviorally, athletes showed better conflict resolution accuracy in incongruent trials despite similar reaction times, indicating enhanced inhibitory control. Neuroimaging revealed athlete-specific activation patterns: heightened right lingual gyrus and cerebellar lobule VI during congruent trials, while incongruent trials triggered left precuneus activation and reduced right middle frontal gyrus engagement. Athletes also exhibited attenuated left inferior parietal lobule responses in incongruent > congruent contrasts. Notably, Dynamic Causal Modeling identified that athletes exhibited reduced right Crus I→left inferior frontal gyrus (IFG) connectivity versus controls, contrasting with enhanced bidirectional Crus II-IFG coupling. Crucially, stronger IFG→Crus II connectivity positively predicted incongruent trial accuracy in athletes.

Conclusions

This athlete-specific connectivity-behavior relationship, absent in controls, indicates that long-term sport training is associated with distinct neural circuit configurations supporting cognitive control via cerebellar-frontal integration. This work contributes to our understanding of the neural correlates related to long-term sport training, particularly highlighting the cerebellum’s involvement in cognitive control. These findings inform our understanding of the cognitive benefits of long-term sport training and open up possibilities for targeted interventions that could enhance cognitive control in various populations, offering hope for future research and applications in cognitive enhancement.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13102-025-01281-5.

Introduction

The human cognitive system routinely navigates competing information streams, a fundamental challenge requiring precise inhibitory control and attentional resource allocation. Cognitive control involves a core neural network, including the anterior cingulate cortex, lateral prefrontal cortex, anterior insula, and parietal lobe, for interference resolution [1, 2]. Critically, contemporary research implicates the cerebellum as an integral component of cognitive control circuits. The ‘cognitive cerebellum’ hypothesis, which suggests that the cerebellum plays a significant role in cognitive control beyond its traditional role in motor coordination, guides our focus on the cerebellum in this study [3, 5]. Anatomical reciprocity via cerebello-thalamo-cortical circuits, functional topography (Crus I/II activation during executive tasks), and physiological mechanisms (conflict prediction, cortical efficiency optimization) substantiate this hypothesis [3, 4, 5]. Crucially, the cerebellum contributes to cognitive control through prediction-error signaling, modulating prefrontal-thalamic connectivity, and regulating inhibition timing [4]. However, key unknowns persist: the differential contributions of specific cerebellar subregions (e.g., Crus I vs. Crus II) to distinct control processes and their integration with frontoparietal networks [6, 7].

Intensive athletic training, particularly in dynamic, open-skill sports like basketball, drives experience-dependent neuroplasticity in cognitive control systems. Such sports demand rapid decision-making, complex visuospatial processing, and constant adaptation, imposing significant demands on executive functions (attention, inhibitory control, working memory) [6, 8]. Frameworks like the ‘cognitive expertise hypothesis’, which posits that prolonged practice enhances neural efficiency and cognitive performance, guide our understanding of the potential cognitive benefits of long-term basketball training [9, 10]. Empirically, athletes—especially from team sports—often outperform non-athletes on cognitive control tasks, showing faster reaction times, lower error rates, and reduced interference effects [11, 12, 13, 14]. Neuroimaging evidence further supports this, showing alterations in brain structure (e.g., increased gray matter volume, cortical thickness) and function (e.g., altered activation patterns, enhanced functional connectivity) in regions associated with executive control, including the anterior cingulate gyrus, prefrontal cortex, parietal cortex, and increasingly recognized, the cerebellum [15, 16, 17]. This aligns with athletes’ proficiency in resolving ecological conflicts, exemplified by the basketball head-fake, where incongruent gaze direction causes defender errors, paralleling conflict tasks [18, 19]. Behavioral evidence further suggests basketball training enhances domain-general cognitive performance, particularly working memory and conflict monitoring [20].

While cerebellar involvement in motor aspects of sports is established, its specific contributions to athletic advantages in cognitive control remain poorly characterized. Investigations targeting cerebellar mechanisms in athletes using established cognitive control paradigms are notably scarce. This scarcity underscores the urgency of our research and its potential impact on our understanding of the cognitive benefits of athletic training. Neuroimaging studies on athletic cognition have predominantly focused on cortical regions, leaving the role of cerebellar subregions (Crus I/II) in athletic cognitive expertise unresolved [3, 7]. Crucially, it is unknown whether athletic training modulates the cerebellum’s engagement in cognitive control circuits or how different subregions contribute to training-related neural adaptations for superior conflict resolution.

To address these gaps, we used functional magnetic resonance imaging (fMRI) during a spatial conflict task, known as the ‘Swimmy’ paradigm [21]. The Swimmy task creates conflict through conflicting local versus global directional processing (e.g., the orientation of the outline and individual triangles), offering better ecological validity for basketball cognition compared to traditional Flanker paradigms. Specifically, its local-global conflict mechanism matches athletes’ needs to quickly distinguish partial deceptive cues (like an opponent’s limb movements) from overall movement intentions while also processing individual-team coordination contexts [22, 23, 24]. The choice of the Swimmy task avoids linguistic processing limitations caused by the traditional Stroop task and more effectively captures sport-related cognitive needs than the classic Flanker task [21, 22]. Furthermore, it can effectively trigger the conflict suppression process, similar to the classic Stroop or Flanker paradigms [21, 23].

Leveraging fMRI’s spatial resolution, we captured localized activity and interregional interactions. Guided by evidence of sport-related neuroplasticity, we hypothesize that: (1) Behaviorally: Athletes will demonstrate superior conflict resolution efficiency, reflected in higher accuracy during incongruent trials. (2) Neurally: Athletes will exhibit altered activation patterns in key networks: visual-spatial processing (occipito-parietal regions), conflict monitoring (fronto-parietal networks), and cerebellum. (3) Connectivity: Long-term training will reshape functional integration between specific cerebellar subregions (Crus I/II) and prefrontal cognitive control regions, potentially correlating with behavioral performance. Through whole-brain activation mapping and dynamic causal modeling (DCM) of cortico-cerebellar circuits, this study elucidates how sport-specific expertise sculpts the neurocognitive architecture of cognitive flexibility, specifically probing the under-explored role of cerebellar subregions.

Materials and methods

Participants

This study recruited 67 active university basketball athletes along with 58 gender- and age-matched healthy controls. Due to data quality issues, such as excessive head motion or low task accuracy, 20 participants were excluded from the analysis, resulting in a final sample of 105 participants, including 50 athletes and 58 controls. All participants provided informed consent and completed a basic information questionnaire (See Supplementary Materials) before the experiment. Basketball athletes and healthy controls were concurrently recruited from Shanghai University of Sport. Inclusion criteria were as follows: (1) aged 18–28 years; (2) normal or corrected-to-normal vision; (3) right-handedness [24]; (4) physically healthy with no implanted medical devices (e.g., dental prostheses, artificial limbs, pacemakers); (5) normal cognitive function, no history of traumatic brain injury and no personal or familial history of mental disorders (See Supplementary Materials: Basic Information Form), (6) no medication, including alcohol and coffee, should be taken before scanning. Additional criteria for the athlete group: certified as national level 2 athletes or above and ≥ 5 years of basketball-specific training experience. Additional criteria for the control group: No prior professional athletic training; No history of continuous exercise habit for more than 1 year [25].

Swimmy task

We adapted the spatial conflict paradigm from Okayasu et al. (2023) to develop the Swimmy Task, a language-independent design that mitigates lateralization biases [21]. This task generates cognitive conflict through hierarchical stimuli featuring misaligned global and local directional cues (Fig. 1). Each stimulus consisted of a large acute triangle (global level) composed of smaller acute triangles sharing identical angular orientation (local level). Thirty-two unique stimuli were created by combining three factors: color (orange or green), large triangle orientation (up, down, left, or right), and small triangle orientation (up, down, left, or right). Stimuli were classified as congruent (matched global/local directions, eight stimuli) or incongruent (mismatched directions, 24 stimuli). During each trial, participants viewed random stimuli and reported either global direction when cued by green (large triangle outline) or local direction when cued by orange (individual small triangles). Responses were made using a four-button response box held in the participants’ right hands. We recorded the reaction time and accuracy for each subject. The experiment employed a blocked design with 4 sequential blocks separated by 10-second breaks. Each block contained 24 congruent and 24 incongruent trials (48 trials/block; 192 total trials, 820 s), presented in randomized order. The average consecutive occurrence per condition was 1.91 ± 1.27 per block—insufficient to elicit oddball effects or proactive control strategies [26, 27]. Each participant performed at least two practice blocks (12 trials per block) before the task, and the formal experiment began when the correct rate of practice was higher than 75%.

Fig. 1.

Swimmy Task: spatial conflict paradigm for global-local conflict. In this experiment, participants performed a cognitive conflict task by judging the apex direction of hierarchical stimuli. Composite figures consisted of a large triangular outline (global level) constructed from smaller triangles (local level). Color cues dictated the focus of judgment: green required identifying the apex direction of the global shape (large triangle), while orange required focusing on the local elements (small triangles). The task included 4 blocks, each separated by 10-second breaks. Each block comprised 24 congruent trials (global/local directional alignment) and 24 incongruent trials (directional conflict), yielding 48 trials per block and 192 trials total, 820 s

ANOVAs (2 conditions × 2 groups) were conducted on reaction time and accuracy to explore group differences in task performance. We also performed a two-sample t-test on the conflict effect (response time of congruent conditioned -incongruent conditioned) between groups.

MRI acquisition

Magnetic resonance imaging (MRI) data were collected on the Siemens Prisma 3T MRI scanner, which is equipped with a stimulation synchronous presentation system and a 64-channel head and neck coil. Firstly, obtain a high-resolution T1 weighted structural image using the following parameters: TR = 2530 ms, TE = 2.98 ms, Flip angle = 7 °, number of slices = 192, voxel size = 1 × 1 × 1 mm³, field of view = 256 × 256 mm². The field maps were collected to correct for distortions and artifacts caused by magnetic field non-uniformity. The geometric parameters of the field image are the same as those of the functional image, and the contrast parameters are as follows: TR = 571 ms, TE = 4.92 ms, 7.38 ms, Flip angle = 90 °, then play the Swimmy task and collect BOLD weighted function images during the task, with the following parameters: TR = 2000 ms, TE = 30 ms, Flip angle = 90 °, FOV = 192 × 192 mm², number of slices = 58, interlayer gap = 20%, voxel size = 2 × 2 × 2 mm³.

MRI analysis

The preprocessing of MRI data is performed using the default pipeline in DPABI software [28]. Preprocessing includes slice timing correction, head motion correction, field map correction, and co-registration using structural images to MNI space, followed by spatial smoothing using a 4 mm Gaussian smoothing kernel. Participants with head movements exceeding 2 millimeters or 2 degrees were excluded from the analysis. In the first level analysis, we include both congruent and incongruent conditions, and model the instructions to consider irrelevant noise. In addition, six head movement parameters were regressed. In the second level analysis, two-sample t-tests (FDR correction P_voxle < 0.001, P_cluster < 0.05) were conducted to compare the differences in brain activity between the two groups under the congruent condition, incongruent conditions, and incongruent minus congruent condition. We extracted the beta values of brain regions that showed differences under their respective conditions and conducted a Pearson correlation analysis with behavioral performance.

Dynamic causal modeling analysis

Referring to the study of Okayasu et al., we selected voxels of interest (VOIs) in the cerebellum and inferior frontal gyrus (IFG) region. Given that all our participants were right-handed and used their right hand to make responses in the task, we selected only the regions located in the left cerebrum and right cerebellum. To make VOIs more representative of the sample, we included all subjects to conduct a one-sample T-test (statistical thresholds were set at a voxel-wise false discovery rate (FDR) correction of p < 0.001, combined with a cluster extent threshold of > 100 contiguous voxels) for the incongruent-congruent condition, and combined the one-sample results (Fig. 2A) to get the following 3 task-related peak coordinates: the left IFG (MNI coordinates: -58, 8, 26), the right Crus I (MNI coordinates: 24, -68, -30), and the right Crus II (MNI coordinates: 16, -80, -46). At the individual level, the VOIs were defined as a sphere with a 6 mm radius, centered on the peak within a 10 mm range of the group peak coordinates. We extracted time series data from the surviving voxels within these VOIs for the incongruent minus congruent contrast (voxel-level p < 0.01, cluster size > 50), which were then used for Dynamic Causal Modeling (DCM) analysis. Eleven participants were excluded due to the absence of surviving voxels within any of the VOIs, resulting in a final sample of 94 participants—45 athletes and 49 controls—being included in the model.

Fig. 2.

Dynamic causal modeling analysis of cerebro-cerebellar effective connectivity. (A) Spatial distribution of task-modulated activation. Significant clusters identified by one-sample t-test (FDR-corrected, p < 0.001, cluster size > 100) for the incongruent > congruent contrast (red) and reverse contrast (blue). Three functionally defined regions served as volumes of interest (VOIs): left inferior frontal gyrus (IFG), right cerebellar Crus I, and right Crus II. (B) Group-level effective connectivity differences. Athletes exhibited significantly attenuated forward connectivity from right Crus I to left IFG compared to controls (blue arrow). Crucially, athletes demonstrated bidirectional connectivity enhancement in the Crus II-IFG circuit, with both strengthened forward connectivity from right Crus II to left IFG (yellow arrow) and enhanced reciprocal connectivity from IFG to Crus II (yellow arrow). Additionally, the connection from IFG to Crus II survived after false discovery rate (FDR) correction (P_FDR = 0.024). (C) The effective connectivity strength between the three brain regions of the two groups

DCM [29] was performed using DCM 12.5 from SPM12 (rev. 7771). This study aimed to investigate the group difference in effective connection strength between VOIs under the incongruent condition, rather than to explore the best model that fits the data. Thus, we specified a fully connected model with the intrinsic connectivity matrix A was defined as fully connected. The B matrix modeled the effects of the incongruent start and was assigned to all connections. No driving input was transferred to the C matrix because the visual stimuli presented in the task do not directly affect selected brain regions. In this setting, the A matrix is effective connectivity intrinsic to the incongruent condition [30]. Finally, we extracted data for each connection from all participants and conducted two-sample t-tests (uncorrected) to compare differences in interactions between VOIs under the incongruent condition.

Results

There were no significant differences between the two groups in gender, age, or years of education (Table 1).

Table 1.

Demographic information of the two groups

	Athlete	Control	T/χ2	P
N(male/female)	25:25	23:32	0.02	0.878
Age	20.46 ± 1.91	20.67 ± 2.17	0.13	0.341
Years of schooling	14.64 ± 2.01	14.56 ± 1.69	0.37	0.187
Years of basketball training	9.00 ± 3.09	0
Training time(hours/week)	7.20 ± 4.58	0
MET	5883 ± 3619	1865 ± 1678	6.94	< 0.001

MET: Metabolic Equivalent of Task, derived from physical activity scale dismissal (Craig et al. 2003)

Behavior results

The Swimmy task performance revealed distinct cognitive patterns between basketball athletes and non-athlete controls (Fig. 3). Analysis of response accuracy demonstrated a significant main effect of group (F (1,103) = 5.41, p = 0.037, η² = 0.041), with athletes showing superior conflict resolution abilities. Under congruent conditions, both groups exhibited comparable accuracy levels (athletes: 95.0% ± 3.0%; controls: 94.0% ± 3.1%; p = 0.100, Cohen’s d = 0.167). However, the group difference became pronounced during incongruent trials, where athletes maintained significantly higher accuracy (87.0% ± 6.8%) compared to controls (84.1% ± 7.5%; p = 0.045, Cohen’s d = 0.481).

Fig. 3.

Behavior results of the Swimmy task. RT: reaction time; ACC: accuracy

Reaction time analysis showed no significant group main effect (F (1,103) = 0.466, p = 0.496, η² = 0.005). Athletes demonstrated marginally faster responses across conditions (congruent: 1067 ± 142 ms; incongruent: 1231 ± 136 ms) relative to controls (congruent: 1076 ± 101 ms; incongruent: 1254 ± 114 ms), though these differences did not reach statistical significance. The conflict effect did not show statistical significance (T = -0.27, p = 0.786, Cohen’s d = -0.05) between groups. Notably, the characteristic conflict interference effect (increased latency during incongruent trials) was observed in both groups.

fMRI results

Neuroimaging analyses revealed distinct task-dependent neural signatures between professional athletes and controls under experimental conditions (Table 2; Fig. 4). As depicted in Fig. 4A, both groups demonstrated comparable positive and negative activation networks during congruent trials. Notably, athletes exhibited significantly enhanced BOLD responses within the right lingual gyrus (Brodmann area (BA) 18; Montreal Neurological Institute (MNI) coordinates: 26, -80, 22) and right cerebellar lobule VI relative to controls, as visualized in Fig. 5A. During incongruent trials, while maintaining similar fundamental activation profiles to controls (Fig. 4B), the athletic cohort displayed condition-specific neural modulations characterized by increased activation in the left precuneus (BA 31, as visualized in Fig. 5B) concomitant with reduced activity in the right middle frontal gyrus (BA 8). Whole-brain contrast analyses (incongruent > congruent) identified convergent activation patterns across groups (Fig. 4C), yet between-group comparisons revealed significant response attenuation in the left inferior parietal lobule (BA 40; peak voxel: -48, -30, 26) among athletes (Fig. 5C). In addition, follow-up correlation analyses showed no significant relationships between these differential neural signatures and Swimmy task performance behavioral measures.

Table 2.

Differences in brain activation during the swimmy task among basketball athletes compared to the non-athlete control group

Contrast	Cluster	Area	Size	T	MNI(mm)
Congruent	1	Right cerebellum posterior lobe	77(6)	4.14	26	-80	-22
		right lingual gyrus	77(53)	3.75	18	-86	-10
		right middle occipital gyrus	77(8)	3.74	-28	-92	4
Incongruent	1	Left precuneus	47 (41)	-4.29	-12	-66	28
	2	Left middle frontal gyrus	48 (13)	4.93	-26	14	38
Incongruent-Congruent	1	Left inferior parietal lobule	113 (108)	-5.67	-48	-30	26

Fig. 4.

Panels A, B, and C show the results of activation of positive and negative networks of basketball and control groups during the congruent, incongruent, and incongruent minus congruent conditions, respectively. (FDR correction P_voxle < 0.001, P_cluster < 0.05)

Fig. 5.

The differences in the congruent, incongruent, and incongruent minus congruent conditions between groups and the corresponding activation of each group

DCM results

Independent samples t-tests revealed significant between-group differences in directed neural connectivity among professional athletes (Table 3; Fig. 2). Specifically, athletes demonstrated reduced effective connectivity from the right cerebellar Crus I to the left IFG compared with controls (p < 0.029). Conversely, bidirectional connectivity alterations were observed between Crus II and IFG, manifesting as enhanced effective connectivity from right cerebellar Crus II to left IFG (p = 0.023, uncorrected) and strengthened reciprocal connectivity from left IFG to Crus II (p = 0.004, FDR correction). Crucially, task-dependent connectivity analysis identified a positive association between IFG→Crus II connectivity strength and behavioral accuracy during incongruent trials in the athletic group (r = 0.39, p = 0.024, uncorrected), a relationship absent in controls (r = 0.09, p = 0.58, uncorrected).

Table 3.

Dynamic causal modeling analysis of cerebro-cerebellar effective connectivity during incongruent conditions between-group differences

Athlete-Control	t	df	p
Crus1-Crus2:	–0.92	92	0.361
Crus1-IFG	–2.21	92	0.029	*
Crus2-Crus1	–0.61	92	0.542
Crus2-IFG	2.32	92	0.023	*
IFG-Crus1	–1.93	92	0.057	.
IFG-Crus2	3.00	92	0.004	**

inferior frontal gyrus (IFG), * P < 0.05, ** P < 0.001

Discussion

The current study revealed distinct neurocognitive profiles between basketball athletes and non-athletes during Swimmy’s task performance. Behaviorally, athletes showed better conflict resolution accuracy in incongruent trials despite similar reaction times, indicating enhanced inhibitory control. Neuroimaging revealed athlete-specific activation patterns: heightened right lingual gyrus and cerebellar lobule VI during congruent trials, while incongruent trials triggered left precuneus activation and reduced right middle frontal gyrus engagement. Athletes also exhibited attenuated left inferior parietal lobule responses in incongruent > congruent contrasts. Notably, DCM identified that athletes exhibited reduced right Crus I→left IFG connectivity versus controls, contrasting with enhanced bidirectional Crus II-IFG coupling. Crucially, stronger IFG→Crus II connectivity positively predicted incongruent trial accuracy in athletes. This athlete-specific connectivity-behavior relationship contrasts with controls and suggests that long-term sport training may sculpt adaptive neural circuits that optimize cognitive control through cerebellar-frontal integration.

We employed the Swimmy task, which integrates spatially localized and global conflicts [21]. Critically, the cognitive demands of this paradigm—such as the quick resolution of directional conflicts under uncertainty—closely resemble decision-making processes triggered by basketball head-fakes [19]. Furthermore, Swimmy effectively models conflict dynamics while extending beyond sport-specific contexts, thereby uncovering domain-general neural signatures of cognitive control. This design offers advantages over the Stroop task [21] by eliminating linguistic influences and hemispheric lateralization biases. It also aptly models the tension between individual actions and team coordination inherent in basketball. During congruent trials, athletes showed increased activation in the right lingual gyrus and cerebellar lobule VI. The lingual gyrus is necessary for both basic and higher-level visual processing required for the orientation and direction of stimuli [31, 32]. The cerebellar lobule VI is strongly connected with core neurocognitive networks such as the default mode and executive control networks [33]. The results suggest enhanced perception of stimulus orientation and improved motor control in athletes under non-conflict conditions.

Under incongruent trials, athletes demonstrated stronger cuneus activation and reduced right inferior IFG engagement. The cuneus is implicated in processing fundamental visual features—including stimulus orientation, direction, and color [16, 17]—and in higher-order integration of these attributes. This dual functionality may facilitate superior conflict resolution through contextual cue integration. Conversely, diminished IFG activation—a region critical for the conflict control effect [34]—implies heightened cognitive efficiency in athletes [10]. Contrast analyses (incongruent > congruent) revealed attenuated activation in the left inferior parietal lobule (IPL), which is involved in conflict control [34], a hub for spatial integration, indicating more efficient neural resource allocation during conflict processing. Collectively, these findings suggest that athletes exhibit optimized visual information processing, streamlined cognitive control, and resource-efficient spatial integration during Swimmy task performance.

We used DCM to investigate the cerebellum’s role in the Stroop interference effect. DCM identified that athletes exhibited reduced right Crus I→left IFG connectivity versus controls, contrasting with enhanced bidirectional Crus II-IFG coupling. An ALE meta-analysis confirmed that the frontal cortex, especially IFG, played an important role in cognitive control in the Stroop task [34]. The theories of the universal cerebellar transform and dysmetria of thought, and recent studies have recontextualized the cerebellum as an integral node in the distributed neural circuits subserving all brain-based behaviors [4]. There is a triple representation (first = lobules VI/Crus I; second = lobules Crus II/VIIB; and third = lobules IX/X) of nonmotor/cognitive task activation in the cerebellum [7]. Crus I and Crus II, the first and second representations of cognition [7], are functional connector hubs in the cerebellum, respectively [33]. Crus I also plays an important role in attention allocation and execution functions, especially when it is necessary to suppress irrelevant information or adjust task strategies [7, 35]. Crus II also plays an important role in executing functions, especially when dynamically adjusting task strategies or processing complex information is required [36]. Fronto-cerebellar connectivity mediating cognitive processing speed [37] and is involved in conflict control [21]. Athletes exhibit a decrease in right Crus I → left IFG connectivity, which may indicate a reduction in sensory input and interference information. The enhanced bidirectional connectivity between Crus II and IFG in athletes, particularly the positive association of IFG→Crus II connectivity strength with incongruent-trial accuracy (r = 0.39, p = 0.024), suggests that long-term training supports cortico-cerebellar integration for cognitive regulation. Importantly, this neural-behavioral coupling was absent in controls (r = 0.09, p = 0.58), demonstrating that athletic expertise selectively shapes cerebellar-frontal circuits for conflict resolution. We propose that such enhancement reflects cerebellar-mediated tuning of cognitive control, whereas controls exhibit weaker cerebellar-cortical modulation disconnected from behavior. Specific changes in these brain regions’ connectivity indicate cerebellar neuroplasticity, enhancing cognitive efficiency.

Limitations

Our findings reveal distinct neurocognitive signatures in athletes during the spatial conflict (Swimmy) task, marked by improved conflict resolution accuracy and specific neural activation patterns. Despite these results, several limitations must be addressed. These include the cross-sectional design, which prevents definitive attribution of differences solely to training, and a sport-specific sampling bias (only basketball athletes). Future longitudinal studies with diverse athletic groups are necessary to separate training-related neuroplasticity from pre-existing individual differences. Additionally, the observed cerebellar-forebrain interactions are temporally integrated signals due to the inherent temporal limitations of fMRI, which shows averaged neural activity over seconds rather than milliseconds. To study the real-time dynamics of these circuits, techniques with higher temporal resolution are needed. Furthermore, future research could explore the similarities and differences in cognitive control between basketball athletes performing specialized tasks (head-fake) and general cognitive tasks (Swimmy).

Conclusions

In conclusion, long-term basketball training significantly impacts the Swimmy task, characterized by better conflict resolution accuracy in behavioral performance, and increased right lingual gyrus/cerebellar VI activation during congruent trials and left precuneus activation with reduced frontal involvement during incongruent trials. Athletes exhibited weaker Crus I→IFG connectivity but stronger bidirectional Crus II-IFG coupling, where IFG→Crus II strength positively predicted incongruent trial accuracy. These results indicate that long-term basketball training is associated with enhanced cognitive control performance, alongside differences in Crus I/Crus II-IFG circuit interactions linked to this performance. This work contributes to our understanding of neural patterns associated with long-term sport training, highlighting the cerebellum’s involvement in cognitive control. These findings may inform future research exploring the potential links between targeted training and cognitive enhancement.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

fMRI

functional magnetic resonance imaging

VOIs

voxels of interest

IFG

inferior frontal gyrus

DCM

Dynamic Causal Modeling

BA

Brodmann area

Author contributions

W. F., Y. Q., Y., W., X. L. and B. S collected the data, and Y. Q., Y. W., and X. D. analyzed the data, and W. F., Y. Q., and B. S. prepared Figs. 1, 2, 3 and 4. All authors reviewed the manuscript.

Funding

The study was supported by the Social Sciences and Humanities project of the Ministry of Education (23A10277004); 2022 Shanghai Philosophy and Social Science Planning Project (2022ZTY001); Key Lab of HumanPerformance, Shanghai University of sport(11DZ2261100) and the Research and Innovation Grant for Graduate Students, Shanghai University of Sport (YJSCX-2024-032).

Data availability

Data and analysis scripts are available upon request.

Declarations

Ethics approval and consent to participate

The study was approved by the Shanghai University of Sport Committee on Human Research (Project No. 102772023RT052). All participants provided written informed consent before taking part. The study adhered to the ethical principles for medical research involving human subjects outlined in the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.