Throughout the lifespan individuals are able to achieve new task performance goals through acquiring functional coordination patterns over time, with a process of refining acquired skills continuing at advanced levels of learning. Progress towards increasingly skilled performance consists of the acquisition and stabilization of more effective movement patterns (Vereijken et al., 1997). In this dynamic process of skill acquisition, adaptation and refinement, it has been proposed that movement variability may have different roles (Davids et al., 2003; Newell, 1985). For example, within-participant variability in movement coordination, over trials, has been defined as having a functional role, providing necessary fluctuations that allow individuals to refine and adapt acquired movement patterns (Davids et al., 2003; Newell, 1985).
Inter-trial variability has previously been examined by assessing its magnitude (Barris et al., 2014; Clark and Phillips, 1993; Hamill et al., 1999; Polk et al., 2008; Williams et al., 2015a; Wilson et al., 2008). Low values of variability here is considered a behavioural state that remains stable over time, while high variability has been characterised as system exploration during transitions to new or more refined movement patterns (Clark, 1995; Clark and Phillips, 1993; Hamill et al., 1999). It is proposed that an optimal range of variability is needed to learn and adapt motor skills (Stergiou et al., 2006). Values below this optimal amount of variability could make the system too rigid and values over the optimal variability would make the system too unstable. Within the optimal range of variability, early in learning, inter-trial variability may be high due to exploration of new coordination modes during practice. But in more skilled performers, variability can also need to be high to provide flexibility in adapting and refining movements to new performance contexts or challenges (Davids et al., 2003; Hamill et al., 1999; Wilson et al., 2008). Some initial suggestions have implied that magnitude of inter-trial variability conforms to a U-shaped function with skill progression (Wilson et al., 2008). A U-shaped function characterising movement variability might indicate that stable performance outcomes can be achieved in a number of ways in sport performers, varying in skill level, because different performance conditions may require different coordination modes during task performance (Edelman and Gally, 2001; Seifert et al., 2013).
The amount of variability in the performance and coordination dimensions can change in accordance with the skill level (Schollorn et al., 2009; Scholz et al., 2000). The Uncontrolled Manifold hypothesis proposes that the relationship between performance and coordination variability and the global performance of the task must be taken into account to interpret the functionality of the role of observed variability at different levels of motor expertise (Scholz and Schoner, 1999). Observed variability over trials can be associated with achievement of the key performance outcome in two ways: (a) low inter-trial variability would restrict variability in key performance measures ([V.sub.REST]), yet lead to task improvements; and (b), high inter-trial variability allows individuals to explore new coordination modes ([V.sub.EXPL]), resulting in simultaneous improvements in key performance measures.
To elucidate these different roles of movement variability as skill level changes, in this study we investigated performance in a multi-articular gymnastic skill as a task vehicle: the 'basic' longswing on the high bar (Irwin and Kerwin, 2005). During the longswing, gymnasts move from handstand to handstand position (at the top of the bar) by rotating around the high bar with a relatively straight body. Full extension of the arms and legs during the whole movement are expected to reach the criteria for the quality gymnastic movements defined in the Federation Internationale de Gymnastique (FIG) Code of Points (2015). A gymnast executes a backward swing as the body rotates to the rear with the front of the body leading throughout the movement. Individuals involved in a full gymnastics training programme successfully perform longswings after extended periods of practice, during which small longswing amplitudes of beginners increase progressively towards complete longswings from handstand to handstand in advanced performers. In addition, when gymnasts increase in competency, the longswing becomes a complementary skill to link other skills with higher difficulty levels, such as dismounts or flight elements, in a performance sequence (Arampatzis and Bruggemann, 1999; Hiley and Yeadon, 2003; Irwin and Kerwin, 2005; 2007).
Several previous studies have revealed the importance of hip and shoulder flexion and extension in successful execution of the longswing (e.g. Yeadon and Hiley, 2000). Irwin and Kerwin (2005) defined two functional mechanical phases during 'basic' longswing execution: (a) a rapid hyper-extension to flexion (i.e. closing angle) of the hip after the gymnast passes through the lowest part of the circle, and (b), extension to flexion (i.e. opening angle) of the shoulder joint just before reaching the highest point of the circle. Hip movements can be analyzed by observing coordination between trunk and leg segments, while the segmental arm-trunk coordination can provide insights on shoulder movements. Previous studies (e.g. Busquets et al., 2011; Williams et al., 2012; 2015a) have reported that novices, after a short period of practice (around two months), show more variability in functional phases of movement than experts. In addition, Williams et al. (2015a) found that novices who completed full longswings (360[degrees]) presented higher variability where the variability observed in expert longswings were low. They suggested that the high variability presented by the novice gymnasts allowed them to explore different motorperceptual strategies. That is, an increase in task experience changes the coordination and performance outcomes of the longswing (Busquets et al., 2013a), and likely their variability levels.
Understanding progression in skill level of a task can be achieved by studying performance in different age groups (Fleisig et al., 1999; Streepey and Angulo-Kinzler, 2002), since older participants typically accumulate more task experience than younger groups. Performance comparisons across different age groups are a good proxy for skill level since it affords cross-sectional observations of task experience effects. Although the true process of learning cannot be characterized, this study design allows researchers to gain insights into relevant changes in task performance, from beginner status to more advanced levels. In a previous study of experience effects on longswing performance, we investigated performance and coordination across competition age groups (Busquets et al., 2013a). This work demonstrated that the younger gymnasts displayed performance and coordination of earlier key events, while more expert gymnasts also revealed later key events in longswing performance. However, the roles of movement variability in establishing and refining functional coordination modes, as age-based experience increased were not evaluated in that study. With those findings in mind, the objective of this study was to examine the relationship between movement variability and performance outcomes as a function of expertise level.
Here, we assessed changes in inter-trial variability of coordination and performance outcomes in the longswing, from beginners to more advanced performers, differing in age. Based on current theorising in Ecological Dynamics (e.g. Seifert et al., 2013), we hypothesized that a U-shaped function would be observed in movement pattern variability across groups of beginner, intermediate and advanced gymnasts. As already outlined, we expected the magnitude of inter-trial variability of performance outcome measures and relevant coordination variables to be larger in younger age groups (beginners) than in intermediate level gymnasts, and to increase again in more experienced groups. Additionally, we hypothesized that relations between inter-trial variability and a key performance outcome variable, considering individual data points for each of the three younger groups, would determine which variables revealed restrictive ([V.sub.REST]) and exploratory constraints ([V.sub.EXPL]) during practice. We expected that beginners would display emergent coordination modes with more exploratory variability ([V.sub.EXPL]) than other groups, while intermediate level gymnasts would seek to stabilise emerging coordination patterns by increasing [V.sub.REST]. From a practical perspective, results of this research study could contribute towards a more accurate focusing of the use of practice variability across competition age groups to improve performance outcomes like swing amplitude or longswing proficiency.
The participants, age-groups and task performed in this study are the same as those presented in the previous study by Busquets et al. (2013b). Male gymnasts (n = 113), classified into five competition-age groups (G1, G2, G3, G4, and G5) participated in the study (Table 1). Gymnastic experience of participants (expressed in years) increased from younger to older individuals. All participants declared themselves to be fit and injury free, had two or more years of experience in gymnastics' training, including experience of the longswing on the high bar, and had competed at national level for their age group. All participants older than 18 years of age, and the parent or legal guardian of participants younger than 18, signed a written consent form to participate. In addition, verbal consent was also obtained from the young children. The Ethics Committee of Clinical Research of the Catalan Sport Administration...