The relationships among executive functions,self-regulation,and physical exercise in children with autism spectrum disorder

Abstract

Physical exercise is widely reported beneficial to executive functions in children with autism spectrum disorder. However, its impact on self-regulation in the population remains unknown. This study is to test whether two types of physical exercise (cognitively engaging vs non-cognitively engaging) benefited self-regulation and whether the social, emotional, and physical needs of an individual mediated the exercise–executive function and exercise–self-regulation relationships. Sixty-four children diagnosed with autism spectrum disorder were randomly assigned into 1 of 3 groups: learning to ride a bicycle (n = 23), stationary cycling (n = 19), or an active control with walking (n = 22). Two executive functions (flexibility and inhibition), self-regulation and the mediating roles of perceived social support, enjoyment, stress, physical self-efficacy, and perceived physical fitness were assessed. Participants in the learning to ride a bicycle group significantly improved their executive functions (p values < .01). The learning to ride a bicycle group and the stationary cycling group also significantly enhanced their self-regulation (p values < .001). Mediation analyses showed that physical self-efficacy and perceived physical fitness partially mediated the exercise–executive function relationship. Meanwhile, perceived social support significantly mediated the exercise–self-regulation relationship (p < .05). Our findings highlight the value of cognitively engaging exercise on enhancing executive functions in children with autism spectrum disorder in part by improving their physical self-efficacy and perceptions of fitness.

Lay abstract

This study examined the impacts of two types of physical exercises (two-wheel cycling vs stationary cycling) on cognition and self-regulation among 64 children with autism spectrum disorder. It also explored the role of social, emotional, and physical needs of an individual in the relationship between exercise, cognition, and self-regulation. Results showed that participants in the two-wheel cycling group showed significant improvements in their cognition and that the two exercise groups also enhanced their self-regulation. Moreover, this study also revealed that the social need is crucial in mediating the relationship between exercise and self-regulation. This study strengthens the notion that cognitively engaging exercise is more beneficial than the non-cognitively engaging exercise in enhancing cognition in children with autism spectrum disorder.

Keywords

autism children cognitive function executive function physical exercise self-regulation

Introduction

Given the well-evidenced cognitive benefits of physical exercise for executive functions (EFs) in children with typical development (TD; Xue et al., 2019), there is growing research interest in analyzing whether such benefits could also be translated to children with autism spectrum disorder (ASD). Previously, we examined the effectiveness of a 12-week basketball training intervention on inhibition control and working memory in children with ASD (Tse et al., 2019). Results showed that the training improved their inhibition control. More recently, Liang and colleagues (2022) conducted a meta-analysis of seven studies examining the effect of physical exercise interventions on EFs in children and adolescents with ASD. They concluded that chronic exercise interventions were beneficial to overall EFs in the population, particularly for cognitive flexibility and inhibitory control (Liang et al., 2022). While physical exercise appears to be beneficial in this population, the mechanism by which physical exercise potentially impacts EFs in children with ASD remains a question. It is important to understand the mechanism in order to design an effective physical exercise intervention to promote the development of EFs among children with ASD.

Over the past few decades, most of the studies examining the exercise–cognition relation in the general population have viewed the mediating mechanism via a neurobiological framework, expressed most clearly by the neurotrophic hypothesis (see Lippi et al., 2020 for a review). The hypothesis states that physical activity increases metabolic demands and triggers a cascade of biochemical changes, such as enhancing cerebral blood flow and increasing the availability of brain-derived neurotrophic factor, which strengthens brain plasticity for higher-level cognitive activities such as those involved in EFs (Khan & Hillman, 2014). Research in ASD population also focused on the neurobiological effects of physical exercise on cognitive functions. For example, (Moradi and colleagues (2018) showed that nerve growth factor concentration increased in children with ASD after perceptual-motor exercise intervention (Moradi et al., 2018). Bremer and colleagues (2020) demonstrated that acute exercise was able to increase the cerebral oxygenation significantly and improve inhibition control in children with ASD (Bremer et al., 2020). It is not until recently that scientists started questioning whether the exercise–cognition relation could also be mediated by a person’s social, emotional, and physical needs (Diamond & Ling, 2016). Diamond and Ling (2016) hypothesized that the most successful approaches for improving EFs would address social, emotional, and physical needs and that cognitively engaging physical activity (e.g. martial arts and dance) that enhances social interaction and joy would be more beneficial to EFs than less cognitively engaging physical activity (Diamond & Ling, 2016). To the best of our knowledge, no previous studies have examined the possible mediating roles of social support, emotion, and physical fitness in the exercise–EF relation in either children with TD or children with ASD.

Another dimension of the exercise–EF relation that has not examined is the impact of physical exercise on self-regulation (SR). SR is a psychological construct that encompasses a range of functional behaviors, such as interacting with peers, remembering rules and regulations, controlling emotions and inhibiting inappropriate and aggressive actions (Pandey et al., 2018). The lack of attention to the impact of physical exercise on SR is surprising, given these behaviors call upon the higher-order cognitive processes associated with EFs (e.g. shifting attention, working memory, and inhibition) and the fact that SR and EFs predict many of the same positive outcomes (e.g. physical health, mental health, and academic achievement; Baumeister & Vohs, 2003; Inzlicht et al., 2021), SR has long been thought of as the behavioral manifestation of EFs (Finders et al., 2021; Pandey et al., 2018; Tse et al., 2021). However, several recent studies provided compelling evidence that SR and EFs are distinguishable and should be treated independently (Howard & Vasseleu, 2020; Inzlicht et al., 2021; Saunders et al., 2018). Confusion may arise when measuring one without the other (Inzlicht et al., 2021). Therefore, it is important to investigate whether physical exercise could yield similar benefits in SR as those in EFs, particularly in children with ASD where SR difficulties are common (Nowell et al., 2019).

Therefore, the purposes of this study were to examine the exercise–SR relation and to investigate the possible mediating roles of social and emotional experience and physical perception in the exercise–EF and exercise–SR relationships in children with ASD. In this study, these needs were expressed by individuals’ perceived social support, enjoyment, stress, physical self-efficacy, and perceived physical fitness. Similar to our previous study (Tse et al., 2021), we compared EFs among three groups: (1) learning to ride a bicycle, (2) stationary cycling, and (3) active control (walking) before and immediately after the 2-week intervention period. Following the suggestion by Diamond and Ling (2016), the active control group with walking (instead of a no-treatment control group) was used to control for potential Hawthorne effects (McCarney et al., 2007). Walking was chosen because it was a low-intensity physical activity that enabled us to assess the same potential mediators as those in the intervention groups. Unlike our previous study, only inhibition and flexibility were measured in this study because exercise interventions were shown effective to improve these two EFs in children with ASD (see Liang et al., 2022 for review) and to enhance the feasibility of the study (to avoid overburdening participants with the additional mediation assessments compared to our previous study). To examine the mediating effects, perceived social support, enjoyment, stress, physical self-efficacy, and perceived physical fitness were measured during the baseline period, mid-intervention, and post-intervention.

Methods

Study design

The study was a three-armed randomized controlled trial (RCT) design with equal allocation ratio to the two intervention groups and one active control group (1:1:1). This study is registered on ClinicalTrials.gov with the identification code: NCT05503459.

Data collection

EFs, SR, and potential mediators were assessed for all participants in their respective schools. EF and SR assessments were conducted a day before the intervention (baseline) and immediately after the last intervention session or regular walking program (post-intervention). The sequence of EF and SR assessments were counterbalanced to prevent order effects. To investigate the mediating effects, the mediation assessments were conducted in the fifth intervention session (mid-intervention). The sequence of the mediation assessments was also counterbalanced to prevent order effects. Figure 1 is the CONSORT flow diagram.

Figure 1.

CONSORT diagram.

Sample size calculation

The sample size was based on the two most relevant prior studies examining the impact of physical exercise on EFs and SR in children with ASD (Razza et al., 2015; Tse et al., 2021), which showed that physical exercise had strong enhancement effects, corresponding to a Cohen’s d of .69 on EFs (Tse et al., 2021) and .85 on SR (Razza et al., 2015). If the effects of our intervention were similar to those in these two studies, a sample of 21 participants per group was required to achieve a power of 80% and a level of significance of 5%. Assuming 20% attrition, 27 participants were required in each group.

Participants

Initially, 82 participants were recruited from four local special schools for children with mild intellectual disability and from parents’ verbal referrals. The inclusion criteria were (1) age 8–12 years (matching with our previous study); (2) mild to moderate ASD (i.e. level 1–2 support classification (Weitlauf et al., 2014) diagnosis from physicians or psychologists based on the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5: American Psychiatric Association, 2016) and Autism Diagnostic Observation Schedule, 2nd Edition (ADOS-2: Lord et al., 2012); (3) non-verbal IQ over 50 using a brief version of the Wechsler Intelligence Scale for Children (Chinese revised) (C-WISC; see Gong & Cai, 1993 for more information); (4) able to follow instructions with the assistance (e.g. prompting, re-read the instructions) of research staff; (5) able to perform the requested physical intervention, EF measures and mediator measures with the assistance of the research staff; (6) no additional regular participation in physical exercise other than school physical education classes for at least 1 month prior to the study, and (7) novice at riding a two-wheel bicycle (i.e. cannot ride the bicycle alone for more than 10 consecutive seconds). The exclusion criteria were (1) other medical conditions that limited physical exercise capacities (e.g. asthma, seizure, and cardiac disease); (2) a complex neurologic disorder (e.g. epilepsy, phenylketonuria, fragile X syndrome, and tuberous sclerosis), (3) suffering from obesity (i.e. >95th percentile of age-gender-specific body mass index (BMI) cutoff (Cole & Lobstein, 2012) such that it would be difficult for research staff to catch them if they began to fall when riding, and (4) self-reported color blindness.

In addition, we collected parent ratings of autistic traits and autism behaviors using the Social Responsiveness Scale, Second Edition (SRS-2: Constantino & Gruber, 2012), information for each participant from the parents, including records in after-school group therapy (e.g. occupational therapy and speech therapy) and medication usage. After screening, a total of 75 participants joined the study and they were randomly assigned to the two exercise intervention groups and the active control group. However, two participants from the learning to ride a bicycle group, six participants from the stationary cycling group and three participants from the active control group dropped out in the middle of the study. Consequently, 64 participants (23 in the learning to ride a bicycle group, 19 in the stationary cycling group, and 22 in the active control group) successfully completed the study as the intervention was intended when designed. Written consent was obtained from participants’ parents/guardians. The study was approved by the university’s ethics committee. There was no community involvement in this study. Demographic data for the two groups is shown in Table 1.

Table 1.

Demographic of participants.

	Learning to bicycle group (n = 23)	Stationary cycling group (n = 19)	Control group (n = 22)	p
Gender	20 boys and 3 girls	13 boys and 6 girls	19 boys and 3 girls
Age (years)	9.61 ± 1.41	10.16 ± 1.38	10.275 ± 1.39	.24
Weight (kg)	37.19 ± 9.10	42.19 ± 9.53	40.02 ± 7.59	.18
Height (m)	1.36 ± 0.11	1.39 ± 0.20	1.41 ± 0.11	.42
BMI (kg/m²)	19.67 ± 3.00	21.45 ± 3.05	20.33 ± 2.75	.09
Non-verbal IQ	58.00 ± 8.15	57.89 ± 8.08	56.18 ± 8.27	.71
SRS-2 Raw scores	70.82 ± 7.09	68.68 ± 9.52	69.63 ± 9.14	.72
Medication (n)
Yes	3	2	3
No	20	17	19

BMI: body mass index; IQ: intelligence quotient; SRS: Social Responsiveness Scale, Second Edition.

Randomization

After screening, all the eligible participants were randomly assigned to the two intervention groups and the active control group. To ensure equal allocation ratios for the intervention groups and the active control group, block randomization (Efird, 2011) was used. A block size of 5 was used in the proposed study. A trained research assistant completed the block randomization process.

Intervention

Learning to ride a bicycle group: The protocol for this intervention group was identical to that in our previous study (Tse et al., 2021), which was a 2-week bicycle training program consisting of 10 sessions (five sessions per week, 60 mins per session) in a hall/gymnasium of each participating school and the Education University of Hong Kong. Each intervention session was conducted by a professional cycling instructor assisted by student helpers. The staff-to-participant ratio was 1:1. Each intervention session was conducted in an identical format, comprising three activities: warm-up (10 min), bicycle training (40 min), and cool-down (10 min). In the bicycle training activity, participants were asked to ride on a training bicycle with training wheels to gain better control of the bike in a gradual way. Participants then progressed from the training bicycle to a two-wheel bicycle. To keep participants on the learning curve, they were asked to ride through an obstacle course that was progressively more difficult to negotiate. The obstacles were designed by a focus group, which consisted of four physical education teachers from participating schools and one experienced cycling coach with more than 5 years of coaching experience. Similar to our previous study (Tse et al., 2021), the exercise intensity was measured by asking participants every 10 mins during exercise to indicate their rating of perceived exertion (target range: 3–5) with the OMNI scale (Robertson et al., 2000). Participants were positively reinforced verbally with compliments for their efforts in the training program and their daily improvements were visualized through graphs kept in the child’s home.

Stationary cycling group: the protocol for this intervention group was identical to that in our previous study (Tse et al., 2021).

Active control group: Participants were asked to walk with their major caregivers (e.g. parents, grandparents) for 20 min every day during the study period. After the study, they were taught how to ride a bicycle to recognize their contribution as controls.

Measures

EFs: As in our previous study (Tse et al., 2021), the Stroop Color and Word Test (SCWT: Stroop, 1935) and a Go/No-go (GNG) task were used to assess cognitive flexibility and inhibition, respectively. In the SCWT, the participants were required to read three different tables as fast as they could. The three different tables were classified into two conditions: congruent and incongruent conditions. In the congruent condition, participants were first required to read the names of the colors printed in black (W; first table) and name different color patches (C; second table; Scarpina & Tagini, 2017). After that they were asked to read the third table, where the color names were printed in different colors (e.g. the word “yellow” is printed in red ink; CW; Scarpina & Tagini, 2017). The flexibility score, which is represented by the interference score (IG), is calculated using the formula IG = CW − (W × C)(W − C). This formula was used because it has been the most frequently used in previous studies (see Scarpina & Tagini, 2017 for a review). Lower IG scores indicate better cognitive flexibility.

In the GNG task, participants were asked to press a left or right key as quickly as possible when the corresponding arrow appeared on the center of a computer screen (Go response) and not to press any key whenever the up arrow appeared on the screen (No-go response). Following 20 practice trials, participants completed 300 trials: 220 trials requiring a Go response (110 left and 110 right) and 80 trials (26.7%) requiring a No-go response (not pressing any key; Tse et al., 2021). The stimuli were randomly presented, one at a time, for 500 ms followed by 1000 ms of blank interval using E-Prime 3.0 software. After blocks of 60 trials, children were offered a break of 2 min. No feedback was given upon response, and the response time was recorded but not analyzed because of the unreliability of the recording procedure. As in Uzefovsky et al. (2016)’s study, a Go response in a No-go trial was coded as a false alarm (FA). FA errors are considered an indicator of inhibition, and the lower the error, the better the inhibition (Uzefovsky et al., 2016). Considering the cognitive ability of the participants, all the tasks were conducted with the assistance of trained student helpers. For example, some of the participants were instructed several times with visual instructions (e.g. what the computer screen would display and how they should answer the questions) to ensure their understandings toward the assessments. Each participant took approximately 20 minutes to complete each EF task.

SR: SR of each participant was assessed by another group of student helpers (the helper–participant ratio 1:1) using the Response to Challenge Scale (RCS: Lakes, 2012). The assessment was conducted when the participants were learning how to perform a stationary dribbling task, which was identical to the one used in our previous study (Tse et al., 2019), before and after the intervention. The RCS is an observer-rated instrument measuring children’s cognitive, affective, and motor regulation in response to a physically challenging task (Lakes, 2012). The RCS consists of 16 items (e.g. attentive–inattentive) rated on 7-point scales. The measurements were carried out at pre- and post-interventions. Due to funding limitations, the SR level of each participant was rated by one rater only instead of multiple raters, as suggested by a previous study (Lakes, 2012). The total score for each participant was computed for analyses. Lower scores reflected stronger SR.

Perceived social support: The perceived social support of the participant was assessed with a self-reported scale referenced on the Athlete Received Support Questionnaire (Freeman et al., 2014). Considering the comprehension difficulties of the participants, six items were chosen by the focus group (consisting of the authors, caregivers, and teachers of the participants; see Appendix 1). Each participant was asked to rate their perceived social support with verbal prompting from their partnered student helpers.

Enjoyment: Participants were assisted in rating their enjoyment of the interventions and walking (active control) using the Physical Activity Enjoyment Scale (Vitali et al., 2019). The scale was rated on 5-point Likert-type scale (see Appendix 2). The total score was computed from the scale. The higher the total score, the better the enjoyment.

Stress level: Referenced on the Feeling Scale (Hardy & Rejeski, 1989), a stress scale was designed by the focus group in this study (see Scale, Appendix 3). Similar to the Feeling Scale, participants were asked to indicate their stress level from “very relaxed” (+5) to “very stressed” (−5) during the middle of the intervention.

Physical self-efficacy: Participants were assisted in rating their self-efficacy in the interventions and walking (active control) using the Physical Self-efficacy Scale (Vitali et al., 2019). This scale is rated on 4-point Likert-type scale (see Appendix 2, which measure the physical self-efficacy and physical activity enjoyment). The total score was computed. The higher the total score, the better an individual’s physical self-efficacy.

Perceived physical fitness: Similar to a previous study by Sollerhed and colleagues (2007) participants were asked “How could you evaluate our own physical fitness when comparing with others?” and told to answer using a 5-point scale (see Questionnaire, Appendix 4; Sollerhed et al., 2007). The higher the rating, the better.

Masking

The staff responsible for the cognitive assessments and data analyses were not aware of the participants’ group assignments and were not involved in data collection.

Statistical methods

RCT analysis: All statistical analyses were conducted using SPSS (version 27.0) for Windows (SPSS Inc., Chicago, IL, USA). All the data were entered into SPSS by a research assistant. One-way (three groups: learning to bicycle vs stationary cycling vs active control) analyses of covariance (ANCOVAs) with repeated measures were performed for each EF outcome and SR to compare the changes between and within groups over different time periods. Considering the potential confounding effects of developmental factors, age and IQ were controlled as covariates. Post hoc analyses were performed when any significant difference was found in any of the outcome variables. Bonferroni correction was used to adjust the alpha levels (i.e. p = .05/3 = .017).

Mediation analysis: The analysis was conducted to assess whether social, emotional, and physical needs were mediators in the exercise–EF and exercise–SR relationships. The SPSS PROCESS Macro version 3.1 was used for the analyses, with Model 4 (mediation effect) and 5.000 biased-corrected bootstrap samples (Hayes, 2013). Mediations were assessed by the indirect effects (IEs) of the physical activity interventions on EFs and SR through different possible mediating variables (i.e. perceived social support, enjoyment, physical self-efficacy, stress level, and perceived physical fitness). The total (c path), direct (c’ path), IE (a*b paths) coefficients, as well as IEs with 95% confident intervals, were presented, regardless of the significance of the total effects (physical exercise interventions on EF and SR) and direct effects (the effect on EF and SR when physical exercise interventions and different perceived variables are included as predictors; Hayes, 2013). Simple mediation models were tested by introducing independently the five different mediating variables. Similar to a previous study (Visier-Alfonso et al., 2020), a parallel serial mediation model was then tested in which each possible mediating variable was simultaneously entered into the model in such a way that each mediation (PM) was calculated as (IE/total effect) × 100 to estimate the percentage of the total effect explained by the mediation path, whenever the total effect was larger than the IE and in the same direction (Visier-Alfonso et al., 2020).

Results

All the EF and SR values were comparable between groups at baseline (see Table 2).

Table 2.

Comparisons of neuropsychological measures between groups and within groups between pre- and post-interventions.

Neuropsychological assessment	Learning to ride a bicycle group (SD)	Stationary cycling group (SD)	Active control group (SD)	p value (group effect)	p value (interaction effect)
Flexibility (SCWT IG)					0.008
Pre	−5.76 (3.56)	−6.81 (4.17)	−6.54 (3.33)	0.62
Post	−8.99 (5.12)	−7.62 (5.48)	−5.79 (4.84)	0.12
p value (time effect)	0.01	0.61	0.34
Cohen’s d effect size (95% CI)	0.73 (0.12, 0.58)	0.17 (–0.47, 0.80)	0.10 (–0.77, 0.41)
Inhibition (GNG FA error)					0.004
Pre	16.52 (8.92)	15.53 (9.32)	20.18 (9.53)	0.24
Post	10.09 (7.34)	14.47 (10.12)	21.19 (10.77)	0.001
p value (time effect)	0.001	0.56	0.58
Cohen’s d effect size (95% CI)	0.79 (0.32, 1.26)	0.14 (–0.32, 0.59)	−0.12 (–0.54, 0.30)
Self-regulation (Total RCS score)					0.001
Pre	62.52 (19.78)	73.42 (20.13)	75.05 (14.45)	0.05
Post	41.00 (17.48)	55.26 (15.56)	79.14 (18.24)	0.001
p value (time effect)	0.001	0.001	0.33
Cohen’s d effect size (95% CI)	1.48 (0.88, 2.07)	1.64 (0.94, 2.33)	−0.21 (–0.63, 0.21)

SD: standard deviation; SCWT IG: Stroop Color Word Test Interference Score; CI: confidence interval; GNG FA error: Go/No-go False Alarm error; RCS: Response to Challenge Scale.

Flexibility: As shown in Table 2, there was a significant group × time interaction effect (p < .02). Follow-up tests showed that the bicycle learning group had a significantly lower SCWT IG score than the active control group (p < .02) with a moderate effect size (d = −.64) after the intervention. No significant difference in the score was observed between the other groups (p > .02) at that time point. Within group, the bicycle learning group showed a significant drop in SCWTIG after the intervention (p < .02) with a moderate effect size (d = .73), whereas the other two groups showed no differences in the score (p values > .02).

Inhibition: A significant group × time interaction effect was observed for the GNG FA error (p < .02). Subsequent tests revealed that the FA error of the bicycle learning group was significantly smaller than that of the active control group (p < .02) with a strong effect size (d = –1.21) while there was no significant difference in the error between the other two groups (p > .02) after the intervention. Within group, there was a significant reduction of FA error from baseline to post-intervention in the bicycle learning group (p < .02) with a moderate effect size (d = .79). By contrast, no significant differences were observed in either the stationary cycling group or the control group (p values > .02) between these time points.

SR: There was a significant interaction effect for SR total score (p < .001). The bicycle learning group showed a significantly lower score than the other two groups after the intervention (p values < .02). The stationary cycling group also showed a significantly lower score than the active control group (p < .02). Within group, the bicycle learning group and the stationary cycling group showed a significant improvement in SR score after the intervention (p values < .001) with very strong effect size (d = 1.48 and d = 1.64; see Table 2 for more information), whereas the active control group showed no differences in the SR score between pre- and post-intervention.

The mediation model scheme is displayed in Figure 2, and the results of the mediation analysis are shown in Table 3.

Figure 2.

Mediation diagram.

Table 3.

Mediation analyses.

Y	Mediator	Total effect (c)	Direct effect (cʹ)	Indirect effect (ab)	95%CI		P_M (%)
Y	Mediator	Total effect (c)	Direct effect (cʹ)	Indirect effect (ab)	Lower	Upper	P_M (%)
Learning to ride a bicycle vs Stationary cycling
Flexibility	Perceived social support	1.38	1.28	0.10	−4.34	3.22	7.25
	Enjoyment	1.38	2.19	−0.82	−3.35	2.21	−59.42
	Stress	1.38	0.85	0.52	−1.05	2.46	37.68
	Perceived self-efficacy	1.38	0.01	1.36	−1.20	4.63	98.55
	Perceived physical fitness	1.38	0.24	1.14	−1.58	4.69	82.61
Inhibition	Perceived social support	4.39*	6.18	−1.79	−7.39	3.44	−40.77
	Enjoyment	4.39*	3.65	0.73	−3.17	4.84	16.63
	Stress	4.39*	4.86	−0.48	−3.59	3.18	−10.93
	Perceived self-efficacy	4.39*	2.62	1.76	−2.26	5.21	40.09
	Perceived physical fitness	4.39*	0.19	4.20*	0.94	8.32	95.67
Self-regulation	Perceived social support	14.26*	7.20	7.06	−3.56	19.50	49.51
	Enjoyment	14.26*	9.37	4.89	−2.60	17.09	34.29
	Stress	14.26*	12.84*	1.43	−4.27	8.55	10.03
	Perceived self-efficacy	14.26*	12.13	2.13	−4.65	10.35	14.94
	Perceived physical fitness	14.26*	11.20	3.06	−3.23	12.97	21.46
Learning to ride a bicycle vs Active control group
Flexibility	Perceived social support	1.60*	2.73	−1.13	−5.96	1.66	−70.63
	Enjoyment	1.60*	1.28	0.32	−1.38	1.96	20.00
	Stress	1.60*	1.53	0.07	−0.74	0.85	4.38
	Perceived self-efficacy	1.60*	0.35	1.25*	0.13	3.90	78.13
	Perceived physical fitness	1.60*	1.89	−0.29	−1.95	1.21	−18.13
Inhibition	Perceived social support	5.55**	4.00	1.55	−2.81	5.84	27.93
	Enjoyment	5.55**	4.85*	0.70	−3.39	3.14	12.61
	Stress	5.55**	5.27**	0.27	−1.00	2.19	4.86
	Perceived self-efficacy	5.55**	5.80*	−0.25	−2.59	2.11	−4.50
	Perceived physical fitness	5.55**	3.40	2.15*	0.06	4.65	38.74
Self-regulation	Perceived social support	19.07**	8.78	10.29*	1.07	22.36	53.96
	Enjoyment	19.07**	18.43**	0.64	−5.06	8.58	3.36
	Stress	19.07**	17.41**	1.66	−0.90	5.35	8.70
	Perceived self-efficacy	19.07**	18.96**	0.11	−4.44	5.07	0.58
	Perceived physical fitness	19.07**	17.24**	1.82	−2.82	6.60	9.54
Stationary cycling vs Active control group
Flexibility	Perceived social support	1.83*	2.04	−0.21	−3.14	2.06	−11.48
	Enjoyment	1.83*	2.42	−0.59	−3.04	0.81	−32.24
	Stress	1.83*	1.81	0.01	−0.81	0.52	0.55
	Perceived self-efficacy	1.83*	1.55	1.64*	0.43	1.13	89.62
	Perceived physical fitness	1.83*	1.83	0.00	−0.66	0.49	0.00
Inhibition	Perceived social support	6.71*	8.63*	−1.93	−7.46	2.92	−28.76
	Enjoyment	6.71*	6.36	0.35	−3.75	2.35	5.22
	Stress	6.71*	7.13*	−0.42	−2.28	1.24	−6.26
	Perceived self-efficacy	6.71*	6.45	0.25	−0.60	2.49	3.73
	Perceived physical fitness	6.71*	6.71*	0.00	−2.36	2.56	0.00
Self-regulation	Perceived social support	23.87**	15.56*	8.31*	1.45	18.15	34.81
	Enjoyment	23.87**	23.58**	0.29	−3.08	7.68	1.21
	Stress	23.87**	24.03**	−0.16	−3.29	1.33	−0.67
	Perceived self-efficacy	23.87**	23.98**	−0.10	−2.88	1.67	−0.42
	Perceived physical fitness	23.87**	23.87**	0.00	−4.83	2.63	0.00

Total, direct, and indirect effects of the simple mediation analyses investigating perceived social support, enjoyment, stress, perceived self-efficacy, and perceived physical fitness as mediators between physical exercise interventions and EFs and SR.

CI: confidence interval; EF: executive function; SR: self-regulation.

p < 0.05; **p < 0.001.

As shown in Table 3, mediation analysis revealed significant IEs of physical self-efficacy on cognitive flexibility for the bicycle learning group versus active control group (Effect = 1.25, 95% CI: (0.13, 3.90)) and for the stationary cycling group versus active control group (Effect = 1.64, 95% CI: (0.43, 1.13)). It also revealed that the IEs of perceived physical fitness on inhibition were statistically significant for the bicycle learning group versus stationary cycling group (Effect = 4.20, 95% CI: (0.94, 8.32)) and for bicycle learning group versus active control group (Effect = 2.15, 95% CI: (0.06, 4.65)). In terms of the exercise–SR relationship, the mediation analyses showed significant IEs of perceived social support on SR for the comparison of learning to a ride a bicycle group versus active control group (Effect = 10.29, 95% CI: (1.07, 22.36)), and for the stationary cycling group versus active control group (Effect = 8.31, 95% CI: (1.45, 18.15)).

Discussion

This study is the first study to examine the impacts of different exercise interventions on SR in children with ASD and to investigate whether the social and emotional experiences and physical perceptions of an individual mediate the exercise–EF and exercise–SR relationships in the population. Results revealed that both exercise interventions improved SR. Mediation analyses revealed significant mediating roles of physical self-efficacy in the exercise–cognitive flexibility relationship, as well as a mediating role of perceived physical fitness in the exercise–inhibition relationship.

The findings on exercise–SR relationship are in line with previous studies in TD population, which showed that greater exercise participation is positively related to SR in preschool children with TD (Becker et al., 2014; Carson et al., 2016; Williams, 2018). For example, Becker and colleagues (2014) showed that children who engaged more in physical activity (active play) had better SR and academic achievements. The authors explain that this might be due to the embodied nature of cognition, where mental processes such as those associated with EFs (particularly attention and inhibition) and problem-solving are facilitated through the interactions between physical movements and the environment (Becker et al., 2014). As SR relies heavily on inhibition control, attentional control, and working memory, SR is therefore facilitated through these interactions (Becker et al., 2014). The present results further support the translation of this SR benefit to children with ASD.

It is important to note that learning to ride a bicycle was more effective than stationary cycling in enhancing SR. One possible explanation may lie in the different behavioral and cognitive requirements of each task. Compared with the participants in the stationary cycling group, participants in the learning to ride a bicycle group were believed to exercise more control on their thoughts and behaviors to accomplish various tasks such as maintaining body balance, navigating one’s position relative to the environment and obstacles, focusing on the coach’s instructions, inhibiting any thought of riding the bicycle too fast or too slow or too recklessly, and progressively adapting from the four-wheel to two-wheel bikes. All these tasks require considerable perceptual, motor, and cognitive resources and were presumably more challenging than cycling on a stationary bicycle, which, therefore, required more SR. This explanation is supported by our EF findings, as learning to ride a bicycle was associated with better cognitive flexibility and inhibition than stationary cycling and walking. These findings further resonate with the idea that cognitively engaging exercise, particularly when it involves skill-acquisition, yields better cognitive benefits than non-cognitively engaging exercise (Diamond, 2012; Tomporowski & Pesce, 2019).

Meanwhile, the results of mediation analyses further support the notion that physical exercise improves EF by enhancing perceived motor competence of the participants (Diamond & Ling, 2016, 2019; Willoughby et al., 2021). Generally, children with ASD have low physical self-efficacy and perceived physical fitness, which discourages them from participating in physical exercise (Bandini et al., 2013; Ketcheson et al., 2018; Liang et al., 2020). Considering physical self-efficacy and perceived physical fitness were highly associated with EFs in children in previous studies (Albuquerque et al., 2022; Verburgh et al., 2014) and that these two variables can be modified by physical exercise interventions in children with ASD (Healy et al., 2018; Ruggeri et al., 2019), it is reasonable to believe that one of the pathways that underlie the exercise–EF relationship is through the enhancement of these two variables. This was supported by this study; however, future research should seek to explain why this relationship exists. It may be that improving physical self-efficacy and perceived physical fitness increases an individual’s engagement in the physical activity (enhancing procedural and motor learning and increasing physical exertion), thereby furthering positive EF outcomes. This relationship may form a loop with increased self-efficacy and fitness in turn further increasing engagement.

In this study, we also hypothesized that perceived social support, enjoyment, physical self-efficacy, stress and physical fitness would mediate the exercise–SR relationship. However, the hypotheses were not all supported by the data. Contrary to our hypotheses, only perceived social support was shown to mediate the improvement in SR in children with ASD assigned to both exercise intervention groups. We believe that this might be due to the high ratio of research staff to participants. In each cycling activity, each participant was partnered with one university student helper. The student helpers provided continuous guidance and encouragement to the participants throughout the whole intervention. Based on the verbal reports from teachers and parents, the presence of the student helpers was crucial for the successful implementation of the interventions. Many parents reported that their children felt less motivated and more tired in the middle of the intervention. The assistance and encouragement of student helpers were critical in motivating them to complete the task. This kind of support may therefore address the social needs of the participants to adhere to the task requirements. Further investigation may consider assessing the motivation level of the participants during the exercise intervention and the impact of positive feedback on sustained participation.

Incorporating many assessments into one large and comprehensive RCT, the current findings help to clarify the current state of knowledge on the exercise–SR and exercise–EF relationships and their underlying pathways, particularly in children with ASD. However, several important issues require further investigation. First, the sample size of this study was relatively small for mediation analysis. Fritz and MacKinnon (2007) suggested that the sample size necessary to test a mediating effect for .8 power ranged between 71 and 562 (Fritz & Mackinnon, 2007). The relatively small sample size in this study may lead to low statistical power to detect existing associations of the other mediating variables. It also may be that the relatively brief intervention period (2 weeks) limited the opportunity to sufficiently examine potential mediating factors. We recommend future study should incorporate a longer period of intervention (e.g. 5 weeks or above). Second, we did not measure cognitive planning and working memory, which are the other two core components of EF. Further research should consider adding the measurements of these two components in order to offer a comprehensive measure of EF. Third, without an objective measurement of exercise intensity, it is difficult to control for the effect of this variable on the results, future studies may consider using the wrist-worn assessments (e.g. Actigraphy; Fitbit) which are valid and handy methods for measuring exercise intensity. Finally, this study criteria excluded children who had severe intellectual difficulties and severe ASD symptoms, it should therefore be cautious that the present findings cannot be applied to all children with ASD and those who have more difficulties in these areas.

Conclusion

In summary, this study reaffirmed the value of cognitively engaging exercise in improving EFs in children with ASD. It adds empirical evidence to the contention that exercise may positively influence EFs by enhancing perceived motor competence in children with ASD. Improving perceived motor competence and perceived physical fitness are likely two mechanisms by which physical exercise contributes to cognitive improvement in the population. It may be that increasing perceptions of competence in one domain (motor) generalizes to increased perceptions of competence in other areas (cognitive). It may also be that improving physical self-efficacy and perceived physical fitness increases an individual’s engagement in the physical activity (enhancing procedural and motor learning and increasing physical exertion), thereby furthering positive EF outcomes. Meanwhile, regardless of the types of physical exercise, exercise has been shown to enhance SR in children with ASD, and it is likely that such benefits are driven—at least in part—by providing social support for children.

Footnotes

Appendix 1

Appendix 2

Appendix 3

Appendix 4 Acknowledgements

The authors would like to thank the principals and the teachers from the special schools for their support with participant recruitment and the intervention implementation. The authors would also like to express their gratitude to all the children who participated in this study and to the participants’ teachers and parents for their support. The authors declare that the results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation and that presentation of the result does not constitute endorsement by ACSM.

Authors’ contributions

A.C.Y.T. designed the study and wrote the manuscript. V.H.L.L. coordinated the data collection and implemented exercise interventions. P.H.L. conducted the statistical analyses. D.I.A. and K.D.L. edited and revised the manuscript. All authors read and approved the final manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work described in this paper was supported by two grants: general research fund from Research Grants Council (project no. EdUHK 18603818) and funding support to General Research Fund from EdUHK (project no. RG21/2019-2020R).

ORCID iD

Andy CY Tse

References

Albuquerque

M. R.

Rennó

G. V. C.

Bruzi

A. T.

Fortes

L., d, S.

Malloy-Diniz

L. F.

(2022). Association between motor competence and executive functions in children. Applied Neuropsychology: Child, 11, 495–503.

American Psychiatric Association. (2016). Diagnostic and statistical manual of mental disorders: DSM-5.

Bandini

L. G.

Gleason

Curtin

Lividini

Anderson

S. E.

Cermak

S. A.

. . .Must

(2013). Comparison of physical activity between children with autism spectrum disorders and typically developing children. Autism, 17, 44–54.

Baumeister

R. F.

Vohs

K. D.

(2003). Self-regulation and the executive function of the self. In Leary

M. R.

Tangney

J. P.

(Eds.), Handbook of self and identity (pp. 197–217). Guilford Press.

Becker

D. R.

McClelland

M. M.

Loprinzi

Trost

S. G.

(2014). Physical activity, self-regulation, and early academic achievement in preschool children. Early Education and Development, 25, 56–70.

Bremer

Graham

J. D.

Heisz

J. J.

Cairney

(2020). Effect of acute exercise on prefrontal oxygenation and inhibitory control among male children with autism spectrum disorder: An exploratory study. Frontiers in Behavioral Neuroscience, 14, Article 84.

Carson

Tremblay

M. S.

Chaput

J.-P.

Chastin

S. F. M.

(2016). Associations between sleep duration, sedentary time, physical activity, and health indicators among Canadian children and youth using compositional analyses. Applied Physiology, Nutrition, and Metabolism, 41, S294–S302.

Cole

T. J.

Lobstein

(2012). Extended international (IOTF) body mass index cut-offs for thinness, overweight and obesity. Pediatric Obesity, 7, 284–294.

Constantino

Gruber

(2012). Social responsiveness scale (SRS-2). Western Psychological Services.

10.

Diamond

(2012). Activities and programs That improve children’s executive functions. Current Directions in Psychological Science, 21, 335–341.

11.

Diamond

Ling

D. S.

(2016). Conclusions about interventions, programs, and approaches for improving executive functions that appear justified and those that, despite much hype, do not. Developmental Cognitive Neuroscience, 18, 34–48.

12.

Diamond

Ling

D. S.

(2019). Aerobic-exercise and resistance-training interventions have been among the least effective ways to improve executive functions of any method tried thus far. Developmental Cognitive Neuroscience, 10, 100572.

13.

Efird

(2011). Blocked randomization with randomly selected block sizes. International Journal of Environmental Research and Public Health, 8, 15–20.

14.

Finders

J. K.

McClelland

M. M.

Geldhof

G. J.

Rothwell

D. W.

Hatfield

B. E.

(2021). Explaining achievement gaps in kindergarten and third grade: The role of self-regulation and executive function skills. Early Childhood Research Quarterly, 54, 72–85.

15.

Freeman

Coffee

Moll

Rees

Sammy

(2014). The ARSQ: The athletes’ received support questionnaire. Journal of Sport Exercise Psychology, 36, 189–202.

16.

Fritz

M. S.

Mackinnon

D. P.

(2007). Required sample size to detect the mediated effect. Psychological Science, 18, 233–239.

17.

Gong

Y. X.

Cai

T. S.

(1993). The wechsler intelligence scale for children revised in China (C-WISC). Hunan Maps Press. [in Chinese].

18.

Hardy

C. J.

Rejeski

W. J.

(1989). Not what, but how one feels: The measurement of affect during exercise. Journal of Sport and Exercise Psychology, 11, 304–317.

19.

Hayes

A. F.

(2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. Guilford Press.

20.

Healy

Nacario

Braithwaite

R. E.

Hopper

(2018). The effect of physical activity interventions on youth with autism spectrum disorder: A meta-analysis. Autism Research, 11, 818–833.

21.

Howard

S. J.

Vasseleu

(2020). Self-regulation and executive function longitudinally predict advanced learning in preschool. Frontiers in Psychology, 11, Article 49.

22.

Inzlicht

Werner

K. M.

Briskin

J. L.

Roberts

B. W.

(2021). Integrating models of self-regulation. Annual Review of Psychology, 72, 319–345.

23.

Ketcheson

Hauck

J. L.

Ulrich

(2018). The levels of physical activity and motor skills in young children with and without autism spectrum disorder, aged 2-5 years. Autism, 22, 414–423.

24.

Khan

N. A.

Hillman

C. H.

(2014). The relation of childhood physical activity and aerobic fitness to brain function and cognition: A review. Pediatric Exercise Science, 26, 138–146.

25.

Lakes

K. D.

(2012). The Response to Challenge Scale (RCS): The development and construct validity of an observer-rated measure of children’s self-regulation. The International Journal of Educational and Psychological Assessment, 10, 83–96.

26.

Liang

Wong

S. H. S.

Sum

R. K. W.

Sit

C. H. P.

(2020). Accelerometer-measured physical activity levels in children and adolescents with autism spectrum disorder: A systematic review. Preventive Medicine Reports, 19, 101147.

27.

Liang

Wong

S. H. S.

Sum

R. K. W.

Wang

Yang

Sit

C. H. P.

(2022). The effects of exercise interventions on executive functions in children and adolescents with autism spectrum disorder: A systematic review and meta-analysis. Sports Medicine, 52, 75–88.

28.

Lippi

Mattiuzzi

Sanchis-Gomar

(2020). Updated overview on interplay between physical exercise, neurotrophins, and cognitive function in humans. Journal of Sport and Health Science, 9, 74–81.

29.

Lord

Rutter

DiLavore

P. C.

Risi

Gotham

Bishop

S. L.

(2012). Autism diagnostic observation schedule (2nd ed.). Western Psychological Services.

30.

McCarney

Warner

Iliffe

van Haselen

Griffin

Fisher

(2007). The Hawthorne effect: A randomised, controlled trial. BMC Medical Research Methodology, 7, Article 30.

31.

Moradi

Sohrabi

Taheri

Khodashenas

Movahedi

(2018). The effects of different combinations of perceptual-motor exercises, music, and vitamin D supplementation on the nerve growth factor in children with high-functioning autism. Complementary Therapies in Clinical Practice, 31, 139–145.

32.

Nowell

S. W.

Watson

L. R.

Boyd

Klinger

L. G.

(2019). Efficacy study of a social communication and self-regulation intervention for school-age children with autism spectrum disorder: A randomized controlled trial. Language, Speech, and Hearing Services in Schools, 50, 416–433.

33.

Pandey

Hale

Das

Goddings

A.-L.

Blakemore

S.-J.

Viner

R. M.

(2018). Effectiveness of universal self-regulation–based interventions in children and adolescents: A systematic review and meta-analysis. JAMA Pediatrics, 172, 566–575.

34.

Razza

R. A.

Bergen-Cico

Raymond

(2015). Enhancing preschoolers’ self-regulation via mindful yoga. Journal of Child and Family Studies, 24, 372–385.

35.

Robertson

R. J.

Goss

F. L.

Boer

N. F.

Peoples

J. A.

Foreman

A. J.

Debayebeh

I. M.

Millich

N. B.

Balasekaran

Riechman

S. E.

Gallagher

J. D.

Thompkins

(2000). Children’s OMNI scale of perceived exertion: Mixed gender and race validation. Medicine & Science in Sports & Exercise, 32, 452–458.

36.

Ruggeri

Dancel

Johnson

Sargent

(2019). The effect of motor and physical activity intervention on motor outcomes of children with autism spectrum disorder: A systematic review. Autism, 24, 544–568.

37.

Saunders

Milyavskaya

Etz

Randles

Inzlicht

(2018). Reported self-control is not meaningfully associated with inhibition-related executive function: A Bayesian analysis. Collabra: Psychology, 4, 39.

38.

Scarpina

Tagini

(2017). The Stroop Color and Word Test. Frontiers in Psychology, 8, Article 557.

39.

Sollerhed

A.-C.

Apitzsch

Råstam

Ejlertsson

(2007). Factors associated with young children’s self-perceived physical competence and self-reported physical activity. Health Education Research, 23, 125–136.

40.

Stroop

J. R.

(1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, 643–662.

41.

Tomporowski

P. D.

Pesce

(2019). Exercise, sports, and performance arts benefit cognition via a common process. Psychological Bulletin, 145, 929–951.

42.

Tse

A. C. Y.

Anderson

D. I.

Liu

V. H. L.

Tsui

S. S. L.

(2021). Improving executive function of children with autism spectrum disorder through cycling skill acquisition. Medicine & Science in Sports & Exercise, 53, 1417–1424.

43.

Tse

A. C. Y.

Lee

P. H.

Chan

K. S. K.

Edgar

V. B.

Wilkinson-Smith

Lai

W. H. E.

(2019). Examining the impact of physical activity on sleep quality and executive functions in children with autism spectrum disorder: A randomized controlled trial. Autism, 23, 1699–1710.

44.

Uzefovsky

Allison

Smith

Baron-Cohen

(2016). Brief report: The go/no-go task online: Inhibitory control deficits in autism in a large sample. Journal of Autism and Developmental Disorder, 46(8), 2774–2779.

45.

Verburgh

Königs

Scherder

E. J. A.

Oosterlaan

(2014). Physical exercise and executive functions in preadolescent children, adolescents and young adults: A meta-analysis. British Journal of Sports Medicine, 48, 973–979.

46.

Visier-Alfonso

M. E.

Sánchez-López

Martínez-Vizcaíno

Jiménez-López

Redondo-Tébar

Nieto-López

(2020). Executive functions mediate the relationship between cardiorespiratory fitness and academic achievement in Spanish schoolchildren aged 8 to 11 years. PLOS ONE, 15, Article e0231246.

47.

Vitali

Robazza

Bortoli

Bertinato

Schena

Lanza

(2019). Enhancing fitness, enjoyment, and physical self-efficacy in primary school children: A DEDIPAC naturalistic study. PeerJ, 7, e6436.

48.

Weitlauf

A. S.

Gotham

K. O.

Vehorn

A. C.

Warren

Z. E.

(2014). Brief report: DSM-5 “levels of support”: A comment on discrepant conceptualizations of severity in ASD. Journal of Autism Developmental Disorders, 44, 471–476.

49.

Williams

K. E.

(2018). Moving to the beat: Using music, rhythm, and movement to enhance self-regulation in early childhood classrooms. International Journal of Early Childhood, 50, 85–100.

50.

Willoughby

Hudson

Hong

Wylie

(2021). Improvements in motor competence skills are associated with improvements in executive function and math problem-solving skills in early childhood. Developmental Psychology, 57, 1463–1470.

51.

Xue

Yang

Huang

(2019). Effects of chronic exercise interventions on executive function among children and adolescents: A systematic review with meta-analysis. British Journal of Sports Medicine, 53, 1397–1404. https://doi.org/10.1136/bjsports-2018-099825