Sports participation & childhood neurocognitive development

Abstract

Various psychosocial factors like collaboration inherent to team sports might provide a more dynamic environment for cognitive challenges that could foster enhanced neurocognitive development compared to individual sports. We investigated the impact of different organised sports on neurocognitive development in children (N = 11,878; aged 9–11) from the Adolescent Brain Cognitive Development (ABCD) study. Participants were classified into four categories based on their sports involvement at baseline and two years later: none, individual-based, team-based, or both. Cross-sectional and longitudinal analyses were conducted on 11 cognitive tests and neuroimaging metrics (i.e., resting-state functional connectivity and various grey matter (GM) and white matter (WM) measurements) between sport groups. A comparison between team and individual sports yielded no significant differences in cognitive measures at baseline and follow-up. Similarly, although WM microstructural differences were significant, the effect size was small. However, participation in any sport at baseline was associated with superior performance in various cognitive domains (i.e. inhibition, processing speed, and others), greater subcortical GM volume (i.e. cerebellum cortex, amygdala, hippocampus, and others), and whole-brain WM integrity compared to non-participants. Results suggest a positive association between organised sports participation, specifically individual and team-based sports, and neurocognitive development. However, further investigation is warranted to determine the nuanced effects of different sports on neurocognitive development.

Highlights

• •We studied the impact of different sport types on neurocognitive development.
• •Sports participation is associated with enhanced cognitive performance.
• •Sports participation demonstrated larger subcortical grey matter volume.
• •Sports participation demonstrated greater white matter integrity.
• •No significant neurocognitive differences of team over individual sports.

1. Introduction

Childhood is a critical period for brain development and the establishment of stable, long-lasting cognitive abilities (Gilmore et al., 2018). Therefore, it is critical to understand and promote factors involved in healthy childhood neurocognitive development. Modifiable lifestyle behaviours such as physical activity (PA) are one such factor shown to benefit healthy cognitive and brain development from childhood (Carson et al., 2016, Estévez-López et al., 2023). PA refers to any bodily movement produced by skeletal muscles that requires energy expenditure (World Health Organization, 2010). Within the broad spectrum of PA include exercise and sport. Exercise embodies “planned, structured, and repetitive bodily movement, through which the objective is to improve or maintain physical fitness” (World Health Organization, 2010). Organised sport can be considered a subset of exercise that is goal-directed, encompassing elements of rules, specific motor skills, competition, and interpersonal interactions (Khan et al., 2012)

It is well-established that PA helps regulate the production of various such as brain-derived neurotrophic factor and insulin-like growth factor 1, and downregulates inhibitory neurotransmitter ɣ-aminobutyric acid (Molteni et al., 2002). These biomolecular mechanisms work synergistically to promote neuroplasticity by inducing neurogenesis and cerebral vascularisation, facilitating nutrient and oxygen supply, ultimately enhancing long-term potentiation and brain development (Tomporowski et al., 2008). Mounting evidence has shown that regular engagement in PA and exercise is beneficial for cognitive and brain development in children and adolescents (Biddle and Asare, 2011, Carson et al., 2016, Chaddock-Heyman et al., 2018, Chaddock-Heyman et al., 2013, Sibley and Etnier, 2003). For example, studies have shown that higher levels of PA and aerobic fitness corresponds to greater grey (GM) and white matter (WM) volumes within brain areas critical for inhibitory control, higher-order cognition, and declarative memory, such as the dorsal striatum, frontoparietal area, and hippocampal regions (Chaddock et al., 2010, Chaddock-Heyman et al., 2014, Colcombe et al., 2006). Likewise, 9-months of PA in children improved microstructural integrity within the genu of the corpus callosum (Chaddock-Heyman et al., 2018). Functional magnetic resonance imaging (fMRI) studies have shown that children who engage in sports display greater activation during attentional control tasks within the prefrontal and parietal cortices, such as the bilateral middle frontal gyrus, supplementary motor area, anterior cingulate cortex (ACC), and precentral gyrus (Chaddock et al., 2012). Differences in brain regions, connectivity and their development are often considered to overlap with alterations to cognition (Reineberg et al., 2018, Schmithorst et al., 2005). Cross-sectional evidence demonstrates that the greater frequency of PA is associated with improved non-linguistic cognitive performance in domains such as executive functioning and inhibitory control, as well as in the language domain (Biddle and Asare, 2011, Carson et al., 2016, Chaddock et al., 2012, Chaddock-Heyman et al., 2013, Davis et al., 2011, Sibley and Etnier, 2003).

It can be postulated that the benefit on cognitive and brain development from sports participation could be attributed, in part, to an underlying increase in PA and aerobic fitness. However, the complexity of sports-specific coordinative motor skills, coupled with elements of rules and regulation, team-based engagement, and competition that are unique to sport may confer differential benefits on child cognitive and brain development. Moreover, sports are readily available in school-based and leisure settings, offering an opportunity for implementation pathways to enhance childhood cognitive and brain development. Indeed, previous meta-analysis and primary investigations showed that sport-based interventions improved executive function categories of working memory, inhibitory control, and cognitive flexibility in children and adolescents (Campos et al., 2024, Contreras-Osorio et al., 2021). Sports participation has also been associated with greater cortical thickness in motor and premotor areas (López-Vicente et al., 2017), plus hippocampal volume (Gorham et al., 2019). Taken together, sport participation has the potential to promote cognitive development in children, however more longitudinal data is needed to understand the neural mechanisms underpinning this relationship.

Unique to sport are the type of motor tasks, rules, and engagement placed on the child (Khan et al., 2012), which has potential to mediate the relationship between exercise, brain, and cognitive function to different extents (Tomporowski et al., 2008). Traditional conceptualisations of sport are often categorised into individual or team, where social interdependence and collaboration act as key characteristics distinguishing the two categories (Evans et al., 2012). Team sports entail athletes training and competing together, requiring continuous collaboration and interaction amongst teammates to achieve a common group-level objective (Evans et al., 2012, Landkammer et al., 2019). Children who participate in organised team sports may experience different experience-dependent changes in the brain compared to children who participate in individual-based sports, possible due to differences in social collaboration (López-Vicente et al., 2017). Although social competency appears early in life and has been implicated as an important factor associated with healthy neurocognitive development, the degree of its influence in shaping brain and cognitive development varies depending on the level of exposure to individual socialisation one receives (Champagne and Curley, 2005, Li and Shao, 2022). Data has shown that increased social play positively correlates with improved socio-cognitive functioning in rodents (Pellis and Pellis, 2007). Similarly, human population-based cohort research has found that childhood team sports participation is associated with a thinner cortex in several frontal regions linked to cognitive control, such as the orbitofrontal and middle frontal cortices, relative to individual sports (López-Vicente et al., 2017). This suggests that team sports could be associated with the maturation of regions linked to regulatory ability, typically indicated in normative data as a reduction in cortical thickness in these areas during childhood brain development (López-Vicente et al., 2017, Sowell et al., 2004). Additionally, childhood team sports participation in cross-sectional studies has been associated with greater hippocampal volume compared to individual sports, potentially demonstrating the benefits of team sports in enhancing memory encoding and emotional regulation (Gorham et al., 2019, Kunitoki et al., 2023). Preliminary evidence on cognitive functioning showed that childhood team sports participation is associated with superior executive functioning scores compared to those in self-paced/individual sports (De Waelle et al., 2021). In line with the traditionally theorised social collaborative differences between team and individual sports, children who participate in team sports might enjoy a more enriching, dynamic social environment offering greater opportunities for social interactions and building social competencies, which could be associated with better cognitive functioning and neurodevelopment, compared to participation solely in individual sports.

Despite the promising evidence for the relationship between sports participation, brain health and cognitive functioning, much of the previous evidence is derived from the benefits of general PA or exercise (Biddle and Asare, 2011, Carson et al., 2016, Chaddock-Heyman et al., 2018, Chaddock-Heyman et al., 2013, Sibley and Etnier, 2003), with most studies focusing on executive functioning (Carson et al., 2016, Chaddock-Heyman et al., 2014, Davis et al., 2011). Moreover, a recent umbrella review of randomised controlled trials (RCTs) highlighted uncertainty, variability, and low statistical power on the effect on exercise on cognition (Ciria et al., 2023), warranting a need for more longitudinal studies and well-designed RCTs. This also underscores the necessity to investigate in greater detail not only how participation in different sports could influence cognition, but also how various socio-environmental confounds could account for the variability in cognitive results. Certainly, several covariates have also shown in previous studies to affect childhood neurocognitive development. (1) Age has been implicated as an important determinant of variations in brain morphology and neural circuitry, corresponding to cognitive functioning differences (Casey et al., 2005, Casey et al., 2000, Sowell et al., 2004). (2) Sex differences are present during the brain maturational process, potentially influencing neurocognitive development (De Bellis, 2001). (3) PA levels vary between sports, and increased PA levels have been shown to facilitate cognitive functioning and neurological development (Tomporowski et al., 2008). (4) Disparities in racial and household income have also been associated with variations in cognitive achievements and brain development (Akhlaghipour and Assari, 2020, Farah et al., 2006). Taken together, there is a need to examine longitudinally, the relationship between sports participation, cognition, and underlying neural correlates, considering various sociodemographic variables.

Hence, the aim of this study was to employ robust statistical analysis to examine holistically, the effect of sport participation across multiple domains of cognition and explore potential underlying mechanisms through both structural (i.e. GM volume and WM architecture) and functional (i.e. resting-state functional connectivity (rs-FC)) MRI markers. We hypothesised that:

H1. Participation in organised sports, particularly team-based sports, would be associated with higher baseline cognitive scores in inhibition, processing speed, episodic memory, verbal learning, immediate and delayed recall, visuospatial processing, language comprehension, language decoding, and crystallised composite cognitive domains, compared to no sports or individual-based sports participation respectively.

H2. Participation in organised sports, specifically team-based sports, would be associated with a greater rate of improvement in the aforementioned cognitive domains from baseline to follow-up, compared to no sports or individual-based sports participation.

H3. At baseline, participation in organised sports, particularly team-based sports, would be associated with alterations in cortical thickness (CT), subcortical grey matter (subGM) volumes, decreased mean diffusivity (MD), increased fractional anisotropy (FA), and rs-FC measures when compared to no sports or individual-based sports participation respectively.

H4. Interval changes in MRI measures would show alterations in CT, subGM volume, MD, FA, and rs-FC during longitudinal participation in organised sports, particularly team-based sports, compared to no sports or individual-based sports participation.

2. Methods

2.1. Participants

Data from the National Institute of Mental Health National Data Archive (NDA) Adolescent Brain Cognitive Development (ABCD) study release 5.0 was used. ABCD is an ongoing longitudinal study that examines the brain and cognitive development of children aged 9–11 years into early adulthood (Auchter et al., 2018). The data used for this study consisted of in-person assessments of an initial sample of 11,878 children (aged 9–11) at baseline (2016–2018) and follow-up (2018–2021) two years later, from 21 testing sites across the US (Anokhin et al., 2022). Ethics clearance was obtained from the relevant institutional and ethics committees. Parental consent and assent were obtained from minors participating in the study. All data collected were subjected to quality control checks by the ABCD Data Analysis and Informatics Core (DAIRC) (Auchter et al., 2018).

2.2. Organised sports classification

Organised sports participation was assessed using the Sport and Activities Involvement Questionnaire (SAIQ) at baseline and follow-up (Ács et al., 2021). The SAIQ has an internal consistency of Cronbach’s α = 0.65 (Ács et al., 2021). In the ABCD study, caregivers were initially asked to indicate “yes” or “no” regarding their child’s lifetime involvement from a list of activities provided. They were then asked to respond “yes” or “no” if the activity was part of an organised programme in or outside of school (Hoffmann et al., 2022). Children who participated in organised sports both in and outside school were included (Hoffmann et al., 2022). The remaining subjects were classified into four organised sports types at baseline and follow-up (Table 1) (Hoffmann et al., 2022):

• 1.No organised sports participation: Activities that were (1) reflective of pure PA rather than organised sports (e.g. yoga); (2) organised non-PA (e.g. chess); and (3) those who indicated previous sports involvement but had no participation 12-months prior to data collection.
• 2.Individual-based sports participation: Organised PAs where participants compete individually against other athletes for personal achievements.
• 3.Team-based sports participation: Organised PAs where subjects work together in collaboration towards a common goal.
• 4.Both individual and team-based sports participation.

Table 1.

Organised Sport Classification.

S/N	Team-based sports	Individual-based sports
1.	Soccer	Ballet/Dance
2.	Baseball	Gymnastics
3.	Basketball	Horseback Riding
4.	Field Hockey	Martial Arts
5.	Football	Wrestling
6.	Ice hockey	Swimming/Water Polo
7.	Lacrosse	Tennis
8.	Rugby	Track/Running/Cross-country
9.	Volleyball	-

Swimming and water polo were combined into one item in the SAIQ, resulting in two sports with significantly different levels of interdependence classified as one item. However, since water polo failed to make the list of core youth sports in the US, it was assumed that children who played water polo also participated in swimming. As a result, water polo and swimming were combined into one sports group (Hoffmann et al., 2022).

2.3. Cognitive measures

Cognitive tasks were electronically administered using an iPad in a supervised laboratory setting, and only available at both baseline and follow-up (Luciana et al., 2018). Uncorrected standardised scores from 11 cognitive tests were utilised, as corrected scores would rule out any meaningful analyses of longitudinal change. A summary of the cognition assessed for each task is presented in Table 2. For a comprehensive description of the ABCD cognitive battery, refer to Luciana et al. (2018).

Table 2.

ABCD Cognitive Measures.

S/N	Cognitive Measure	Cognitive Process
1.	NIH Toolbox Picture Vocabulary task (TPVT)	Language Comprehension
2.	NIH Toolbox Flanker (TFT)	Inhibition
3.	NIH Toolbox Pattern Comparison Processing Speed Test (TPCPST)	Processing Speed
4.	NIH Toolbox Picture Sequence Memory Test (TPSMT)	Episodic Memory
5.	NIH Toolbox Oral Reading Recognition Task (TORRT)	Language Decoding
6.	NIH Crystallised Composite	Crystallised Cognition
7.	Rey Auditory Verbal Learning Task (RAVLT) Trial V	Verbal Learning
8.	RAVLT Trial VI	Immediate Recall
9.	RAVLT Trial VII	Delayed Recall
10.	Little Man Task (LMT) Proportion Correct	Visuospatial Processing (Mental Rotation)
11.	LMT Reaction Time (RT) Correct

2.4. Neuroimaging metrics

A standard ABCD MRI scanning session during baseline and follow-up sessions consisted of sMRI, resting-state fMRI (rs-fMRI), and diffusion tensor imaging (DTI) sequences (Hagler et al., 2019). MRI data were acquired and pre-processed in accordance with the ABCD pipeline (Hallquist et al., 2013).

sMRI metrics were generated by the segmentation of T1-weighted images onto a cortical surface atlas using FreeSurfer v5.3. This includes white matter segmentation, initial mesh creation, topological correction, surface optimisation, and nonlinear registration (Hagler et al., 2019). CT measurements were parcellated and labelled using the Destrieux atlas (Destrieux et al., 2010), and subGM volumetric measurements using FreeSurfer’s automated brain segmentation (ASEG) atlas (Fischl, 2012, Fischl et al., 2002).

rs-fMRI pre-processed time-course measures were band-pass filtered between 0.009 and 0.08 Hz (Hallquist et al., 2013), before being sampled onto the cortical surface, followed by motion censoring, which involved exclusion of time points with framewise displacement (FD) greater than 0.2 mm and/or standard deviation (SD) values three times the median absolute deviation away from the median SD, from variance correlation calculations (Hagler et al., 2019). The resting-state functional connectome was constructed using Gordon Atlas (Gordon et al., 2016). The edges were averaged within their respective networks, resulting in a 12 × 12 functional connectivity (FC) matrix (Gordon et al., 2016).

Microstructural tissue properties related to DTI, such as MD and FA of WM tracts, were extracted and labelled using AtlasTrack (Hagler et al., 2019). Voxels which primarily contained GM and/or cerebrospinal fluid volumes were excluded from subsequent analyses (Hagler et al., 2019).

2.5. Covariates

Additional variables of (1) age (in months); (2) sex (ref: Male); (3) race (ref: White, Black, Asian, Others); (4) family income (0 = don’t know/refuse to answer/missing data; 1 = < USD 5,000; 2 = USD 5,000–11,999; 3 = USD 12,000–15,999; 4 = USD 16,000–24,999; 5 = USD 25,000–34,999; 6 = USD 35000–49,999; 7 = USD 50,000–74,999; 8 = USD 75000–99,999; 9 = USD 100,000–199,999; 10 = ≥ USD 200,000); (5) family ID; (6) testing site (21 sites); and (7) PA levels acquired using Fitbit (weekly average metabolic equivalents (METs/min) and weekly average step count), were also collected at baseline and follow-up (Heeringa and Berglund, 2020).

3. Statistical analyses

Two separate analyses were conducted: (1) cross-sectional at baseline and (2) longitudinal. For longitudinal analysis, only participants who were classified under the same sports category at baseline and follow-up were retained. Furthermore, interval changes in (1) MRI metrics, (2) age, and (3) family income were calculated by subtracting the baseline from the follow-up values for each participant (Kaltenhauser et al., 2023). To explore the differences in cognitive and brain measures between different organised sport types during baseline and longitudinal analyses, six different pairwise comparisons between the four different sports categories were conducted. As there were no significant differences in PA levels between individual, team and both sport types, it was excluded from subsequent analyses as a potential covariate (see Table S2).

3.1. Cognitive measures

Mass univariate mixed effect models (MUMMs) with false discovery rate (FDR) correction were performed using the LmerTest v3.1.3 package to examine the associations between each organised sports pairwise comparison across all cognitive tasks. As shown in Eq. 1, each model during baseline analyses included fixed effect variables of the organised sport types being pairwise compared, while adjusting for the covariates of age, sex, race, and family income. Family ID nested within testing site was included as a random effect. This allowed the model to concurrently account for the oversampling of twins in the ABCD study, along with the clustering of data within the 21 testing sites (Heeringa and Berglund, 2020). For longitudinal analysis (Eq. 2), the model was similar to that used during the baseline analyses. Additional variables of the main effect of time (ref: Baseline), sport type × time product term measuring differences in cognitive scores between sport types across time, and an additional random effect of subject ID, to account for repeated sampling, were included. Altogether, 11 cognitive tests were analysed at baseline and longitudinally. False discovery rate (FDR) correction was applied to control for the multiple cognitive tests analysed.

$$Y_i={\beta }_0+{\beta }_1\mathit{sport}+{\beta }_2\mathit{age}+{\beta }_3\mathit{sex}+{\beta }_{4}R1+{\beta }_{5}R2+{\beta }_{6}R3+{\beta }_7\mathit{income}+u_{\mathit{site}\left[j\right]}+v_{\mathit{family\; ID}\left[k,|,j\right]}+{ϵ}_i$$ (1)

$$Y_i={\beta }_0+{\beta }_1\mathit{sport}+{\beta }_2\mathit{time}+{\beta }_3\left(\mathit{sport}*\mathit{time}\right)+{\beta }_4\mathit{age}+{\beta }_5\mathit{sex}+{\beta }_{6}R1+{\beta }_{7}R2+{\beta }_{8}R3+{\beta }_9\mathit{income}+u_{\mathit{site}\left[j\right]}+v_{\mathit{family\; ID}\left[k|,j\right]}+z_{\mathit{subject\; ID}\left[l\right]}+{ϵ}_i$$ (2)

Where,

Y _i = i ^th cognitive measure

sport = dummy coded pairwise compared sport groups (e.g. None=0, Team=1)

time = dummy coded timepoint (ref: Baseline)

age = age in months

sex = dummy coded sex variable (ref: Males)

R1 - R3 = dummy coded race variables (ref: White; R1 = Black; R2 = Asian; R3 = Others)

β _n = fixed effect coefficient of the n ^th variable

u _site[j] = random effect of j ^th site

v _{family ID[k|j]} = random effect of k ^th family ID nested within j ^th site

z _{subject ID [l]} = random effect of ID of the l ^th subject

ϵ = residual error

3.2. Neuroimaging measures

As the ABCD study acquired multimodal MRI data from 29 scanners across 21 sites (Casey et al., 2018), variance attributable to scanner-induced effects across scanner models can be a source of bias, potentially leading to a reduction in statistical power (Dudley et al., 2023). Therefore, MRI measurements at each time point were subjected to ComBat harmonisation using the neuroCombat v1.0.13 package before subsequent statistical analyses (Fortin et al., 2017).

3.2.1. Resting-state functional MRI

For baseline and longitudinal analyses, separate mixed-effect implemented network-based statistics (NBS) were conducted using the NBR v0.1.5 package. This identified network components which exhibited statistically significant differences when comparing different sport types (Gracia-Tabuenca and Alcauter, 2020). Partial correlation analyses were carried out between each pairwise compared sport types, and edges within the FC matrix, while adjusting for age, sex, race, and income, and including the random effect of family ID. NBS was then implemented to correct for multiple comparisons across the FC edges using a conservative p-value threshold of 0.001, defined a priori, to minimise false-positive rates.

Significant FC edges within each network were subsequently clustered and summed to obtain their respective network strengths. These network strengths were then tested against a null distribution generated by 1000 permutations, where the analyses were repeated with the subject ID labels of the FC data randomly shuffled (Gracia-Tabuenca and Alcauter, 2020).

3.2.2. Structural MRI

To account for the multiple parcels stemming from the Destrieux and ASEG atlases representing the CT and subGM metrics respectively, MUMMs with FDR correction were employed to analyse each of the six pairwise comparisons between sport types at baseline and longitudinally. The fixed effects of the MUMMs for sMRI data analysis were identical to those mentioned above for cognitive measure analysis. The model included the variables of sport types being pairwise compared, with each regression model adjusting for age, sex, race, and income. However, only family ID was included as a random effect (Eq. 3).

$$Y_i={\beta }_0+{\beta }_1\mathit{sport}+{\beta }_2\mathit{age}+{\beta }_3\mathit{sex}+{\beta }_{4}R1+{\beta }_{5}R2+{\beta }_{6}R3+{\beta }_7\mathit{income}+v_{\mathit{family\; ID}[j]}+{ϵ}_i$$ (3)

Where,

Y _i = i ^th parcel measure

sport = dummy coded pairwise compared sport groups (e.g. None=0, Team=1)

age = age in months

sex = dummy coded sex variable (ref: Males)

R1 - R3 = dummy coded race variables (ref: White; R1 = Black; R2 = Asian; R3 = Others)

β _n = fixed effect coefficient of the n ^th variable

v _{family ID[j]} = random effect of j ^th family ID

ϵ = residual error

3.2.3. Diffusion tensor imaging

Similar to sMRI analysis, MUMMs were used to analyse each of the six pairwise compared sport types at baseline and the longitudinal change of MD and FA measurements (Eq. 3). There were 42 DTI parcels for both MD and FA measurements. 37 represented unique ROIs, with the remaining five containing independent measurements of (1) whole brain fibres; (2) all fibres in the left hemisphere; (3) all fibres in the right hemisphere; (4) all fibres in the left hemisphere excluding the corpus callosum; and (5) all fibres in the right hemisphere excluding the corpus callosum. Hence, FDR corrections were applied separately to account for multiple comparisons between the (1) unique ROIs; (2) left versus right hemispheres; and (3) left versus right hemispheres excluding the corpus callosum.

3.3. General assumptions

All mixed-effect models were estimated using restricted maximum likelihood (REML). Owing to six pairwise comparisons between sport types, the alpha value of significance for each model was set to.05/6 (α ≤.0083). For all MUMMs, pairwise comparisons of sport types and their associations with measures were reported using standardised regression coefficients. Regression diagnostics were applied to each mixed-effect model to determine if there were any violations of the assumptions of linearity, normality of residuals, and homogeneity of residuals. All statistical analyses were conducted using R statistical software version 4.1.0.

4. Results

4.1. Demographics

Owing to the presence of invalid/missing cases for the different types of data, a number of subjects had to be excluded from the different analyses, as shown in Fig. 1. Breakdown of participant demographics for each sport type during cognitive and neurological baseline and longitudinal analyses are shown in Table 3 (see Table S3).

Fig. 1.

Study flowchart.

Table 3.

Descriptives.

Cognitive Measures
Baseline
Characteristic	Total (N = 11285)	None (N = 3279)	Individual (N = 2667)	Team (N = 2799)	Both (N = 2540)
Age [months], mean (SD)	118.96 (7.51)	118.71 (7.49)	118.90 (7.66)	119.43 (7.49)	118.85 (7.37)
Sex, N
Females	5398	1680	1847	727	1144
Males	5887	1599	820	2072	1396
Race, N
White	5886	1241	1448	1592	1605
Black	1694	769	327	360	238
Asian	245	50	93	39	63
Others	3460	1219	799	808	634
Income, mean (SD)	6.62 (3.08)	5.24 (3.15)	7.06 (2.81)	6.90 (2.97)	7.62 (2.74)
Longitudinal
Characteristic	Total (N = 13924)	None (N = 10468)	Individual (N = 2080)	Team (N = 1212)	Both (N = 164)
Age [months], mean (SD)	131.39 (14.74)	131.31 (14.73)	131.41 (14.72)	132.09 (14.89)	130.60 (14.81)
Sex, N
Females	3612	2652	758	158	44
Males	3350	2582	282	448	38
Race, N
White	3170	2104	642	368	56
Black	1262	1144	66	50	2
Asian	154	72	72	4	6
Others	2376	1914	260	184	18
Income, mean (SD)	6.14 (3.15)	5.59 (3.11)	7.82 (2.58)	7.68 (2.82)	8.56 (2.28)
Structural MRI
Baseline
Characteristic	Total (N = 11721)	None (N = 3382)	Individual (N = 2780)	Team (N = 2909)	Both (N = 2650)
Age [months], mean (SD)	118.97 (7.50)	118.71 (7.50)	118.90 (7.66)	119.42 (7.46)	118.88 (7.35)
Sex, N
Females	5595	1723	1929	751	1192
Males	6126	1659	851	2158	1458
Race, N
White	6122	1284	1510	1655	1673
Black	1731	773	333	373	252
Asian	251	51	97	38	65
Others	3617	1274	840	843	660
Income, mean (SD)	6.62 (3.07)	5.24 (3.14)	7.05 (2.81)	6.91 (2.95)	7.61 (2.74)
Longitudinal Interval Change
Characteristic	Total (N = 2889)	None (N = 2172)	Individual (N = 431)	Team (N = 256)	Both (N = 30)
Age Interval [months], mean (SD)	24.73 (2.70)	24.67 (2.69)	24.76 (2.59)	25.18 (2.87)	25.00 (2.89)
Sex, N
Females	1462	1081	310	54	17
Males	1427	1091	121	202	13
Race, N
White	1333	884	273	154	22
Black	509	461	27	21	0
Asian	58	27	27	2	2
Others	989	800	104	79	6
Income Interval, mean (SD)	0.30 (2.31)	0.30 (2.35)	0.28 (2.50)	0.28 (1.64)	0.13 (0.62)
Resting-State Functional MRI
Baseline
Characteristic	Total (N = 11218)	None (N = 3217)	Individual (N = 2672)	Team (N = 2794)	Both (N = 2535)
Age [months], mean (SD)	119.03 (7.50)	118.77 (7.51)	118.93 (7.65)	119.50 (7.45)	118.94 (7.34)
Sex, N
Females	5383	1655	1862	724	1142
Males	5835	1562	810	2070	1393
Race, N
White	5893	1231	1458	1599	1605
Black	1638	721	320	357	240
Asian	237	47	95	35	60
Others	3450	1218	799	803	630
Income, mean (SD)	6.65 (3.05)	5.27 (3.13)	7.05 (2.81)	6.94 (2.93)	7.64 (2.72)
Longitudinal Interval Change
Characteristic	Total (N = 2718)	None (N = 2030)	Individual (N = 410)	Team (N = 249)	Both (N = 29)
Age Interval [months], mean (SD)	24.73 (2.70)	24.67 (2.71)	24.76 (2.56)	25.15 (2.85)	25.03 (2.93)
Sex, N
Females	1385	1027	291	51	16
Males	1333	1003	119	198	13
Race, N
White	1273	836	264	152	21
Black	474	429	24	21	0
Asian	54	24	26	2	2
Others	917	741	96	74	6
Income Interval, mean (SD)	0.28 (2.29)	0.28 (2.33)	0.25 (2.51)	0.27 (1.64)	0.14 (0.64)
Diffusion Tensor Imaging
Baseline
Characteristic	Total (N = 11095)	None (N = 3163)	Individual (N = 2657)	Team (N = 2751)	Both (N = 2524)
Age [months], mean (SD)	119.04 (7.50)	118.79 (7.50)	118.95 (7.67)	119.53 (7.47)	118.91 (7.34)
Sex, N
Females	5330	1632	1846	721	1131
Males	5765	1531	811	2030	1393
Race, N
White	5852	1216	1449	1590	1602
Black	1609	703	322	346	238
Asian	232	45	93	35	59
Others	3402	1199	793	786	625
Income, mean (SD)	6.66 (3.06)	5.28 (3.14)	7.07 (2.82)	6.95 (2.93)	7.64 (2.71)
Longitudinal Interval Change
Characteristic	Total (N = 2664)	None (N = 1982)	Individual (N = 411)	Team (N = 246)	Both (N = 25)
Age Interval [months], mean (SD)	24.75 (2.71)	24.68 (2.71)	24.79 (2.58)	25.21 (2.87)	25.20 (3.01)
Sex, N
Females	1366	1008	294	52	12
Males	1298	974	117	194	13
Race, N
White	1251	819	265	150	17
Black	462	417	25	20	0
Asian	52	23	25	2	2
Others	899	723	96	74	6
Income Interval, mean (SD)	0.28 (2.29)	0.28 (2.32)	0.27 (2.55)	0.26 (1.55)	0.00 (0.50)

Note. N = number of observations; SD = standard deviation.

4.2. Cognitive measures

Pairwise comparison at baseline revealed that participation in individual sport types, relative to no sports participation, was positively associated with all cognitive measures except the Little Man Task (RT correct) and Pattern Comparison Processing Speed Test (TPCPST) (Fig. 2) (see Table S4a). Likewise, participation in team-based sports, compared to no sports participation, was significantly and positively associated with the Flanker Test (TFT), TPCPST, Picture Sequence Memory Test (TPSMT), Oral Reading Recognition Task (TORRT), Rey Auditory Verbal Learning Task (RAVLT) Trials V to VII, and LMT (proportion correct) measures but negatively associated with the Picture Vocabulary Task (TPVT) (Fig. 2) (see Table S4b). Participation in both sports relative to none, was positively associated with all cognitive measures, except TPVT, crystallised composite score, and LMT RT correct scores (Fig. 2). Participation in both sports relative to no sports was positively associated with all cognitive measures, except TPVT, crystallised composite score, and LMT RT correct scores (see Table S4c).

Fig. 2.

Baseline analysis of cognitive measures. TPVT = Toolbox Picture Vocabulary Task; TORRT = Toolbox Roal Reading Recognition Task; Cryst_Comp = Crystallised Composite; RAVLT_V to VII= Rey Auditory Verbal Learning Task Trials V to VII; TFT = Toolbox Flanker Task; TPCPST = Tollbox Pattern Comparison Processing Speed Test; TPSMT = Toolbox Picture Sequence Memory Task; LMT_Prop_Correct = Little Man Task Proportion Correct; LMT_RT_Correct = Little Man Task Reaction Time Correct. Values to the right of 0 on the x-axis indicate greater performance. The error bars represent confidence intervals of each respective sport type comparison calculated using Benjamini-Hochberg procedure to account for FDR correction.

On the other hand, participation in team-based sports was negatively associated with only TPVT and crystallised composite scores relative to individual sports type (Fig. 2) (see Table S4d). Similar patterns emerged when comparing both relative to individual sports types (Fig. 2) (see Figure S4e). There were no significant associations of both sports type relative to team with cognitive performance (Fig. 2) (see Table S4f).

Comparison of individual sports, relative to no sports participation, indicated a main effect of time across all measures, with improvement in performance related to cognitive processing (i.e. TFT, TPCPST, TPSMT, and LMT) over time in all subjects (Fig. 3) (see Table S5a). However, there was a significant decrease in performance on tasks associated with language processing (i.e. TPVT, TORRT, crystallised composite scores, and RAVLT trials V–VII) for all participants (Fig. 3) (see Table S5a). The main effect of sports type results indicated that participation in individual sports type relative to none was also positively associated with all cognitive measures except TPCPST (Fig. 3) (see Table S5a). Only the variable LMT correct RT showed a significant sports type × time interaction effect (Fig. 3) (see Table S5a).

Fig. 3.

Longitudinal analysis of cognitive measures. TPVT = Toolbox Picture Vocabulary Task; TORRT = Toolbox Roal Reading Recognition Task; Cryst_Comp = Crystallised Composite; RAVLT_V to VII= Rey Auditory Verbal Learning Task Trials V to VII; TFT = Toolbox Flanker Task; TPCPST = Tollbox Pattern Comparison Processing Speed Test; TPSMT = Toolbox Picture Sequence Memory Task; LMT_Prop_Correct = Little Man Task Proportion Correct; LMT_RT_Correct = Little Man Task Reaction Time Correct. Values to the right of 0 on the x-axis indicate greater performance. The error bars represent confidence intervals of each respective sport type comparison calculated using Benjamini-Hochberg procedure to account for FDR correction.

Similarly, results on the main effect of time highlighted that children who participated in team sports and those who did not participate in any organised sports showed significant improvements in tasks associated with non-linguistic cognitive functioning (i.e. TFT, TPCPST, TPSMT, and LMT) but showed a significant decrease in performance for tasks associated with language processing (Fig. 3) (see Table S5b). However, results on the main effect of sport type demonstrated that participation in team sports, relative to no sports, was negatively associated with cognitive measures of TPVT, RAVLT trial V, and LMT proportion correct (Fig. 3) (see Table S5b). No significant sports type × time interaction effects were observed for any of the measures (Fig. 3) (see Table S5b).

The results of the main effect of time showed significant performance improvements across all cognitive measures, with the exception of TORRT, when comparing both sports to no sports participation (Fig. 3) (see Table S5c). Results regarding the main effect of sports type highlighted that participation in both sports, relative to no sports participation, was significantly and positively associated with LMT accuracy (Fig. 3) (see Table S5c). No significant sports type × time interaction effects were observed (Fig. 3) (see Table S5c).

There were significant and positive improvements in TPCPST, TPSMT, LMT proportion correct, and LMT RT correct task performance across time for all participants when comparing team sport to individual sport types (Fig. 3) (see Table S5d). Team sports participation, relative to individual sports, was significantly and negatively associated with TPVT and crystallised composite scores (Fig. 3) (see Table S5d). The sports type × time interaction effects across all cognitive measures were not significant (Fig. 3) (see Table S5d).

Significant and positive improvements in TPCPST, LMT proportion correct, and LMT RT correct scores were observed for all participants across time when comparing both sports to individual sports participation (Fig. 3) (see Table S5e). However, participation in both sports, relative to individual sports, was not significantly associated with any cognitive measure (see Table S5e). There were also no significant sports type × time interaction effects across any of the cognitive measures (Fig. 3) (see Table S5e).

Likewise, when comparing team sports to both sports, all participants only significantly improved in LMT RT correct performance over time (Fig. 3) (see Table S5f). Participation in both sports, relative to team sports, was not significantly associated with any cognitive measure (Fig. 3) (see Table S5f). There were also no significant sports type × time interaction effects across any of the cognitive measures (Fig. 3) (see Table S5f).

4.3. Neuroimaging outcomes

Five networks were initially identified during cross-sectional NBS analyses at baseline. However, these networks were not statistically significant (see Table S6). This suggests that there were no FC networks that were significantly different between children who participated in the various types of sports. Similarly, analyses of CT measurements did not identify any significant differences in all sports type comparisons at baseline and longitudinally (see Tables S7 & S8).

4.3.1. Subcortical grey matter

As shown in Fig. 4, baseline comparison of children who participated in team-based sports, as compared to no sports participation, showed significantly greater subcortical grey matter volume in the (1) bilateral putamen; (2) bilateral pallidum; (3) bilateral accumbens area; (4) left amygdala; (5) left cerebellum cortex; and (6) right ventral diencephalon. Participation in both types of sports at baseline, relative to no sports participation at baseline, showed significantly greater subcortical grey matter volume in the (1) bilateral cerebellum cortex; (2) bilateral thalamus proper; (3) bilateral putamen; (4) brain stem; (5) bilateral hippocampus; (6) bilateral amygdala; (7) bilateral accumbens area; (8) bilateral ventral diencephalon; and (9) right pallidum (Fig. 4). There were no significant differences between the remaining sports types compared cross-sectionally at baseline and longitudinally (see Tables S9 & S10).

Fig. 4.

Baseline subGM associations during none vs team and both sport type comparisons. The error bars represent confidence intervals of each respective sport type comparison calculated using Benjamini-Hochberg procedure to account for FDR correction.

4.3.2. Mean diffusivity

At baseline, children who participated in team-based sports exhibited significant negative MD associations across various white matter regions compared with those who engaged in only individual-based sports. Specifically, negative associations were observed in the (1) whole brain fibres; (2) all fibres in the left hemisphere; (3) corpus callosum; (4) bilateral fornix (excluding fimbria); and (5) left fornix (including fimbria) (Fig. 5). There were no significant differences between the remaining pairwise comparisons of sports types at baseline and longitudinally (see Tables S11 & S12).

Fig. 5.

Baseline MD associations during individual vs team sport type comparison (ref: individual). The error bars represent confidence intervals of each respective sport type comparison calculated using Benjamini-Hochberg procedure to account for FDR correction.

4.3.3. Fractional anisotropy

At baseline, children who engaged in team-based sports showed significantly greater FA associations in (1) whole brain fibres; (2) all fibres within the left hemisphere; and (3) corpus callosum (Fig. 6A) than those who did not participate in sports. Team-based sports participation was significantly associated with greater FA within (1) whole brain fibres; (2) left hemisphere; (3) left anterior thalamus; (4) corpus callosum; (5) forceps minor; and (6) right fornix (excluding fimbria), compared to those who only participated in individual-based sports (Fig. 6B). On the other hand, analyses of FA longitudinal change demonstrated that participation in team-based sports was negatively associated with interval changes in FA within the right corticospinal region, compared to individual-based sports (Fig. 6C) (see Tables S13 & S14).

Fig. 6.

FA associations during baseline (A) none vs team; (B) individual vs team; and (C) longitudinal individual vs team comparisons. The error bars represent confidence intervals of each respective sport type comparison calculated using Benjamini-Hochberg procedure to account for FDR correction.

5. Discussion

The primary aim of this study was to evaluate the cross-sectional and longitudinal effects of different types of organised sports participation on neurocognitive outcomes in children. Results indicate that children who participated in any form of sports displayed significantly greater cognitive performance at baseline and two years later. Additionally, those who participated in any sport had greater subGM in several regions (i.e. cerebellum cortex, hippocampus, and others). They also demonstrated significantly improved whole-brain WM microstructural integrity. However, there were no significant differences between team and individual sports in terms of cognitive performance at baseline or follow-up, except for a marginal improvement in FA observed within the right corticospinal tract in individuals who participated in individual sports during longitudinal analysis. No significant differences in rs-fMRI and CT measurements were found during baseline and longitudinal analyses.

5.1. Cognitive outcomes

As expected, children who participated in any form of sport showed significantly better performance in most cognitive measures when assessed at baseline compared to those who did not engage in any organised sports. This was supported by longitudinal results, where sports participants demonstrated significantly better performance on most cognitive measures over time. This is consistent with current literature indicating that increased PA as a potential factor underlying sports participation is associated with significant beneficial effects on cognitive developmental outcomes (Carson et al., 2016, Sibley and Etnier, 2003). Exercise has also shown to improve attentional, interference control and working memory task performance in children (Chaddock et al., 2010, Chaddock-Heyman et al., 2013). Furthermore, physically fitter children demonstrated smaller error-related negativity amplitudes in event-related potentials, indicating the need for fewer neural resources during fast response speed tasks and more efficient neural processing (Hillman et al., 2009, Pontifex et al., 2011). This suggests that superior physical fitness, which may underpin habitual sports participation, may be associated with improved cognitive performance, attributable to the capacity for more effective allocation of neural resources during mentally demanding tasks, compared to non-participants (Chaddock-Heyman et al., 2014). Overall, this study reaffirms existing literature that children who do not participate in organised sports tend to perform worse on cognitive control tasks (Campos et al., 2024, Contreras-Osorio et al., 2021).

This study also found that team sports participation did not yield significantly superior performance in various cognitive test scores at baseline and longitudinally, compared to individual sports. This finding contradicts our initial hypothesis. Although there is evidence regarding the benefits of general sports participation, the literature on the effects of various sport types on neurocognitive development is inconsistent. Research has demonstrated that children participating in team sports outperform individual sports in comprehensive cognitive test batteries (De Waelle et al., 2021). These studies suggest that team sports involvement may enhance cognitive flexibility in processing real-time cues from teammates, such as inhibiting planned actions that might not be the best course of action, and communicating effectively (De Waelle et al., 2021, Diamond and Ling, 2016, Pesce, 2012). The dynamic social collaborative environment inherent in team sports may provide additional cognitive challenges, ultimately improving executive functioning and fostering sustained improvements in cognitive development. Conversely, no significant benefits stemming from team over individual sports were observed. Giordano et al. (2021) demonstrated that children practising martial arts, an individual sport, performed significantly better on similar executive functioning tasks compared to team sports. Additionally, it has been shown that the sport type young adults participate in does not significantly predict cognitive test performance (Chakraborty et al., 2023).

It is commonly theorised that the differences in neurocognitive development resulting from team or individual sports can be attributed to differences in their levels of social collaboration (Eime et al., 2013). Hence, it could be conjectured that instead of focusing on the social collaborative element of team sports as the benchmark in promoting cognitive development, other psychosocial elements, such as perceived and actual competence, level of competitiveness, and level of parental encouragement and involvement, that various sports provide for children may have a greater influence in mediating the effect of sports type on cognitive development (Choi et al., 2014, Eime et al., 2013, Giordano et al., 2021, Smith, 2003). However, such arguments should be considered speculative, as there were no measurements of the quantity of social interaction between individual and team sports. Additionally, no study has investigated the relationship between different sports and their influence on cognitive development. Future research designed specifically to elucidate any potential psychosocial mechanisms associated with various sport types and their influence on cognitive development is necessary.

Another possible explanation for the differing results of this study from previous research could be due to variations in study design. Past studies investigating the effect of team versus individual sports on cognitive functioning have differed in the number and type of sports classified as either team or individual sport types (De Waelle et al., 2021, Giordano et al., 2021). Additionally, varying cognitive tasks employed across studies may have offered varying levels of cognitive challenge and/or measured cognitive sub-domains which are different from this study (De Waelle et al., 2021, Giordano et al., 2021). The discrepancies in task selection and study design may account for the contrasting findings.

Another notable finding during secondary analysis was the performance decline in language-specific tests for all participants, even after accounting for age-induced differences. One possible explanation is that alternative versions of these tests are administered at baseline and follow-up (Anokhin et al., 2022). For example, RAVLT Form 1 was administered at baseline and Form 5 during follow-up, although findings reported that Form 1 was marginally easier than Form 5 (Hawkins et al., 2004). This inference was drawn from adult samples, and the difficulty might be more pronounced when alternative RAVLT versions are administered to children, resulting in a “reverse practice effect” (Anokhin et al., 2022). Moreover, the TPVT and TORRT administration protocols utilised computerised adaptive testing, with items calibrated using item response theory. This could result in different variations of the same test administered at baseline and follow-up, causing a mismatch of difficulty levels between sessions and negatively affecting performance (Luciana et al., 2018). Any longitudinal findings involving language processing measures should be interpreted with caution (Anokhin et al., 2022).

5.2. Neuroimaging outcomes

Our results did not demonstrate any significant rs-FC differences between different sport types during baseline and longitudinal analyses, contrary to similar fMRI studies. Studies have shown that increased PA can enhance FC between the anterior hippocampus and frontal brain regions, which are associated with memory and cognitive control (Esteban-Cornejo et al., 2021, Voss et al., 2011). Furthermore, increased exercise could facilitate greater refinement of other networks over time, characterised by reduced between-network synchrony and increased within-network synchrony, reflecting a more coherent and specialised pattern of resting-state synchrony. Such changes could indicate enhanced flexibility in adapting to cognitive demands with different networks able to act independently without mutual internetwork interference (Krafft et al., 2014, Valkenborghs et al., 2019).

However, rs-fMRI studies exploring the effects of childhood organised sports participation on FC are limited. Most evidence stems from PA and/or exercise-related task-based fMRI (Chaddock et al., 2010, Chaddock-Heyman et al., 2013, Davis et al., 2011), typically conducted on children with underlying health conditions (Esteban-Cornejo et al., 2021, Krafft et al., 2014). Furthermore, no studies have directly compared FC differences between team and individual sports. Therefore, no definitive conclusion can be drawn from the current null results, as this study is the first to explore how different sport types may elicit different FC patterns. It could be also postulated that the non-significant results may be due to the use of a 12 × 12 FC atlas, which might not provide a resolution fine enough to elucidate significant FC differences between sport types longitudinally. Additionally, rs-fMRI’s weaker predictive power when associated with cognitive measures, might further explain why the current rs-FC results do not correspond to cognitive results (Zhao et al., 2023). This suggests that despite the majority of studies using task-based fMRI (Chaddock et al., 2010, Chaddock-Heyman et al., 2013, Davis et al., 2011), it currently stands as a more effective approach to evoking neural processes relevant to the cognitive phenotypes being explored, providing a better avenue to examine the impact of organised sports on FC (Zhao et al., 2023).

There were no significant CT differences between the different sports categories at baseline or longitudinally. However, baseline analyses showed that greater subGM volumes within the cerebellum cortex, thalamus (including ventral diencephalon), pallidum, amygdala, hippocampus, brain stem, ventral striatum (i.e. nucleus accumbens), and dorsal striatum (i.e. putamen) was significantly associated with sports participants, compared to non-participants. These brain regions are pivotal for voluntary motor tasks, encompassing action planning, coordination, inhibition, and memory formation (Khan and Hillman, 2014). This aligns with evidence demonstrating that children with greater sports participation, exhibit larger volumes in these regions while showing enhancements in their ability to buffer against behavioural interference (Estévez-López et al., 2023, Khan and Hillman, 2014). Team sports participation was also associated with greater whole-brain FA, specifically in the genu of the corpus callosum, relative to non-participants at baseline. The corpus callosum is vital for integrating sensorimotor and cognitive information across the brain hemispheres, supporting attention, memory, and cognitive processing speed (Chaddock-Heyman et al., 2018). Consequently, results suggest an association between team sports participation and improved estimates of microstructural integrity within the genu of the corpus callosum, facilitating neurotrophic growth factor upregulation, associated with improved whole-brain WM microstructure in children (Chaddock-Heyman et al., 2018). In summary, findings indicate that organised sports participation could be associated with individual GM and WM differences in regions vital to goal-directed cognition underlying perception, memory, and action, thereby facilitating optimal learning and cognition.

When comparing team to individual sport types, team sports participation was associated with lower MD and higher FA across the whole brain, including specific regions like the corpus callosum, fornix, and forceps minor at baseline. These findings suggests that team sports are associated with healthier WM microstructure in these regions. In contrast, longitudinal analyses demonstrated that individual sports are associated with greater microstructure within the right corticospinal WM tract, as shown by smaller FA interval changes compared to team sports. However, it is important to note that the DTI effect sizes were small, possibly explaining the absence of a significantly advantage of team over individual sports observed across various cognitive measures.

Therefore, the subtle differences in brain morphology and non-significant differences cognitive results when comparing team-based versus individual-based sports might be more reflective of the limited explanatory power of focusing solely on the social collaborative element when categorising different sports, as this perspective might be overly simplistic. The complex relationship between various social, environmental, cognitive, and motor demands that different sports might impose on children could have a greater influence in mediating the effect of sports type on cognitive development and brain health (Giordano et al., 2021). As aforementioned, this suggests that while social collaboration is important, it may not be exclusive in determining how different sports would influence childhood neurocognitive development. Further research is needed to establish the underlying mechanisms linking various sports and their influence on neurocognitive development.

This study has several major implications for neurocognitive development. First, this comes at a time when youth sports participation is significantly decreasing (Hoffmann et al., 2022). By exploring the impact of participation in different organised sport types on childhood neurocognitive development, this study aims to better inform evidence-based public health policies in promoting youth sports participation. Furthermore, while the results highlight the benefit of sports on childhood neurocognitive development, they also stress looking beyond the simple dichotomies of individual versus team sports and a singular focus on the factor of social collaboration between the two sport types. It is apparent that other socio-cognitive requirements that various sports place on a child might have a greater influence in mediating the effect of sport type on neurocognitive development. Conversely, children's inherent cognitive functioning and neurological characteristics may influence their participation in sports and/or a specific sports type. Therefore, it must be highlighted that this study was a cohort study – with no random assignment of participants to any sports type – which does not allow any causal inferences to be made.

5.3. Limitations and future directions

This study has several limitations. First, the SAIQ data used to classify children into different organised sport types were based on parental reports. It is possible that the caregivers may incorrectly estimate the type and extent of the sport(s) in which their child participated. Second, labelling of metrics using the 12 × 12 Gordon FC atlas and parcels for sMRI limits the detail which differences attributed to sports type can be discerned from MRI data. Hence, studies working on the ABCD data may want to expend additional resources to preprocess neuroimaging data to extract information at much granular scales, taking the form of finely segregated FC edges and vertex-wise/voxel-wise data on brain structures. Third, the ABCD study was conducted in the US population. Future investigations are necessary to determine the generalisability of the current findings to other populations. Fourthly, current analyses only utilised data from a 2-year period with participants being excluded for various reasons. Future research should use longer time intervals to better assess the association between sports participation and neurocognitive development. Additionally, fluid and total composite scores were removed from analysis as they were not administered at both baseline and follow-up. However, measures of IQ have been shown to be associated with cognitive function and neuroimaging measures (Burgaleta et al., 2014, Tillman et al., 2009). Future studies should aim to include IQ scores as a covariate when analysing cognitive test scores.

Lastly, information on the level of amateur or professional training provided to the subjects who participated in sports was unavailable. It would be interesting for future research to study how varying levels of professionalism in sports coaching and resources children receive during sports participation would influence their neurocognitive development.

6. Conclusion

Overall, organised sports participation was shown to be associated with better cognitive performance, complementing previous research. Analyses of brain morphology revealed increased GM and WM development in subcortical regions linked to goal-directed cognition, memory, and action in sports participants. However, contrary to what was hypothesised, children who participated in team sports did not systematically show significantly better performance across various cognitive tasks and time, and vice versa, compared to individual sports. The absence of significant rs-FC differences between team and individual sports further corroborates the cognitive results. Interestingly, while children who engaged in individual-based sports exhibited greater microstructural integrity in the right corticospinal tract, the magnitude of this effect was small. This could be linked to the lack of significant differences in cognitive outcomes between team and individual sports. To guide the development of more effective youth sports programmes to optimise neurocognitive developmental outcomes, future research should aim to elucidate the nuanced mechanisms through which various sports serve to influence cognitive development and brain health.

CRediT authorship contribution statement

Alicia M. Goodwill: Writing – review & editing, Supervision, Investigation, Conceptualization. Junhong Yu: Writing – review & editing, Supervision, Formal analysis, Data curation. Fu-Miao Tan: Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

Supplementary material

Data availability

The codes used for analyses are available online at http://dx.doi.org/10.17605/OSF.IO/Q5WYX. The data are not publicly available as we did not obtain consent for public release of data.