Skills, such as locomotor and object manipulation, are not acquired innately as children grow up (Clark, 2005). Instead, planned programs are needed to facilitate the skills acquisition during early childhood (Logan et al., 2012). How should movement skills be trained in early childhood? Masters et al. (2013) suggested that the stage learning models of motor skill acquisition (Fitts and Posner, 1967) seem inappropriate during early childhood. Stage models suggest that a learner goes through an initial cognitive stage when learning motor skills during which methods to perform movements successfully are formulated, and the validity of perceived feedback is judged. In this stage, there is heavy reliance on cognitive resources and language ability to process and formulate rules and methods for movement performance; cognitive and verbal engagement is reduced later after extensive practice. Considering that cognitive resources and language skills are not fully developed in early childhood, it is unlikely that initiating a cognitive stage of skill learning as described above would be ideal for children at this stage.

In training motor skills during early childhood, it is therefore necessary to consider that skills may be best acquired through less engagement of verbal processes (e.g., instructions and rules). Masters and colleagues have developed this approach under the label of implicit motor learning (Masters, 1992; see Masters et al., 2019, for a recent review). Neurocognitive research has shown that young children tend to rely more on visual codes and less on verbal labels (Alloway et al., 2006) and rely less on executive resources when performing dual tasks compared to older children (Ang and Lee, 2010). Grounded on this evidence, an approach to training motor skills wherein practice errors are minimized may be suitable during early childhood.

When learners commit practice errors, they seek alternative movement solutions, form rules to support successful performance, and consequently draw upon their working memory resources (Maxwell et al., 2001). On the contrary, when practice errors are minimized, motor skills are learnt with less reliance on cognitive resources, which is suitable for children whose cognitive functions are still developing (Capio et al., 2012). This is the basis of the error-reduced approach, which has been shown to improve gross motor skills proficiency of children without dependence on short-term memory capacity (Capio and Eguia, 2020).

In earlier work, the error-reduced approach was tested specifically in the training of overhand throwing by children (Capio et al., 2013a). Practice errors were limited in the initial stage of learning by incrementally manipulating task difficulty (i.e., throwing at targets that were large enough to ensure high success rates). Movement performance was found to improve; children displayed the ability to engage in a secondary cognitive task (i.e., counting backward) without it disrupting movement performance. This suggested that movement performance was not dependent on cognitive resources. The findings were replicated in children with cognitive limitations (Capio et al., 2013b), providing further evidence that this approach is relatively less reliant on cognitive resources. Interestingly, children who trained with an error-reduced approach manifested a greater increase in spontaneous throwing activity during free play. Capio et al. (2013b) suggested that greater experiences of success during practice facilitated subsequent sustained engagement and motivation; these two factors have been shown to mediate the positive effect of movement training on cognitive development (Pesce et al., 2016). As such, the error-reduced training could be an approach that not only improves movement performance but also aids cognitive development. Greater experience of success during practice has been associated with enhanced self-efficacy (Muller and Roder, 2010), which suggests that it could contribute to psychosocial development in early childhood. While converging evidence from studies involving children of different abilities (i.e., Maxwell et al., 2017; Capio et al., 2018) has indicated the value of an error-reduced approach to motor skills training, the potential effects on cognitive and social development have yet to be verified.

It is evident that movement skills proficiency plays a crucial role in optimal cognitive and social development in early childhood. Children who have proficient motor skills tend to have positive outcomes in a number of child development domains. The evidence to date, however, has been largely drawn from studies of association, and the mechanisms underlying the apparent relationships between the motor domain and other domains of development continues to be poorly understood (Libertus and Hauf, 2017). With limited evidence from intervention studies, the antecedent role of movement proficiency in child development is an area that needs further exploration. It remains unclear whether enhanced movement proficiency can directly impact cognitive and social development in early childhood, or whether all developmental domains are similarly influenced by factors, such as general stimulation or the early childhood education environment.

From a theoretical perspective, it has been noted that motor development has been relatively neglected when trying to understand childhood behaviors (Rosenbaum, 2005; Iverson, 2010), even though movement enables exploration and expression. This may be due in part to limited evidence from robust experimental designs, which this present study aims to address. The aim of this study is to examine whether improved movement proficiency leads to benefits in the cognitive and social domains of child development. A fundamental movement skill training program, using the error-reduced learning approach, will be implemented and compared with typical kindergarten activities (i.e., active control). It is hypothesized that significant and meaningful improvements will be observed in gross motor skills proficiency from baseline to immediate and delayed time points following training; improvements will be greater among those in an error-reduced condition compared to those in the active control group. It is further hypothesized that improvements will be found in executive functioning and social competence from baseline to immediate and delayed time points following training; improvements will be greater for those who displayed larger gains in gross motor skills proficiency. Executive functions will be based on standardized tests of inhibitory control and working memory, while social competence will be based on standardized reports by teachers. Factors that could influence the main variables of interest will be accounted for.

The study will be implemented in Hong Kong, where children aged three to five years attend kindergarten classes. All local kindergartens follow the same curriculum which stipulates a learning area of physical fitness and health, inclusive of fundamental movement skills development (Curriculum Development Council, 2017). This is a two-arm (training vs. active control), group-randomized intervention study that is situated in the kindergarten setting; the intervention will be delivered in partnership with teachers.

A purposive sampling approach will be taken with the following inclusion criteria for participants: Children (1) aged three to five years, (2) enrolled in a local kindergarten within the same territory as the research center (i.e., New Territories and Hong Kong), and (3) have no diagnosed neurodevelopmental, medical, or orthopedic condition that is contraindicated to moderate intensity physical activity or that requires special educational needs support. In previous research, a six-week gross motor skills training program led to improved motor proficiency, with effect sizes ranging from 0.28 to 0.45 (Brian and Taunton, 2018). Taking a conservative approach, it was calculated that to achieve 80% power (two-tailed alpha at 0.05), with an effect size of 0.28, and accounting for between-group and within-group interactions, a total sample of 144 is required to compare two groups (i.e., training and control) at four testing points. Accounting for 10% potential attrition rate, 79 participants per group (N = 158) will be recruited. The participants will be randomly allocated as a group (i.e., one intact kindergarten class) to either training or active control. Typical kindergarten classes in Hong Kong have about 20 to 30 pupils; hence, three to four classes will be allocated to each group (i.e., training and active control). To control for a cluster effect and socioeconomic status being a potential confounder, we had recruited a local kindergarten with at least eight kindergarten classes from a middle-class area.

Random allocation will be at group level because the intervention is to be delivered in the context of kindergarten classes. Figure 1 illustrates the planned progress of participants through the study, following the flow diagram recommended by the Consolidated Standards of Reporting Trials (CONSORT; Schulz et al., 2010). Following receipt of parental informed consent, participants’ parents will be invited to complete a questionnaire ascertaining background information about the child, including typical daily activities and other learning (e.g., phonics and music) and physical (e.g., after-school sports and cycling) activities outside of the kindergarten. It is expected that parents will take about 20 to 30 min to complete the questionnaire. Measurement of covariate and outcome variables will be conducted across four time points, with the training program commencing after the first measurement (see Table 1 for schedule of implementation).

Preliminary work was conducted with local kindergarten teachers, which identified a number of key design considerations for the movement training program. First, training will target fundamental movement skills comprising: object control (i.e., throwing, catching, and kicking), locomotion (i.e., running, jumping, and hopping), and stability (i.e., turning, rolling, and balancing; Ulrich, 2000; Donnelly et al., 2017). These specific skills were identified to be consistent with the expectations of the curriculum guide for kindergartens in Hong Kong (Curriculum Development Council, 2017). Second, 20-min training sessions, three times per week, were deemed feasible in the context of the typical programming of local kindergartens. Third, the program duration was balanced to be both evidence-informed and fit for the local context. Significant movement skills improvement was evident following eight weeks of error-reduced training (Capio et al., 2018) or six weeks of training delivered by early childhood teachers (Brian et al., 2017). However, it was noted that eight weeks of fundamental movement skills training did not improve cognitive functioning of early primary school children (Lee et al., 2020). Therefore, the program duration in this study will be extended to enable motor development (as opposed to skill acquisition). Working around the school terms and breaks of the local curriculum, a 10-week intervention was deemed feasible in each of the two semesters of one school year. Finally, it was determined that teachers will be trained to deliver the intervention, supported by project assistants who had been trained in error-reduced approach to fundamental movement skills training. This was deemed an important consideration because teacher-led interventions in early childhood settings have been shown to benefit children (Jones et al., 2016). Teachers’ training will consist of the knowledge base for the error-reduced learning approach, the rationale for each lesson plan, and the performance measures that would serve as the basis for progression.

Errors will be minimized by specifically manipulating equipment size and training conditions, such that the initial sessions are suitably easy to allow successful movement performance. Progressive increments in task difficulty will follow when each child in a class is able to display successful performance ≥75% of the time at a task (e.g., task completion: 6 out of 8 attempts are successful). Equipment manipulation has been shown to be feasible in field-based studies (e.g., Buszard et al., 2014), and delivering error-reduced motor skills training in class settings is feasible (Capio et al., 2015). Class-based training activities were therefore designed in partnership with local kindergarten teachers, where equipment and environment set-up, procedures, and progressions were identified (see Table 2 for a sample).

The active control group will participate in the typical physical activities currently being implemented in their kindergartens in the context of the local curriculum guide. Details of the activities (including the parameters, such as repetitions and bouts) completed during the study period will be documented for both the active control and the training group to ensure that activities are delivered as planned (i.e., fidelity check). The instruction approach for the active control groups will also be identified and documented (e.g., direct instruction).

To measure immediate and sustained effects of training on the outcomes, assessments will be performed prior to training (baseline, M1), at the end of the first term (mid-training, M2), at the end of the second term (post-training, M3), and four months after the end of training (follow-up, M4). Participants’ performances in the tests for fundamental movement skills, executive function, and working memory will be recorded on videos and scored post-hoc by designated research staff who were not involved in the test administration (i.e., blinded to participants’ allocation to training or active control groups). A random sample of video recordings (i.e., 10%) will be scored by a second set of designated research staff to establish inter-rater reliability.

A two-group growth model for intervention studies (Muthen and Curran, 1997) will be fitted to the outcomes data. Growth curves will be fitted to the repeated measures with separate random intercepts and slopes estimated for the training and active control groups. The random intercepts capture individual differences in skills at commencement. The intercepts will be regressed on participant characteristics (i.e., sex, age, body composition, fine motor skills, and physical activity), with the latter acting as explanatory variables for the observed individual differences. A separate but parallel model will be estimated with the intercepts set at the final follow-up to examine the extent to which the same characteristics explain differences at that final data point.

Training effects will be evaluated using random slopes fitted to the data from the training and active control groups. For both groups, separate slopes will be fitted to describe (a) the normative growth observed and (b) growth that deviates from the normative pattern that can be attributed to training. Whether the mean of this second curve deviates significantly from zero serves as a test for the effects of training. Shared variance attributable to children belonging to the same class will be corrected for potential clustering effects by using the complex sampling procedure in Mplus 8.0 (Muthén and Muthén, 2017).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.