Implicit motor learning in primary school children: A systematic review

ABSTRACT

The aim of this study was to assess the current state of evidence and methodological quality of studies on implicit and explicit motor learning in both typically developing children and children with developmental disorders. A systematic literature review was conducted on the experimental literature published up to April 2020. A total of 25 studies were included. Studies were evaluated on methodological quality, paradigm used, and level of evidence. The results showed that implicit paradigms are as effective as explicit paradigms in both groups of children. Studies are predominantly experimental in nature involving mostly upper limb aiming tasks. The few studies that were performed outside the lab (n = 5) suggest superior efficacy of the implicit paradigm. Methodological quality varied between studies and was not always of sufficient standard to allow conclusions. In particular, manipulation checks were only performed in 13 studies (52% of all studies), limiting conclusions. Further progress can be made by focussing on improving methodological quality through retention testing by the inclusion of a control group, by the inclusion of a manipulation check, and via assessment of relevant co-variables, such as working memory, age, and motor competence.

KEYWORDS:

• Motor learning
• children
• implicit learning
• explicit learning
• systematic review

Introduction

Having an active lifestyle, which includes participation in sport, exercise and physical activity, contributes significantly to general health and well-being. An important prerequisite for this active lifestyle is certain base level of motor skill performance. The basis of these skills, also known as Fundamental Movement Skills, is acquired during childhood and early adolescence in home, school, and leisure settings (Gallahue & Donnelly, 2003; Gallahue et al., 2012). Learning these FMS early in life is crucial to sustain an active lifestyle in later life (Faigenbaum & Myer, 2012; Stodden et al., 2008). Physical education (PE) lessons, as well as training at sport clubs and in leisure time, are the predominant opportunities for children to structurally practice and learn FMS. Accordingly, PE teachers and trainers at sport clubs play a very important role to promote the development and mastery of these skills. Given the eminent importance of adequate motor skill performance, it is warranted that professionals apply methods that yield the most efficacy in learning.

Traditionally, motor skill learning is conceptualized as a succession through stages, in which the first stage is directed towards increasing awareness and gaining explicit knowledge about skill execution (Anderson, 1983; Fitts & Posner, 1967). This verbal-cognitive stage is frequently accompanied by extensive explicit instructions. All acquired knowledge needs to be kept active and available for processing and is subsequently manipulated and/or applied to the next attempt in order to improve performance. As such, this type of motor learning places a high demand on cognitive resources and, in particular, working memory. As motor learning progresses, motor control becomes more automated and less dependent on working memory availability (Masters, 1992). In contrast to this explicit mode of learning, it has been argued that an initial verbal-cognitive stage in motor skill learning is not necessary for motor learning to take place. Specifically, with implicit learning, a context is created that aims to prevent or minimize the accumulation of declarative knowledge. With this, unconscious, automated control processes to regulate movement execution are promoted (Masters, 1992).

Different experimental paradigms have been used to promote implicit motor learning for sport-related tasks or FMS. In this review, we will limit the discussion to those paradigms that 1- empirically, have been most tested and reported in the literature (also see Kal et al., 2018 for a systematic review on implicit and explicit motor learning among adults) and/or −2 theoretically, showed the highest consensus among experts as to whether these are implicit learning paradigms (Kleynen et al., 2014). This yielded the following four paradigms for implicit learning 1) dual-task learning, 2) errorless learning, 3) analogy learning, and 4) learning with an external focus of attention. The commonality of these paradigms is that they aim to reduce the degree to which conscious processing is needed when learning a motor task.

Dual-task learning was the first paradigm described to induce implicit learning in a sport-related skill (Masters, 1992). Participants practiced a primary motor task, in this case golf putting, while concurrently performing a secondary task, in this case a random letter generation task. The secondary task loads working memory and prevents the development of declarative knowledge. The results show that participants indeed learn the task and accumulate less knowledge compared to participants that do not perform the secondary task, which was confirmed in later studies (Liao & Masters, 2001; Maxwell et al., 2003). According to Masters (Masters, 1992), this exemplifies that the task is learned implicitly. However, a sidenote in these studies was that implicit learners displayed a lower level of performance in retention tests compared to explicit learners. Additionally, secondary tasks are strenuous and difficult to apply in settings like PE and sport practice. To counteract this, different paradigms have subsequently been developed that do minimize the development of declarative knowledge during motor learning but at the same time can be better applied in practice (also see Poolton & Zachry, (2007)).

The second paradigm, errorless learning, is based on the assumption that performance errors lead to an explicit hypothesis testing strategy (Maxwell et al., 2001). That is, following an error, participants actively engage in generating hypotheses on how to improve performance and test these in the next trial. Through this process explicit knowledge is acquired. With errorless learning the environment is constrained to minimize performance errors. This results in a limited need for hypotheses-testing, which in turn aims to reduce the involvement of working memory and the development of explicit knowledge (Maxwell et al., 2001).

In the third paradigm, analogy learning, the learner is instructed to use an analogy which integrates the complex structure of the to-be-learned skill into a simple biomechanical metaphor (Liao & Masters, 2001). It is assumed that this metaphor relies very little on the manipulation of explicit verbal information or rules, suggesting implicit learning to be promoted. An example of an analogy used to teach a basketball shooting task is “Shoot the ball as if you are trying to put cookies into a cookie jar on a high shelf” (Lam, Maxwell, Masters et al., 2009).

Both, errorless learning and analogy learning, have been shown to lead to similar, or even better, motor skill performance compared to explicit learning (Maxwell et al., 2001, p. 14,15,16,17; Lam et al., 2009; Chauvel et al., 2013; Lam et al., 2009; Lam et al., 2010; Savelsbergh et al., 2012). Nevertheless, evidence about the learning process that takes place is mixed (Kal et al., 2018). This learning process is inferred from the amount of declarative knowledge that is accumulated (which should be less following implicit learning) and/or by having participants perform a secondary, cognitively demanding task (performance following implicit learning should not be hindered by this when motor performance is indeed independent of cognitive recourses). A systematic review showed that most studies only include one measure of the learning process. Most often, there are no differences between the groups or differences are only apparent for the amount of declarative knowledge accrued. The evidence is especially inconsistent in errorless learning, where most studies also fail to demonstrate differences between groups regarding the amount of declarative knowledge. For example, in the study of Maxwell et al. (2001), this difference is only apparent when only statements that indicate the testing of hypotheses were compared, instead of all movement-related rules and facts. As a result, the learning process that encompasses the performance improvements remains unclear.

The fourth implicit motor learning paradigm we included is learning with an external focus of attention. In learning with an external focus, the attention of the learner is diverted towards the effect or outcome of the movement on the environment instead of to the movements of the body as is the case in learning with an internal focus of attention (Wulf et al., 1998). Research in healthy adults has consistently shown that motor performance and learning are enhanced when practicing with an external focus of attention (see (Wulf, 2013) for a review). Theoretically, these beneficial effects are thought to be due to automatic control processes that are promoted with an external focus of attention, whereas an internal focus of attention is supposed to interfere with this automatic control. This explanation is the key feature of the constrained action hypothesis (Wulf et al., 2001). Importantly, in dual-task conditions performance remained stable when the skill was performed concurrently with a secondary task after practice with an external focus of attention, whereas performance deteriorated following practice with an internal focus (Poolton et al., 2006; Wulf, 2013). This highlights that performance was indeed more automatic following an external focus of attention. It needs to be noted that a Delphi study among motor learning experts did not reach consensus as to whether an external focus of attention can be regarded as a paradigm for implicit learning, but there was a trend present (Kleynen et al., 2014, 2015). Nonetheless, based on the empirical similarities, and the rapidly increasing number of studies reporting on this paradigm, we have decided to include learning with an external focus of attention as a paradigm to promote implicit learning.

Taken together, research in adults indicates that implicit paradigms are just as effective, or even more effective, as explicit methods for improving motor skill performance, but questions concerning the learning process and methodology are present (Kal et al., 2018). It has been suggested that implicit methods are also suitable for motor skill learning in children (Masters et al., 2013). This is based on the assumption that implicit methods place a lower demand on cognitive resources, combined with evidence from basic motor skill learning studies that implicit learning is independent of age (Meulemans et al., 1998, p. 26,27; Reber, 1992; Reber, 1991; Thomas & Nelson, 2001). Apart from its suitability in typically developing children, implicit motor learning methods may be especially effective for children with difficulties in (motor) development, such as children with Developmental Coordination Disorder (DCD), intellectual disabilities (ID), or Cerebral Palsy (CP). It is known that these children often experience comorbid problems with cognitive functioning in general and working memory in particular, which may hinder their ability for explicit motor learning (Steenbergen et al., 2010). The aim of the present review is to systematically assess the current state of evidence and methodological quality of studies on implicit and explicit motor learning in typically developing children and children with motor restrictions. We will discuss these findings in relation to the main outcomes, the learning processes taking place and the ecological validity of the findings.

Methods

Our review was registered on PROSPERO (registration number CRD42018116141).

Search strategy

A systematic literature search was conducted using the following databases: PubMed, PsychINFO, SPORTDiscus, ERIC and Web of Science. The initial search was performed on 13^th of December 2019 and updated on 22nd of April 2020 (see appendix A for details of the search strategy). Finally, reference lists of included publications were screened for relevant articles.

Identification

The initial search identified a total of 625 records. After removal of all duplicates (N = 240) a total of 385 records remained. All titles and abstracts were screened and examined by one researcher (WdG. Based on title and abstract, another 353 publications were removed. Of the 32 potentially relevant articles the full-text was retrieved and screened by two researchers (WDG and RM) for final inclusion. The inclusion criteria for the systematic review were defined as:

1. The study presented experimental data including a comparison between an implicit learning paradigm (i.e., dual-task learning, errorless learning, analogy learning, external focus of attention) and an explicit-learning paradigm (i.e., explicit instructions, internal focus of attention, error-strewn learning) and with an acquisition/practice phase;

2. The study focused on complex motor/sport skills (i.e., balancing, throwing, etc.);

3. Participants are primary school children, aged 4–12 years old, either with typical development or with developmental disabilities (motor or cognitive).

Articles were included upon agreement between the two researchers. The third researcher (BS) was consulted for any discrepancies between the two researchers. The search was updated by one researcher (FvA), leading to the inclusion of two additional studies. A total of 26 publications were included in this review for further analysis (See Figure 1 for the PRISMA flow diagram).

Figure 1. Flow chart of the systematic literature search. (adapted from the PRISMA flow diagram (Moher et al., 2009))

Methodological quality

The quality of the studies was evaluated using a rating system adapted from Siebes, Wijnroks, and Vermeer (Siebes et al., 2002). We specifically selected this rating system for its suitability for studies other than randomised controlled studies and its previous use in systematic reviews on motor learning (Houwen et al., 2014). Studies were rated based on a list of criteria on sample characteristics, motor learning characteristics, and methodological aspects (see Table 1). A manipulation check was added to the list. This check is important to ensure that implicit learning indeed occurred. We therefore only included manipulation checks related to the learning process (i.e., dual-task condition, accumulation of declarative knowledge and/or relations with working memory capacity). All included publications were read full-text by two authors (WdG and FvA) and independently scored. Checklist scores were then compared and discussed until consensus was reached (see Table 2). Based on its very low methodological quality, one study (Samsudin & Low, 2017) was deleted in the remainder of the review.

Table 1. Quality of included studies

Criteria		Score	Description
Sample characteristics	Nr and age of participants	1 0	Specified (N, mean/range) Unspecified
	Use of control or contrast group	2 1 0	Experimental, contrast and control group Experimental and contrast/control group No control/contrast group
Intervention characteristics	Contents description	2 1 0	Detailed description Summary only Very limited information
	Frequency, intensity and duration	2 1 0	Detailed description Some information has been given Very limited information
Methodological aspects	Reliability	2 1 0	Explicitly reliable: the reliability test has been published Standardized: the measurement procedure is described Unreliable
	Validity	2 1 0	Explicitly valid: the validity test has been published Validity unknown, but appears reasonable Not valid
	Internal validity	2 1 0	High internal validity Fair internal validity Low internal validity Internal validity can be reduced by various study variations such as subject assignment, contamination, co-intervention, and blind assessment
	Manipulation check	1 0	Manipulation check No manipulation check
	External validity	2 1 0	Generalization is plausible Some possibilities for generalization No information about generalization is given
	Statistical evaluation	2 1 0	Detailed description of statistical methods, p values and F values are presented Only a brief description of statistical methods is described, some values are presented No statistical evaluation
	Total score	18

Table 2. Methodological quality

	Sample characteristics		Intervention characteristics		Methodological aspects						Total
	Nr and age of participants	Use of control or contrast group	Contents description	Frequency, intensity and duration	Reliability	Validity	Internal validity	Manipulation check	External validity	Statistical evaluation
	1	2	2	2	2	2	2	1	2	2	18
Errorless learning
Capio, Poolton, Sit, Holmstrom et al., 2013	1	1	2	2	2	1	2	1	2	1	15
Capio, Poolton, Sit, Eguia et al., 2013	1	1	2	2	2	1	2	1	2	2	16
Maxwell et al., 2017	1	1	2	2	1	1	1	1	1	1	12
Van Abswoude et al., 2015	1	1	2	1	1	1	1	1	1	2	12
Analogy learning
Chatzopoulos et al., 2020	1	1	1	1	2	1	1	0	2	2	12
Tse et al., 2017	0	1	2	1	2	1	1	1	1	1	11
Tse & Masters, 2019	1	2	2	2	2	1	1	0	1	2	14
External focus of attention
Agar et al., 2016	1	1	1	1	1	1	0	0	1	1	8
Becker & Smith, 2013	0	1	1	2	1	1	1	0	1	2	10
Brocken et al., 2016	1	1	2	1	1	1	1	1	1	2	12
Chiviacowsky et al., 2013	1	1	1	1	1	1	1	0	1	2	10
Chow et al., 2014	1	2	2	2	1	1	2	0	1	1	13
Emanuel et al., 2008	1	1	2	1	1	1	1	0	1	2	11
Flores et al., 2015	1	2	1	2	1	1	1	0	2	1	12
Hadler et al., 2014	0	2	1	1	1	1	1	0	1	1	9
Krajenbrink et al., 2018	1	1	2	1	1	1	1	1	1	1	11
Perreault & French, 2015	0	1	1	1	1	1	1	0	1	1	8
Perreault & French, 2016	0	2	2	2	1	1	1	0	1	1	11
Petranek et al., 2019	0	1	2	2	2	1	2	0	2	2	14
Saemi et al., 2013	0	1	2	1	1	1	1	0	1	2	10
Samsudin & Low, 2017	0	1	0	1	0	0	0	0	0	1	3
Teixeira da Silva et al., 2017	1	1	1	1	1	1	1	0	1	1	9
Tse, 2017	1	2	2	2	1	1	2	0	1	1	13
Tse & van Ginneken, 2017	0	2	2	2	1	1	1	1	1	2	13
Van Cappellen – van Maldegem et al., 2018	1	1	2	1	1	1	1	1	2	2	13
Wulf et al., 2010	0	1	1	1	2	1	2	0	1	2	11

Data extraction and synthesis

From the included studies, the following information was independently extracted by two authors (WdG, FvA): 1. Sample characteristics; 2. Intervention characteristics (task, paradigm, intervention details) 3. Measures and outcomes; 4. Co-variables; 5. Results. The measures and outcomes include how the performance of the skill is assessed (based on performance accuracy (product-oriented) and/or movement quality (process-oriented)), realised intervention (i.e., evidence that the intended intervention took place, like adherence to instruction, successful manipulation of errors) as well as the measures that were used as manipulation check (i.e., dual-task performance, number of rules reported). Results are described for study characteristics and methods, differences between implicit and explicit learning (during practice, retention, transfer and realized intervention), the learning process, methodological quality and additional results. By scoring the methodological quality of the studies, we became aware of the different methodological approaches that are used in the field, methodological issues regarding the learning processes (as can be seen in Table 2), and different theoretical foundations (for example, the different views regarding an external focus of attention, also mentioned in the introduction). We therefore decided for a narrative synthesis to describe the results.

Results

Overall study characteristics

A total of 25 studies were included. Eighteen studies included typically developing children and seven studies included children with developmental disabilities. None of the included studies adopted the dual-task learning paradigm, four studies used the errorless learning paradigm and three studies used the analogy learning paradigm. The majority of the included studies (n = 18) examined the effect of an external focus of attention. In the following paragraphs, we describe the results separately for typically developing children (Table 3), and a-typically developing children (Table 4).

Table 3. Included studies with typically developing children

Errorless learning
Reference	Sample characteristics	Intervention characteristics				Outcome measures and covariables	manipulation check	Results
		Task	Implicit	Explicit	Intervention
Capio, Poolton, Sit, Holmstrom et al., 2013	EL-HA N = 26 Age = 9.8 (1.0) EL-LA N = 20 Age = 8.6 (1.0) ES-HA N = 28 Age = 9.5 (1.0) ES-LA N = 34 Age = 8.7 (0.6)	Over-hand throwing /aiming	Gradual decrease target size, staring with largest target	Gradual increase target size, starting with smallest target	Pre-Test – 10 trials Practice: 3 weeks, 1 practice session of 2*20 trials per week (total 120 trials) 1 week later: Retention – 10 trials Dual-task – 10 trials	Outcome: Accuracy Movement form (TGMD-2) Covariables: Gender Intervention: # errors	Dual task: counting backwards	Intervention #errors EL < ES Practice Accuracy: EL > ES, HA > LA, boys> girls Retention Accuracy: EL > ES, only LA groups improve Movement form: EL > ES, only girls improve Dual-task EL > ES Accuracy in ES groups ↓ Accuracy in EL groups =
Maxwell et al., 2017	EL-HA N = 13 Age = 10.1 (0.5) EL-LA N = 11 Age = 9.5 (0.5) ES-HA N = 10 Age = 9.9 (0.6) ES-LA N = 11 Age = 9.7 (0.7)	Golf putting	Gradual increase target distance, starting close	Gradual decrease target distance, starting far	Practice: 6 blocks of 50 trials on 1 day (total 300 trials) Same day: Retention 1–50 trials Dual-task – 50 trials Retention 2–50 trials	Outcome: Putting score	Dual task: tone counting Verbal report: # explicit rules	Practice EL = ES, HA > LA Retention EL-HA = ES-HA EL-LA > ES-HA Dual-task EL-HA = ES-HA, putting score =  EL-LA > ES-LA, putting score EL-LA = , putting score ES-LA ↓ Verbal report EL report less rules than ES
EL: Errorless; ES: Error strewn; -HA: High ability; -LA: Low ability; TGMD: Test of Gross Motor Development.
Analogy learning
Reference	Sample characteristics	Intervention characteristics				Outcome measures and covariables	manipulation check	Results
		Task	Implicit	Explicit	Intervention
Chatzopoulos et al., 2020	AI N = 22 Age = 5.7 (0.7) EI N = 21 Age = 5.5 (0.6)	Run, jump, gallop, balance	1 analogy per skill	4 explicit instructions per skill	Pre-Test: TGMD, 3 trials per skill Practice: 4 lessons (2 per week), 3–5 min per skill Retention: TGMD, 3 trials per skill	Outcome: TGMD score movement form Centre of mass displacement (balance)		Retention Balance: AI > EI Jump, gallop, run: AI = EI
Tse et al., 2017	AI N = 15 EI N = 17 Age total group = 5.8 (0.9)	Rope skipping	6 explicit instructions and 2 analogies (for critical components of skill)	8 explicit instructions	Pre-Test – 1 minute rope skipping Practice: 2 weeks, 3 practice sessions of 40 attempts to skip (total 120 trials). Instructions at start of each lesson 2 days later: Retention – 1 minute rope skipping Dual-task – 1 minute rope skipping	Outcome: Number of skips Rope skipping posture	Dual task: counting backwards	Practice AI > EI in session 1, both #skips and posture Retention AI = EI, both improve #skips and posture Dual-task AI > EI, #skipps and posture for AI = and for EI ↓
AI: analogy instructions; EI: Explicit instructions; TGMD: Test of Gross Motor Development.
Attentional focus
Reference	Sample characteristics	Intervention characteristics				Outcome measures and covariables	manipulation check	Results
		Task	Implicit	Explicit	Intervention
Agar et al., 2016	48 children; age 5–12 years. Allocation over groups EF-Y, EF-O, IF-Y, IF-O unclear	Shuffle-board with a stick	Non-standardized external instructions, cues and feedback	Non-standardized internal instructions, cues and feedback	Practice: 1 day with 3 blocks of 10 practice trials (total of 30 trials). Feedback after every trial. 1–2 days later: Retention – 10 trials Transfer (opposite hand) – 10 trials	Outcome: Accuracy		Practice EF = IF Retention EF = IF Transfer EF = IF
Becker & Smith, 2013	EF-S (N = 12) IF-S (N = 12) EF-C (N = 12) IF-C (N = 12) Age 8–10	Riding a double pedalo with (simple task) or without (complex task) handles	Focus on pushing the boards forwards	Focus on pushing your feet forward	Practice: 2 days with 10 trials (total of 20 trials). Instructions before each trial and reminders halfway each trial. 1 day later: Retention – 4 trials	Outcome: Time to complete 7 metre Covariables:- Gender Intervention: Adherence (5 point likert scale, focus was used never-always)		Practice EF = IF Retention EF > IF, only for males in complex task, otherwise EF = IF Intervention Focus was used most of the time or always
Brocken et al., 2016	EF-Y (N = 15) IF-Y (N = 15) Age Y = 8.9 (0.5) EF-O (N = 15) IF-O (N = 15) Age O = 11.7 (0.4)	Golf putting	Move the golf club like a pendulum	Move the arms like a pendulum	Pre-Test – 30 trials Practice: 1 day, 8 blocks of 10 practice trials (total 80 trials). Instructions before every block. 1 day later: Retention – 30 trials	Outcome: Accuracy Covariable:- Age	Role of V-WM on learning	Retention EF > IF, only EF groups improve accuracy Covariable No effect of age Manipulation check No effect of V-WM
Chow et al., 2014	IF (N = 12) EF (N = 12) C (N = 12) Total age = 9.9 (0.3)	Standing broad jump	4 external instructions	4 internal instructions Control group: 4 motivational instructions	Pre-Test – 4 trials Practice: 3 sessions of 6 trials in 1 week (total 18 trials), reminders before each trial. 1 week later: Retention – 4 trials	Outcome: Jump distance Kinematic data Intervention: Adherence		Retention EF = IF = C on both measures Intervention All participants adhered to the instruction and indicated they were helpful
Emanuel et al., 2008	IF N = 17 Age = 9.0 (0.3) EF N = 17 Age = 9.1 (0.4)	Dart throwing	6 external instructions + 1 reminder	4 internal instructions + 1 reminder	Instructions before practice. Practice: 5 blocks of 10 throws (total 50 trials). Reminder after each block. 1 day later: Retention – 20 trials Transfer (further distance) – 20 trials	Outcome: Accuracy Consistency Intervention: Adherence		Practice No consistent improvement in accuracy or consistency Retention EF = IF Transfer EF < IF Intervention 29.4% non-adherence, shift from EF to IF
Flores et al., 2015	DEF-Y (N = 14) PEF-Y (N = 14) IF-Y (N = 14) C-Y (N = 14) DEF-O (N = 13) PEF-O (N = 13) IF-O (N = 13) C-O (N = 13) Age Y = 6 Age O = 10	Riding a double pedalo (without handles)	DEF: focus on a marker at the finish line PEF: focus on pushing the platforms forward	IF: focus on pushing your feet forward C: No instructions	Pre-Test – 1 trial Practice – 20 trials 1 day later: Transfer 1 (fast) – 5 trials Transfer 2 (hands on head) – 5 trials	Outcome: Time Covariable:- Age		Practice IF < all other groups Transfer 1 IF < DEF and PEF, DEF > C Transfer 2 DEF > IF
Hadler et al., 2014	IF (N = 15) EF (N = 15) C (N = 15) Total age = 11.0 (0.7)	Forehand tennis stroke	Movement of the racquet	Movement of the arm C: no instructions	Practice: 60 trials. Reminder after each 10^th trial 2 days later: Retention – 10 trials Transfer (from other side of court) – 10 trials	Outcome: Accuracy		Practice EF = IF = C Retention EF > IF & C Transfer EF > IF, C = EF & IF
Krajenbrink et al., 2018	IF & EF Total 162 children, age 10.6 (1.2)	Slinger-ball throw	3 external instructions directed at the ball	3 internal instructions directed at the arm	Pre-Test – 20 trials Practice: 80 trials, reminder every 10 trials and repetition of instruction every 20 trials 1 week later: Retention – 20 trials Dual-task – 20 trials	Outcome: Accuracy Intervention:- Adherence	Dual task: tone counting Role of V-WM and VS-WM on learning	Practice EF > IF Pre-Test-Retention EF = IF, both improve Manipulation check Dual task: EF = IF, performance on motor task ↑ and performance on cognitive task↓ No effects for WM Intervention Divers responses, most focus on instructed aspects
Perreault & French, 2015	IF (N = 14) EF (N = 14) Total age 9–11 years	Basketball free-throw (modified)	One of four external feedback statements each time	One of four internal feedback statements each time	Practice: 2 days with 5 blocks of 10 trials per day (100 trials total). Feedback after each 3^rd trial. 1 day later: Retention – 20 trials	Outcome: Accuracy Intervention: Adherence		Practice Only in block 5 EF > IF Retention EF>IF in second retention block Intervention Low reported use of instruction, no crossover
Perreault & French, 2016	IF (N = 14) EF (N = 14) C (N = 14) Total age 9–11 years	Basketball free-throw (modified)	One external cue on day one and one additional external cue on day 2	One internal cue on day one and one additional internal cue on day 2	Practice: 2 days with 5 blocks of 10 trials per day (100 trials total). Reminder after each block, cues at start of each day. 2 days later: Retention – 20 trials	Outcome: Accuracy Intervention:- Adherence		Practice EF = IF = C Retention EF = IF = C Intervention Small number used respective cues, better participants report more external cues
Petranek et al., 2019	EF-HF (N = 14) EF-LF (N = 15) IF-HF (N = 18) IF-LF (N = 18) Total age 6–7 years	Overhand throwing	One statement selected from 4 external focus feedback statements at each feedback moment and video instruction with external focus	One statement selected from 4 internal focus feedback statements at each feedback moment and video instruction with internal focus	Pre-Test – 10 trials and knowledge test Practice: 2 weeks, 2 days per week, 6 blocks of 10 trials (60 trials total). Feedback after ever 2^nd (HF) or 5^th (LF) trial Directly following practice: Retention 1–10 trials Knowledge post-test 1 week later: Retention 2–10 trials Knowledge retention test Transfer (aim at balloons) – 10 trials	Outcome: Movement form Knowledge about movement (selecting correct movement form from pictures)		Practice EF-HF > EF-LF block 2&3 EF-LF > EF-HF block 4, 5, & 6 IF-HF > IF-LF Retention and transfer IF > EF EF-LF > EF-HF IF-HF > IF-LF Knowledge All groups improve IF > EF at retention HF > LF at retention
Teixeira da Silva et al., 2017	IF & EF Total 38 children, age 9.5 (0.8)	Classical ballet pirouette	Focus on spotting point on the wall and fix gaze as long as possible	Focus on initial position of head relative to the wall and keep in that position as long as possible	Pre-Test – 2 trials Practice: 15 trials 2 days later: Retention – 5 trials Transfer (turn other way) – 5 trials	Outcome: Extend of rotation Intervention: Adherence		Practice EF > IF Retention EF > IF Transfer EF > IF Intervention Responses not related to attentional focus
Tse & van Ginneken, 2017	IF-HR (N = 10) IF-LR (N = 10) EF-HR (N = 10) EF-LR (N = 10) C-HR (N = 10) C-LR (N = 10) Total age = 10.4 (1.8)	Dart throwing	6 external instructions + 1 reminder	4 internal instructions + 1 reminder	Practice: 1 day, 5 blocks of 10 trials (total 50 trials). Reminder before each block Transfer (further) – directly following practice 1 week later: Retention – 10 trials	Outcome: Accuracy Intervention: Adherence	Role of MSRS	Practice IF = EF = C Transfer IF-HR > EF-HR & C-HR EF-LR > IF-LR & C-LR Retention IF-HR > EF-HR & C-HR EF-LR > IF-LW & C-LR Intervention 95% adhered to instructions
Wulf et al., 2010	EF100% (N = 12) EF33% (N = 12) IF100% (N = 12) IF33% (N = 12) Total age 10–12 years	Soccer throw-in	One of 8 external feedback statements each time	One of 8 internal feedback statements each time	Practice: 30 practice trials, feedback after every trial (100%) or after every 3^rd trial (33%) Directly following practice: Retention 1–5 trials Transfer 1 (closer target) – 5 trials 1 day later: Retention 2–5 trials Transfer 2 (closer target) – 5 trials	Outcome: Movement form Accuracy		Practice EF100 = EF33 = IF100 = IF33 on both measures Retention 1 to 2 EF100 = EF33 = IF100 = IF33 on both measures Transfer 1 to 2 EF100 > all other groups for movement form, no differences for accuracy

EF: External focus; IF: Internal focus; DEF: Distal external focus; PEF: Proximal external focus; C = Control group; V-WM: verbal working memory; VS-WM: visuospatial working memory; -Y: younger age group; -O: older age group; -HR: High reinversters; LR: Low reinvesters; -S: Simple task; -C: complex task; -HF: High frequency; LF: Low frequency; MSRS: Movement Specific Reinvestment Scale;

Table 4. Included studies with A-typical populations

Reference	Sample characteristics	Intervention characteristics				Outcome measures and covariables	manipulation check	Results
		Task	Implicit	Explicit	Intervention
Errorless learning
Capio, Poolton, Sit, Eguia et al., 2013	Children with mild ID (IQ 50–70) EL N = 18 Age = 7.7 (2.3) ES N = 21 Age = 7.1 (2.0)	Overhand throwing/ aiming	Gradual decrease target size, staring with largest target	Gradual increase target size, starting with smallest target	Pre-Test – 10 trials 1 week before practice Practice: 4 weeks, 1 practice session of 3*10 trials per week (total 120 trials) 1 week later: Retention test – 10 trials Dual-task – 10 trials	Outcome:- Accuracy Movement form (TGMD-2) Throwing frequency in free play Intervention: # errors	Dual-task: singing a children’s song	Retention Accuracy: EL > ES, both improve Movement form: EL > ES, both improve Manipulation check EL > ES Accuracy in ES groups ↓ Accuracy in EL groups =  Intervention #errors EL < ES Throwing frequency EL > ES, both improve
Van Abswoude et al., 2015	Children with CP, GMFCS II–IV. Groups matched on motor (GMFCS & Pegboard) and cognitive (VWM) abilities EL N = 18 Age = 9.8 (3.7) ES N = 20 Age = 9.3 (4.3)	Aiming task (boccia)	Gradual decrease target width, staring with widest target	Gradual increase target width, starting with most narrow target	Pre-Test – 10 trials Practice – 4 blocks of 20 trials (total 80 trials) Same day: Retention 1–10 trials Dual-task – 10 trials Retention 2–10 trials	Outcome: Accuracy	Dual-task: counting Role of VWM	Pre-Test-retention2 EL = ES, both improve Manipulation check Retention 1 – dual task: EL = ES, performance on motor task = , performance on cognitive task ↓ VWM influenced learning across groups
Analogy learning
Reference	Sample characteristics	Intervention characteristics				Outcome measures and covariables	manipulation check	Results
		Task	Implicit	Explicit	Intervention
Tse & Masters, 2019	Children with ASD (IQ > 50). Groups matched for WM AL-vis N = 12 Age = 9.9 (1.2) AL-verb N = 12 Age = 10.3 (1.0) EI N = 12 Age = 9.9 (1.2) C N = 12 Age = 10.8 (1.1)	Modified basketball shooting task	AL-vis: look at a picture of putting cookie in a jar AL-verb: Shoot as if you are trying to put cookies into a cookie jar on a high shelf	Two internal focus instructions	Practice: 6 blocks of 15 trials (total 90 trials). Instructions twice before each block 1 day later: Retention – 15 trials Transfer (further distance) – 15 trials	Outcome:- Accuracy		Practice C < all others, accuracy C = ; other groups ↑ Retention AL-vis > AL-verb & EI & C AL-verb & EI > C Transfer AL-vis > AL-verb & EI & C AL-verb & EI > C
Attentional focus
Reference	Sample characteristics	Intervention characteristics				Outcome measures and covariables	manipulation check	Results
		Task	Implicit	Explicit	Intervention
Chiviacowsky et al., 2013	Children with mild ID (IQ 50–70). (13 boys/11 girls); age 10–14 years (12.21 ± 1.31 years)	Overhand (beanbag) throw	Focus on movement of beanbag	Focus on movement of throwing hand	Practice: 40 trials, reminder after each 3^rd trial One day later: Retention – 10 trials Transfer (further distance) – 10 trials	Outcome: Accuracy		Practice EF = IF Retention EF > IF Transfer EF > IF
Saemi et al., 2013	Children with ADHD-C, low skilled throwers EF (N = 10) IF (N = 10) Total age = 10.1 (0.9)	Throwing skill/ aiming (tennis ball)	Concentrate on the ball, particularly the landing location of the ball	Concentrate on the motion of your hand and wrist that is throwing the ball	Practice: 6 blocks of 30 trials (total 180 trials). Reminder before each block. 2 days later: Retention – 10 trials	Outcome: Accuracy		Practice EF > IF Retention EF > IF
Tse, 2017	Children with ASD, IQ > 70, TGMD-2 object control > 40 IF N = 22 Age = 10.1 (1.2) EF N = 22 Age = 10.0 (0.9) C N = 21 Age = 10.1 (1.0)	Overarm throwing skill (beanbag)	2 short external instructions on target and beanbag flight	2 extensive internal instructions on arm position and elbow movement	Practice: 5 blocks of 10 throws (total 50 trials). Reminder before each block 1 day later: Retention – 10 trials Transfer (further distance) – 10 trials	Outcome: Accuracy Intervention: Adherence		Practice EF = IF = C, all groups improve Retention IF > EF & C Transfer IF > EF & C Intervention 90.7% adhered to the focus
Van Cappellen – van Maldegem et al., 2018	Children with DCD, MABC < 16^th percentile, IQ > 70 EF N = 15 Age = 7.3 (1.9) IF N = 12 Age = 6.8 (1.1)	Slingerball	Standardized feedback with one of 5 external statements each time	Standardized feedback with one of 5 internal statements each time	Pre-Test – 20 trials Practice: 70 trials on day 1, 60 trials on day 2 one week later (total 130 trials). Feedback every 30 trials 1 week later: Retention – 20 trials	Outcome: Accuracy	Role of WM on learning	Retention IF = EF Manipulation check VS-WM role in EF and not in IF No role V-WM

CP: Cerebral Palsy; TD: Typically Developing; ID: Intellectual Disability; EL: Errorless; ES: Error strewn; AL: analogy learning; EI: Explicit instructions; EF: External focus; IF: Internal focus; C = Control group; AL-vis: visual analogy; AL-verb: verbal analogy; V-WM: verbal working memory; VS-WM: visuospatial working memory; ADHD-C: Attention Deficit Hyperactivity Disorder Combined; ASD: Autism Spectrum Disorder; DCD: Developmental Coordination Disorder; TGDM: Test of Gross Motor Development

Errorless learning

Study characteristics and methods

Two included studies used the errorless learning paradigm. Practice was either individually or in pairs. Errorless learning was applied by gradually decreasing target size, or increasing target distance. Both studies also divided children in low- and high-motor ability at the start of the study. In the study of Capio, Poolton, Sit, Holmstrom et al. (2013) this was based on the pre-test score, and in the study of Maxwell et al. (2017) this was based on the assessment of FMSs.

Implicit vs. explicit (practice-retention-transfer-intervention)

Both studies showed that for participants with low-motor ability errorless learning was more beneficial compared to error strewn learning. There was no difference between the groups for children with high ability (Capio, Poolton, Sit, Holmstrom et al., 2013; Maxwell et al., 2017). In addition, Capio, Poolton, Sit, Holmstrom et al. (2013) showed that the quality of the movement improved more for girls in the errorless group. Both studies successfully manipulated the amount of errors between groups.

Manipulation check

Both studies included a dual-task condition. Capio, Poolton, Sit, Holmstrom et al. (2013) showed that performance was stable for children in the errorless group, whereas performance deteriorated for children in the error strewn group. Maxwell et al. (2017) showed similar results, but only for children with low ability. This study also showed that children in the errorless groups reported less rules.

Methodological quality

The total quality score of the studies was 12 (Maxwell et al., 2017) and 15 (Capio, Poolton, Sit, Holmstrom et al., 2013), respectively. None of the studies included a control group. Validity and reliability of the studies is reasonable at least. Additionally, the study of Capio, Poolton, Sit, Holmstrom et al. (2013) included blinded assessment and the use of a valid and reliable measure to assess movement quality, which contributed to the methodological quality of the study.

Additional results and practical implication

The study of Capio, Poolton, Sit, Holmstrom et al. (2013) was performed during physical education classes. Even though the setting was still experimental, children practiced in pairs in a separate part of the gym. This suggests that errorless learning could be used in practice.

Analogy learning

Study characteristics and methods

Two studies examined the effect of analogy learning on complex motor skills in children (Chatzopoulos et al., 2020; Tse et al., 2017). Chatzopoulos et al. (2020) performed the study at a preschool, with practice for the entire class. In the study of Tse et al. (2017) the acquisition phase was performed individually in a public park. Both studies examined motor learning by determining movement quality and Tse et al. (2017) also included the number of skips. In the study of Tse et al. (2017) instructions were provided at the start of each practice day and consisted of eight explicit instructions in the explicit group. In the analogy learning group two of those were replaced by an metaphor. In the study of Chatzopoulos et al. (2020) children received either four explicit instructions or one analogy instruction at the start of practice of a specific skill.

Implicit vs. explicit (practice-retention-transfer-intervention)

Tse et al. (2017) showed that the analogy group performed better at the start of practice compared to the explicit instruction group, but this difference disappeared at retention. Chatzopoulos et al. (2020) only showed better performance for the analogy learning group on the retention test for balancing, but not for the other skills. Neither of the studies checked if the intended intervention was realized.

Manipulation check

Tse et al. (2017) showed stable performance in the dual-task condition for both the number of skips and the quality of the movement in the analogy group, but not in the explicit instruction group.

Methodological quality

The total score on methodological quality of the studies was 11 (Tse et al., 2017) and 12 (Chatzopoulos et al., 2020). Neither of the studies included a check on the realized intervention or a control group. Both studies score high on reliability by reporting inter-rater reliability and using blinded assessment of movement quality.

Additional results and practical implication

The acquisition phase of both studies was performed in a real-life setting, indicating the practical applicability of analogy learning of complex skills in young children.

External focus of attention

Study characteristics and methods

The majority of the studies (n = 14) manipulated the focus of attention to promote implicit learning in a large variety of tasks. Only three of these studies include the theory on implicit learning in their rationale (Brocken et al., 2016; Krajenbrink et al., 2018; Tse & van Ginneken, 2017) and the remainder of the studies describe the constrained action hypothesis as their theoretical framework. Attentional focus was manipulated by using instructions and/or feedback with an internal or external focus of attention. In addition, five studies also included a control condition in which no additional instructions with a specific attentional focus were provided (Chow et al., 2014; Flores et al., 2015; Hadler et al., 2014; Perreault & French, 2016; Tse & van Ginneken, 2017). One study also included a comparison between a proximal (close to the person) and distal (directed towards a goal) external focus of attention (Flores et al., 2015), and two studies also manipulated the frequency of feedback for both attentional foci (high vs. low frequency)(Petranek et al., 2019; Wulf et al., 2010). Eight studies included a transfer test, which are described in (Table 3).

Implicit vs. explicit (practice-retention-transfer-realized intervention)

During practice, seven studies did not show any difference in performance between practice with an internal or external focus of attention (Agar et al., 2016; Becker & Smith, 2013; Emanuel et al., 2008; Hadler et al., 2014; Perreault & French, 2016; Tse & van Ginneken, 2017; Wulf et al., 2010) or a control condition (Chow et al., 2014; Hadler et al., 2014; Tse & van Ginneken, 2017). Three studies showed better performance in the groups that practiced with an external focus of attention (Flores et al., 2015; Krajenbrink et al., 2018; Teixeira da Silva et al., 2017). The study of Perreault and French (2015) only showed better performance of the external focus group in one of the practice blocks. On the retention test, four studies consistently showed better performance of the external focus of attention groups compared to the internal focus of attention groups (Brocken et al., 2016; Hadler et al., 2014; Perreault & French, 2015; Teixeira da Silva et al., 2017) (and control group in the study of Hadler et al. (2014)). In contrast, six studies did not show any differences between the groups (Agar et al., 2016; Chow et al., 2014; Emanuel et al., 2008; Krajenbrink et al., 2018; Perreault & French, 2016; Wulf et al., 2010). Two studies showed inconsistent results (Becker & Smith, 2013; Tse & van Ginneken, 2017). For example, Becker and Smith (2013) only found a benefit of an external focus of attention in the complex task of riding a pedalo without handles, but not for riding the pedalo with handles. One study showed better performance during retention for the internal focus groups (Petranek et al., 2019). In the transfer tests, two studies showed better performance following practice with an external focus of attention (Hadler et al., 2014; Teixeira da Silva et al., 2017), whereas two studies (Emanuel et al., 2008; Petranek et al., 2019) showed better performance in the internal focus group. One study did not show any differences between the groups (Agar et al., 2016). Three studies showed inconsistent results in the transfer test (Flores et al., 2015; Tse & van Ginneken, 2017; Wulf et al., 2010), for example, in the study of Wulf et al. (2010), a better performance was only found for the external focus group that received feedback after every trial, and the effect was only apparent in movement form and not movement accuracy.

Seven studies included a check on the realized intervention. Three of those reported good adherence to the instructed attentional focus (Chow et al., 2014; Krajenbrink et al., 2018; Tse & van Ginneken, 2017). In contrast, Emanuel et al. (2008) found that 29.4% of the participants did not adhere to the instructed focus, with most of the children shifting from the instructed external focus to an internal focus. Similar results were reported by Becker and Smith (2013), where children in the internal focus group more often adopted the designated focus than children in the external focus group. In the studies of Perreault and French (2015, 2016) children did not shift their focus, but only a small number of children reported to use the instructed attentional focus. Adherence was unrelated to the results described during practice, retention or transfer.

Manipulation check

The study of Krajenbrink et al. (2018) included a dual-task condition as a manipulation check, but did not show any differences between the groups. Two studies also examined the role of working memory capacity on learning, but they both showed no effect (Brocken et al., 2016; Krajenbrink et al., 2018)

Methodological quality

The total quality score of the studies varied from 8 to 14. None of the studies score maximal on reliability and validity, most often because of the lack of a blinded assessment, standardised or validated measures, or risk of co-intervention (for example, by including general instructions that may influence the focus of attention of the participants). Five of the studies included a control group and two studies included a manipulation check. In the majority of the studies, detailed description about the provided feedback, attentional focus instructions or general task instructions is missing.

Additional results and practical implication

One study performed the experiment with groups of children during physical education classes, but showing mixed results (Petranek et al., 2019). Therefore, it remains unclear how the results of practice with an internal or external focus of attention translate to a more practical setting.

Children with a-typical development

Study characteristics and methods

Seven studies included children with a specific diagnosis or disorder as their target population. Two studies applied an errorless learning method (Capio, Poolton, Sit, Eguia et al., 2013; Van Abswoude et al., 2015), one used analogy learning (Tse & Masters, 2019), and four studies manipulated the focus of attention (Chiviacowsky et al., 2013; Saemi et al., 2013; Tse, 2017; Van Cappellen – van Maldegem et al., 2018). In four studies, the training was performed individually in the presence of the experimenter(s). In the study of Capio, Poolton, Sit, Eguia et al. (2013) children practiced in pairs under supervision of teacher assistants and in the study of vVan Cappellen – van Maldegem et al. (2018) the experiment took place at the physical therapist practice and the therapist performed the experimental procedures.

Implicit vs. explicit (practice-retention-transfer)

Four studies analysed the performance during practice, three of those did not find a difference in performance with an implicit or explicit learning method (Chiviacowsky et al., 2013; Tse, 2017; Tse & Masters, 2019) and one study showed a beneficial effect of implicit learning, in this case learning with an external focus of attention for children with ADHD (Saemi et al., 2013). On the retention test, four studies showed beneficial effects of implicit learning over explicit learning (Capio, Poolton, Sit, Eguia et al., 2013; Chiviacowsky et al., 2013; Saemi et al., 2013; Tse & Masters, 2019). One study found better performance following explicit learning for children with ASD (Tse, 2017) and two studies did not show a difference between the implicit and explicit groups (Van Abswoude et al., 2015; Van Cappellen – van Maldegem et al., 2018). On the transfer tasks, two studies showed better performance following implicit learning (Chiviacowsky et al., 2013; Tse & Masters, 2019) and one study showed better performance following explicit learning (Tse, 2017). The studies that applied an errorless learning protocol analysed the number of errors. Capio, Poolton, Sit, Eguia et al.\ (2013) showed that participants in the errorless learning group indeed made less errors compared to the error-strewn group, whereas the study of van Abswoude et al. (2015) did not show a difference in errors between the groups. In the study of Tse (2017), attentional focus was manipulated and the verbal reports showed that 90.7% of the participants adhered to the instructed focus.

Manipulation check

Two studies examined the role of working memory capacity on motor learning, with mixed results. In de study of van Abswoude et al. (2015) verbal working memory capacity was related to performance improvements across groups. In contrast, in the study of van Van Cappellen – van Maldegem et al. (2018) visuospatial working memory capacity was related to motor learning in the implicit group.

Methodological quality

The total methodological quality score of the studies ranged between 10 and 16. Only two of the studies included a control group (Tse, 2017; Tse & Masters, 2019) and only two of the studies included a manipulation check (Van Abswoude et al., 2015; Van Cappellen – van Maldegem et al., 2018). Only one study used a validated assessment of performance (Capio, Poolton, Sit, Eguia et al., 2013), and only two studies used blinded assessment and interrater reliability (Capio, Poolton, Sit, Eguia et al., 2013; Tse & Masters, 2019).

Additional results and practical implications

Next to working memory, the study of Capio, Poolton, Sit, Eguia et al. (2013) assessed throwing frequency during free play (i.e., the skill that was practiced), showing that throwing frequency improved most in the implicit learning group. Two of the studies performed the experiment in a practical setting indicting that these motor learning methods could be applied in practice for children with ID and DCD (Capio, Poolton, Sit, Eguia et al., 2013; Van Cappellen – van Maldegem et al., 2018).

Discussion

The aim of the present study was to provide an overview of the current state of evidence on the use of implicit motor learning in primary school children. The results showed that the implicit motor learning methods were just as effective, or in some cases even more effective, compared to the more traditional explicit motor learning methods. Both children with typical and a-typical development were shown to improve their performance on the practiced skill. However, the ecological validity of a majority of the studies is low and methodological issues (e.g., short interventions, small groups, no control on learning process and realized intervention) need to be taken into consideration when interpreting the results. Still, it may be cautiously concluded that implicit motor learning can boost on increasing evidence of its efficacy in children. Below, we will further elaborate on this conclusion.

Using implicit methods to improve motor performance

The studies in this review consistently show that implicit motor learning methods are just as effective, or even more effective, compared to explicit methods to improve motor performance of primary school children with typical or a-typical development. The majority of the studies showed this effect for aiming tasks (e.g., golf putting, throwing objects at a target), and predominantly assessed performance accuracy and not movement form. The overrepresentation of aiming tasks and product-oriented measures raises the question if children also improve the way they perform the movement and if the methods can be used for different types of complex motor skills. The few studies that do include a process-oriented measure also show similar, or even more, improvements in the implicit groups compared to the explicit groups, suggesting that movement form and accuracy may improve simultaneously (Capio, Poolton, Sit, Eguia et al., 2013; Capio, Poolton, Sit, Holmstrom et al., 2013; Petranek et al., 2019; Tse et al., 2017; Wulf et al., 2010). In addition, results from balancing tasks (Becker & Smith, 2013; Chatzopoulos et al., 2020; Flores et al., 2015) and jumping tasks (Chatzopoulos et al., 2020; Chow et al., 2014) do not differ from the aiming tasks, suggesting that methods can be applied to different types of complex motor tasks.

Another aspect that needs to be discussed is that only seven studies included a control group and results of the studies are inconsistent. Most results indicate that instructions (either analogy instructions or with an internal or external focus of attention) are just as effective as the control condition (Chow et al., 2014; Perreault & French, 2016; Teixeira da Silva et al., 2017; Tse, 2017, and the acquitition phase of; Hadler et al., 2014). In contrast, Tse and Masters (2019) showed that all forms of instruction were more effective compared to the control condition, both during practice and retention. Given the large variation in information content that is provided to the control group (and in some cases also to the experimental groups), from no information (Flores et al., 2015) to video instructions and general cues (Perreault & French, 2016), it is at this stage not warranted to conclude that implicit instructions (using an analogy or an external focus of attention) are more effective than practice with only general task instructions. Additionally, control groups have not yet been used in studies examining errorless learning in children. Therefore, it could be suggested that any form of task-specific practice leads to performance improvement. We suggest that a more structured approach in the employment and comparison of the different paradigms is warranted to strengthen the evidence regarding the effectiveness of the interventions.

A final aspect with respect to implicit and explicit learning, is the issue whether the intended intervention and learning process indeed occurred. Ten studies used verbal reports as a check if children adhered to the provided instructions when attentional focus was manipulated. Five of the studies reported that children indeed used their instructed focus, whereas the other five studies did not find conclusive evidence or even showed that children switched their focus, possibly leading to opposite learning processes. It needs to be noted that, especially among children, responses on the verbal reports have questionable reliability. For example, reporting of information is dependent on verbal and cognitive skills, which are still developing in children (Rieber, 1969). As a result, children may not be able to report all the knowledge that they use. Likewise, responses of children can include socially desirable answers, for example, by repeating the instructions of the experimenter without actually using them. On top of that, it was shown that children also report thought processes that are not clear movement-, or task-related rules or facts, but which can still load working memory (Krajenbrink et al., 2018; Van Abswoude, Nuijen et al., 2018). Taken together, the issues concerning the reliability of verbal reports question their value as a single measure to make informed conclusions about the realized intervention in children and cannot be related to the learning process taking place.

In addition to the check on the realized intervention, only nine of the studies reported in this review included some form of a manipulation check and results are inconclusive. Six studies reported results from a dual-task condition to check if motor learning was implicit. The three studies that examined errorless learning concluded that performance is more automated following an implicit motor learning paradigm, but at the same time fail to report on the performance on the secondary task. Interestingly, Krajenbrink et al. (2018) and van Abswoude et al. (2015) did include the change in performance on the secondary task and showed that in both the implicit and explicit groups performance on the motor task was stable (or even improved in the study of Krajenbrink et al. (2018), but this was at the cost of performance on the secondary task which in fact deteriorated. A recent review on movement automaticity following implicit learning in adults also highlighted that most studies overlook the change in secondary task performance when studying the dependence of motor performance on cognitive resources (Kal et al., 2018). In order to draw conclusions about the learning process based on dual-task performance, we recommend future studies to always include single- and dual-task performance of both the primary motor task and the secondary task for more accurate and reliable interpretations of the findings.

Individual differences

The individual abilities of the learner, type of task used, and the stage of motor learning are three factors that are crucial in choosing an appropriate implicit strategy and in turn affect the effectiveness of the chosen strategy and type of intervention (Kleynen et al., 2014, 2015). It is therefore surprising that a vast majority of the studies only analyse performance improvements on a group level and do not take individual differences into account. Nevertheless, a close look at the data in most studies show a large standard deviation in motor learning, indicating individual differences in the effectiveness of both implicit and explicit learning methods. Two studies did include the motor skill level in their analysis (Capio, Poolton, Sit, Holmstrom et al., 2013; Maxwell et al., 2017). Both studies indicate that implicit methods are only beneficial for children with a low level of motor skills. From the perspective of preventing delays in motor development and early intervention, this would suggest that implicit methods, or in this case errorless learning, may be preferred. However, the observed benefits may not be related to implicit learning per se, but can also be the results of the design of errorless learning studies. The early successes attained within this design may enhance self-efficacy or promote functional variability in movement execution, which in turn could result in the learning benefits (Van Ginneken et al., 2018). Additionally, in the studies assessing children with a specific motor disorder (i.e., DCD and CP), no differences between implicit and explicit learning methods are found, indicating that the beneficial effect of implicit methods may disappear when motor difficulties are more pronounced. It needs to be noticed that both cognitive and motor abilities were not comparable across studies in a-typical populations (if this information was presented), and could also vary within studies, again highlighting the importance of individual differences. In sum, the exact role of the level of motor skill on implicit and explicit motor learning remains unclear but warrants further study in order to provide individualized, and likely effective, training programmes.

A second factor that was included in four studies is the role of working memory capacity. Results of these studies are not in line with the expectation that especially verbal working memory is required in explicit learning but not in implicit learning. Two of the studies did not show any role for working memory in either of the learning methods (Brocken et al., 2016; Krajenbrink et al., 2018). Surprisingly, the study in children with DCD by Van Cappellen – van Maldegem et al., (2018) showed that visuospatial working memory capacity was related to learning in the implicit but not in the explicit group. More specifically, children who received feedback with an external focus of attention improved more on the aiming task when they had better visuospatial working memory capacity. Also, in the study of van Abswoude et al. (2015) verbal working memory was related to learning across the implicit and explicit group for children with CP. In addition to this inconsistent picture, on the one hand Buszard, Farrow et al., (2017) showed that children with better working memory capacity perform better during explicit practice and improve more compared to children with low working memory capacity. On the other hand, van Abswoude et al. (Van Abswoude et al., 2020; Van Abswoude, van der Kamp et al., 2018) did not find an effect of either verbal or visuospatial working memory on motor performance or learning following implicit or explicit learning. Similar results were shown by Jongbloed-Pereboom et al. (2018). In addition, Jongbloed-Pereboom et al. (2017) showed a role for visuospatial working memory capacity in both implicit and explicit groups. Taken together, working memory capacity may have a role in motor learning, but the assumption that it is a prerequisite in explicit learning and not for implicit learning cannot be confirmed. As such, it cannot explain the individual differences in the effectiveness of implicit and explicit motor learning methods (also see Janacsek & Nemeth (2013) for a critical review on the role of working memory in implicit and explicit learning). One confounding factor in this overview could be related to the relative load that the intervention poses on working memory capacity. For example, in the study of Brocken et al. (2016), both groups received one short instruction, with the instruction in the implicit group both phrased with an external focus and as an analogy. The load on working memory in this case is very low, compared to the five explicit instructions used in the study of Buszard, Farrow et al. (2017). This may partly explain the differences in the role of working memory capacity, but it does not fit with all of the findings. In addition, the tests used to assess working memory capacity are based on psychological constructs which may be less suitable to assess aspects of working memory capacity related to movement execution. We therefore suggest future research to take the cognitive load into account and to explore different ways to assess working memory capacity (also see Buszard, Masters et al. (2017)).

Ecological validity

Researchers mainly apply implicit motor learning interventions in controlled settings according to a protocol, often within a carefully pre-selected population. Practitioners, however, mainly apply motor learning in circumstances that are less controllable and more variable (Kleynen et al., 2015). A teacher or coach respond to what is observed during the class or training, without following a strict protocol as is the case in experimental studies. It is therefore important to assess if the results from experimental studies translate to a practical setting. A few of the studies described in this review include some elements that enhance the ecological validity. For example, in the study of (Van Cappellen – van Maldegem et al. (2018) the training, instructions and feedback were delivered by the child’s own physical therapist. Still, this therapist did follow a strict protocol. In the studies of Capio et al. (Capio, Poolton, Sit, Eguia et al., 2013; Capio, Poolton, Sit, Holmstrom et al., 2013) children practiced in pairs during the PE classes, but they were in a separate part of the gym and practiced an isolated skill. The studies that resemble a real-life setting the most are the studies of Chatzopulos et al. (2020) and Petranek et al. (2019). In the study of Chatzopulos et al. (2020), the whole class participated during practice and the skills were practiced with music and songs (which is usual for kindergarten) under supervision of their own teacher. In contrast, children in the study of (Petranek et al. 2019) received instructions and feedback from the experimenter, but the feedback was adapted to their actual performance and children practiced in small groups during PE classes.

One promising finding regarding the translation of practiced skills in a real-life setting is from the study of Capio, Poolton, Sit, Eguia et al.(2013). Here, children with an ID more often played games involving throwing during school breaks after the experiment. This increase in throwing frequency was more pronounced following implicit (errorless) learning. It suggests that practicing the skill may have led to enhanced enjoyment or self-efficacy with regard to throwing. These psychological effects are not yet systematically included in the studies, but are likely to be important for the motivation to pursue prolonged skill practice and motor learning.

Taken together, the first results of the studies that apply implicit and explicit motor learning in more practical settings are promising. However, up to date no research has been performed that aims to systematically evaluate how these methods can be applied in practice to yield the most success (also see van der Kamp et al. (2015)). One possible avenue we would suggest is to combine the field of implicit and explicit learning with the field of nonlinear pedagogical design principles, which accounts for the complex interactions between the learner, the task, and the environment (Chow et al., 2015). Skills can be practiced in a meaningful environment (for example, PE classes), while task constraints are manipulated based on the principles of implicit and explicit learning. We believe that the key lies within the profession that applies the learning methods in this meaningful, but also complex, context. Applying principles of implicit learning to constrain the task and the context in a way that children can perform the task successfully, with a minimum of explicit/prescriptive instructions, and suitable for the individual needs and abilities may yield the best learning result. In addition, effectiveness should not only be based on the product-oriented outcome (i.e., accuracy) but also on the learning process (both cognitive processes and movement behaviours). This way, we can better understand how we can promote motor learning in a meaningful setting.

Conclusion

In this systematic review on implicit and explicit motor learning of primary school children, we show that implicit methods are just as effective as explicit methods in improving FMS and sport skills. This holds both for children with typical and a-typical development, indicating that practitioners can use both methods. Nevertheless, several methodological issues need to be resolved to strengthen the evidence, understand the learning processes, and make informed recommendations for practice. To advance the field, structured manipulation checks, control groups, individualised approaches, and more ecologically valid designs are the next steps to be taken.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Hanzehogeschool Groningen, Award Number: HG11731 | Recipient: Wouter de Groot