An increasing popular view in developmental psychology that has been used to understand how developmental changes emerge is the Dynamic Systems Approach (DSA) (e.g., Thelen and Smith, 1994; Lewis, 2011; Molenaar et al., 2014; Blumberg et al., 2017). The hallmark of the DSA is its emphasis on all components of the system including environment and task. The DSA has provided lasting changes in understanding development in a variety of fields (e.g., Thelen et al., 2001; van Geert, 2011; Kahrs and Lockman, 2014; Roche et al., 2016; Bateson, 2017), in particular that of infant reaching (Clifton et al., 1993; Thelen et al., 1993, 1996; Corbetta and Snapp-Childs, 2009). In general, first reaching movements to a target emerge around 3 months (von Hofsten, 1991), while by 6 months, infants develop straighter and smoother reaches (Berthier and Keen, 2006). An important contribution of the DSA is that the intrinsic dynamics of the system affect this development (Thelen et al., 1996). For example, Thelen et al. (1996) showed that infants’ activity prior to reaching and infants’ arm movement speed influenced how infants learn to reach.

In contrast to infant reaching, the DSA is underrepresented in studies of mid-childhood (6- to 10-years of age) reaching. This field is exemplified here to show how the DSA can be used as an explanatory framework to advance understanding of development. During mid-childhood reaching skills are further refined (i.e., reaches become faster and more accurate), marking this as an important developmental period which has received limited attention. Our review of the literature revealed that developmental trends differ among studies, as shown in Figure 1 and explained in more detail later (e.g., Hay, 1979; Chicoine et al., 1992; Ferrel et al., 2001; Van Braeckel et al., 2007). Moreover, proposed explanations were tailored to trends revealed in individual studies, limiting their scope. The DSA is able to explain different developmental trends within one skill, as it focuses on all contributing components. Importantly, the components contributing to reaching are still undergoing developmental changes during mid-childhood. For example, joint coordination changes (Schneiberg et al., 2002), body proportions such as length and mass fluctuate (Malina et al., 2004), postural control accompanying reaching movements develops (van der Heide et al., 2003), and attention and executive functions improve (Olivier et al., 2003; Klimkeit et al., 2004). This shows the continuing complexity in reaching development and the need to understand developmental changes. The goal of this perspective is to offer novel ideas on how to advance the understanding of the development of mid-childhood reaching by approaching the developmental changes from the DSA. We commence with a short synopsis of the existing explanations for developmental changes in mid-childhood reaching.

We focus on studies in which simple reaching movements from a start location to a target were performed. These studies have primarily focused on the performance level of the reach, quantifying spatio-temporal measures of the index finger, such as movement time or accuracy at the target. Most studies compared three age groups, usually 6-, 8-, and 10-years-olds. Based on our literature review, we differentiate developmental trends into three groups, i.e., non-monotonic (not consistently decreasing; Figure 1, left panel), plateauing (decrease ending in plateau; Figure 1, middle panel), and linear (consistently decreasing; Figure 1, right panel).

The literature overview shows that several performance measures improve over development, indicating that important fine-tuning takes place during mid-childhood that demands understanding. Reviewed studies assumed that developmental changes in performance measures follow directly from developmental changes in one single process (i.e., feedback/feedforward mechanisms, sensory integration) or component (i.e., representations) in the system (e.g., Hay, 1979; Chicoine et al., 1992; Ferrel et al., 2001). Note, we will call these approaches ‘single-cause approaches.’ The proposed cause differed across studies which might be partly due to the time of publication, spreading across four decades. Early studies were published at a time in which the information processing approach was popular, whereas more recent studies follow the computational neuroscience tradition referring to internal models and representations as explanations. In sum, over the last decades, useful and interesting explanations of developmental changes in reaching during mid-childhood have been put forward by focusing on specific single causes, fitting within the theoretical framework underlying these studies.

We, however, propose that if one wants to understand the full range and complexity of the revealed developmental trends, one should depart from the assumption that over development the same cause is responsible for all developmental changes. Our reasons for this are: first, from the literature overview it became clear that different developmental trends were found for both, different manipulations and the same manipulations (e.g., vision availability). For example, the studies of Bard et al. (1990) and Fayt et al. (1993) both used similar age-groups, manipulated vision availability and focused on amplitude and directional aspects, but they found different results. If there would be a single cause, the changes brought about by this cause should be found across manipulations. Feedback and feedforward processes, for example, play a role in all described experiments in one way or another, which would mean that the deterioration in performance following from increased usage of feedback in 8-year-olds should be seen in each experiment. As the literature overview revealed, the deterioration around 8-years is not found in all studies which makes it unlikely that there is only one cause. Second, if we follow the reasoning of ‘single-cause approaches,’ measuring one level of the system would suffice (such as the performance level) because all other levels of the system (for instance, joint angles, muscle activation patterns, or brain activation patterns) should also reflect the developmental changes of the single process or component. Studies on reaching during mid-childhood have not focused on other levels so far, however, in other behavior different developmental trends at different levels have been found (e.g., Ricken et al., 2004; Wu et al., 2009). We expect that the same is true in reaching, which argues against the reasoning of ‘single-cause approaches.’ Third, a fundamental issue that previous studies have not addressed is why these processes or components should develop in the way the authors propose. We think that this is an essential question which will be difficult to answer, hampering full understanding of the developmental changes.

We propose that the DSA (Turvey and Fitzpatrick, 1993; Thelen and Smith, 1994; Lewis, 2011; Spencer et al., 2011; Newell and Liu, 2012; Molenaar et al., 2014; Endedijk et al., 2015) can offer an explanation for the full complexity of the development of reaching during mid-childhood. In contrast with ‘single-cause approaches,’ the DSA takes all components of the system into account. Importantly, the system is not confined to the body, but includes the full action-perception cycle. Automatically, this means that the environment and the task are equally important parts of the system. Thus, the DSA’s starting point is that all components of the person, environment and task are equally important and could potentially contribute to the emerging behavior (cf. Newell, 1986).

According to the DSA, the components of the body-environment-task system are interacting. The result of the interaction at any point in time is the system’s current behavior. Hence, if one or multiple components change, the behavior might change. Thus, developmental trends emerge from changes in interactions that are affected by all components of the system. In contrast with ‘single-cause approaches,’ the DSA does not search for causal factors in development, but aims to reveal processes according to which behavior emerges from various contributing components. It also means that the component(s) involved in the emergence of new behavior may differ at each instant in development.

The concept that DSA uses to explain the emergence of new behavior is that of an attractor. Attractors are preferred, but not fixed, behaviors of the system to which the system returns to when perturbed. Attractors emerge from the interaction of the components at a certain point in time. At a given moment more behavioral attractors are present, hence, the attractor landscape represents the dynamic regime and the stability of the attractors emerging from interactions among task, person and environment components. Changes in the attractor landscape (reflecting disappearing behaviors, appearing behaviors, and qualitatively changing behaviors) are indicated in terms of stability and its counterpart variability. Stability of the attractor specifies resistance to change which is indicated by the effort it takes the system to perform a new or a different behavior. Weak attractor stability can result in an easy transition to a different attractor, which is reflected in increased behavioral variability. For development this means that when components of the system change, the interaction changes, which might influence the stability of the attractors in the attractor landscape. This changed attractor landscape can lead to different behavioral patterns becoming stable resulting in changes at the performance level, affecting development.

To understand developmental changes in reaching, the attractor landscape of reaching has to be identified (e.g., Schöner, 1990, 1994; Huys et al., 2014; Knips et al., 2017). Following Schöner (1990, 1994), we suggest that discrete reaching movements can be conceptualized as sequentially stabilizing point attractors (i.e., representing the initial and target location) and limit-cycle attractors (i.e., representing the movement) within a single dynamic system. The reaching movement is engendered by an intentionally destabilizing point attractor of the initial location while the limit cycle concurrently stabilizes (representing the actual displacement of the limb), followed by a destabilization of this limit cycle and subsequent relaxation toward the target location attractor. How do changes in the attractors of reaching lead to different patterns of change at the performance level?

The limit cycle in particular has effects on the performance of the reach because it accounts for the trajectory stability (Beek and Beek, 1988; Mottet and Bootsma, 1999; Zaal et al., 1999; Bongers et al., 2009). For instance, Mottet and Bootsma (1999) showed that to meet different task constraints imposed by modifications in target size and distance in a rhythmic reaching task, limit cycle dynamics systematically varied over conditions. In each condition, the imposed task constraints instantiated a limit cycle attractor of which the characteristics emerged from the interaction of the components involved in the system (target properties and person constraints). The exact dynamics of the limit cycle in turn determined the performance of the reaching movement.

Here, we have set out to provide a perspective that can offer an overarching explanation for a research field; mid-childhood reaching. The advantage of the DSA is that this perspective’s line of reasoning can be applied to other developmental fields because it is about general principles of change. We have already shown how these principles can also be applied to the behavior itself (i.e., the attractor landscape of reaching). These general principles provide a framework that over-arches individual studies and fields of studies. Other fields of development (e.g., language development) can therefore benefit from insights in, for example, the field of reaching. One important point that should be considered in all fields is that the effects of the context in which behavior is performed and the influence of individual characteristics should be determined. Here, we have argued that differences in experimental setups together with differences in developing components involved in reaching may explain different developmental trends. Therefore, in all fields of developmental research inter- and intra-individual variability should be more emphasized (cf. Adolph et al., 2015). Focusing on individual differences implies an experience driven-approach, as opposed to the often-used age-driven approach (e.g., which is also used here to follow the literature). Changes in intra-individual variability could indicate transitions to new behavior (i.e., changes in the attractor landscape). Also, variability may imply exploration which is valuable to understand how new behavior emerges. To conclude, with the example of mid-childhood reaching we have shown that the DSA can increase the understanding of emerging developmental changes.

Following authors contributed to the following aspects of the paper: (1) establishing the goal and main research question (LG, RB, MS, LM, EO); (2) reviewing the literature (LG, MS, RB); (3) drafting the work (LG, RB, MS); (4) establishing the core conceptional aspects (LG, RB); (5) critical revising of the work (LG, RB, MS, LM, EO). All authors gave final approval of the version to be published.

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.