There is growing interest in the prospect of improving rehabilitation by applying the principles underlying the control of motor actions to therapeutic interventions. Here, the word “control” implies the ability to direct, command, or rule the production of motor action. However, there is some misunderstanding among rehabilitation clinicians and neuroscience researchers about motor control theoretical frameworks and terminology. Clinicians prescribe and apply treatment interventions based on hypotheses about the effectiveness of interventions. Assessment of effectiveness follows from the process of clinical reasoning based on examination of the best available evidence, clinical expertise, and patient values, which are the pillars of evidence-based practice. The conceptual framework guiding this process is the International Classification of Functioning Health-related Function and Disability Model of the World Health Organization (WHO, 2001). The International Classification of Functioning model also provides a lexicon to communicate aspects of a patient’s sensorimotor impairment, activity limitations, and participation restrictions. Since its introduction in 2001, this model has helped foster accurate and informative interdisciplinary communication. What is absent from the conceptual framework for clinical practice is a comprehensive theory of how motor actions are controlled. However, there is currently no consensus among neuroscientists and clinicians about a common motor control theory. Consensus would be valuable in order to identify what we mean by normal and abnormal movement and how neuroplasticity and functional recovery occur. Thus, the understanding of how motor actions are controlled is an essential contributor to the conceptual framework for clinical practice, on which evidence for clinical effectiveness should be based.
The situation engendered by the International Classification of Functioning lexicon is not shared by the field of motor control. Indeed, there is considerable variability regarding the use of the term “motor control.” For example, some commonly occurring phrases in the literature are: “altered motor control of the pelvis,” “scapular motor control retraining,” or “motor control of the spine” (e.g., Aldabe et al., 2012; Hodges & Richardson, 1996; Worsley et al., 2013). Reading this, one might assume that the process of control of the pelvis, shoulder, or spine are not the same and that remediation of sensorimotor deficits of diverse body segments should be approached differently. However, this type of incorrect and confusing use of the term “motor control” often occurs in the literature (Low, 2018). The confusion mainly stems from the use of the term “motor control” as a synonym of “motor commands to muscles,” which is based on the widely held assumption that the central nervous system (CNS) specifies muscle activation (i.e., EMG) or force directly. Indeed, there is growing interest in correlating neural activity to motor behavior in order to identify the strategies used by the CNS to produce movement (e.g., Churchland et al., 2012; Suresh et al., 2020). These efforts, including computational approaches using neural networks and modeling (Todorov, 2000), have provided important information primarily about how movement is executed. This area of research suggests that a wide neural network is involved in the planning and execution of voluntary movement (Groenewegen, 2003; Sobinov & Bensmaia, 2021; Svoboda & Li, 2018). Such models are helpful in understanding how movement execution is disrupted following stroke and other CNS lesions. However, the interpretation of findings is largely based on the underlying assumption about direct control of muscle output, which has been questioned on the basis that it rearranges cause and effect in the production of motor actions (see below).
The purpose of this paper is to discuss the underlying assumptions about the control of voluntary movement and to provide suggestions for a unifying framework for the understanding of the control of motor actions.
Assumptions Underlying the Use of the Terminology
Since one of the aims of science is to propose a coherent view of the world, related terminology needs to show internal consistency and a logical structure. As a consequence of scientific advances, definitions or terminology may change according to new discoveries. For example, the theory of special relativity has changed our fundamental understanding of space and time (Einstein, 1916). Therefore, the precision of terminology should follow the epistemological evolution of philosophy and physics.
As one of the most important goals of rehabilitation is the recovery (management) of sensorimotor function at the impairment and activity level, much attention has been given to the recovery of “motor control” processes. Notably, the term “motor control” is often used in the literature with different meanings ranging from descriptions of motor behavior at the muscle or joint level to descriptions of pathways involved in higher order processing of sensory or other information (e.g., Leiras et al., 2022; Zarzycki et al., 2022). For example, several studies report altered kinematics and muscle activity in patients with low back pain or ankle instability compared with healthy subjects (i.e., Labanca et al., 2021; Pinto et al., 2021), with the suggestion that such differences are the result of “poor motor control.” Similarly, following rehabilitation interventions better execution of elbow movement in patients with stroke has been attributed to better motor control (Hammerbeck et al., 2017). However, in both instances, better motor output may largely be due to improvements in the plasticity-related mechanisms involved in the transmission of signals to muscles resulting in improved motoneuronal recruitment and rate coding as well as muscle fiber hypertrophy (Burke, 1981) rather than to changes in motor control processes. While such studies elegantly describe improvements in motor output (i.e., motor behavior), without a clear theoretical foundation about what exactly the CNS controls, they do not provide information about what has changed in the CNS to explain the improvement. An important distinction is that the analytic description of the final output, for example, reaching, walking, and so on, represents motor behavior and not the motor control processes underlying this behavior. In contrast, research based on observations of motor behavior can use inductive reasoning to formulate hypotheses or theories of motor control.
Confusion About Terminology
Terminological inconsistencies exist when referring to hypotheses, theories, and models of motor control. In scientific reasoning, a hypothesis is an assumption made before an experiment has been done to test it, while a theory is a principle used to explain phenomena supported by data. A model is a physical and/or abstract representation of a specific aspect of a system linking a theoretical concept with experimental output and then used to formulate and test predictions of the theory. However, these three terms have often been used interchangeably, resulting in confusion and miscommunication. For example, Sherrington (1947) proposed that movements were produced by the modulation of reflex parameters. In his landmark work, The Integrative Action of the Nervous System, he described the neurophysiological mechanism of reflex behavior, introducing the concept of the synapse. This work should be recognized for its advancement of the understanding of a neural mechanism, but has been referred to as a theory of motor control (i.e., Reflex Theory; Shumway-Cook & Woollacott, 2012). Similarly, the hierarchical organization of the CNS (Foerster, 1977) that elegantly proposed a relationship between anatomical structures and motor actions has also been referred as a theory, while it should only be considered as a model to investigate specific neural structures and lesion effects on movement. The above examples illustrate how models have been misused to describe how the nervous system controls movements while they actually describe different processes by which control leads to the execution of movement, without addressing what is actually controlled.
Lack of a Unifying Theoretical Framework
Awareness is growing about how knowledge of normal control processes may inform the understanding of disordered movement production to effect better functional outcomes (Kitago & Krakauer, 2013), where motor function refers to the ability to perform a task that has a specified purpose. However, confusion has arisen because different models/theories have been used to describe motor training (e.g., McLoughlin, 2020; Shumway-Cook & Woollacott, 2012). The problem of how to apply multiple models/theories of motor control to rehabilitative training is an artificial one, borne from a hodgepodge of different models, hypotheses, principles, and theories alternatively describing features of the control and execution of movement. This situation in which the boundaries between control and execution have been blurred has been propagated from one textbook to the next and has led to an unwieldy set of ideas, without a common unifying framework. This has led to a mechanistic rather than a theory-driven “mash-up” approach whereby one theory may be used to explain one sensorimotor deficit and another theory may be used to describe another deficit, sometimes in the same patient. Such an approach is controversial as it does not recognize the intrinsic complexity of neural processes and interactions with the environment and is a barrier to advancing treatment effectiveness.
In 2017, the Stroke Recovery and Rehabilitation Roundtable task force clarified many concepts and practices in poststroke rehabilitation but a common language for the control of motor actions was not considered (Bernhardt et al., 2017). Therefore, it is time to provide a common definition of motor control and describe a potential theoretical framework to foster better communication and allow interdisciplinary collaboration among clinicians and researchers, and across fields of science.
The Solution
A general solution to the problem of the use of the terminology of motor control in rehabilitation starts with a clear definition of the concept. Control is not simply the production of muscle forces, but rather the processes underlying the production of forces. It is important to consider that force output occurs under different environmental and sensory conditions, based on the context in which the action is performed, and subject to biomechanical, environmental, and contextual constraints (Bernstein, 1967; Latash & Zatsiorsky, 2016; Newell, 1996). The International Society of Motor Control has adopted the following definition of motor control as “an area of physics exploring laws of nature defining how the nervous system interacts with other body parts and the environment to produce purposeful, coordinated actions” (i-s-m-c.org).
This definition does not specify exactly how control is organized in the CNS. However, traditionally, it is thought that in the hierarchically organized CNS, control of sensorimotor processes occurs at the highest levels. Following the ideation of an action, a decision-making process is initiated, likely through a distributed network of connections between the premotor cortex, supplementary motor area, primary motor cortex, association cortex, intraparietal sulcus, dorsolateral prefrontal cortex, and visual areas (Roland et al., 1980; Weinrich & Wise, 1982). Both feedforward and feedback modes of control can be combined where sensory feedback loops connect lower to higher levels to shape descending influences and provide relevant information about the action and the environment. The process results in the execution of a motor action through connections at lower levels (Wolpert & Ghahramani, 2000; Wolpert & Jessell, 2021). Thus, motor output such as force production only represents the consequences of control processes and not the processes themselves. Here, we distinguish between control processes that occur at higher levels of the neuroaxis and motor execution at lower levels.
Is there a unified theory of motor control? Currently, there are two major and opposing theoretical approaches to motor control—the direct (also known as computational, internal model, or biomechanical) and the indirect (also known as physical) frameworks. The direct framework assumes that the CNS outputs muscle force (e.g., direct programming of EMG ouput, reviewed in Kawato, 1999) through the direct activation of muscles from higher brain centers. In contrast, the indirect framework assumes that force production results from the specification of neurophysiological parameters that may influence, but remain independent of, biomechanical variables (Feldman, 2019). Thus, neurophysiological parameters determine the conditions and context in which muscles may act to produce a given task within a given environment. Each framework encompasses hypotheses/theories with similar approaches but different names. Grouped within the direct framework are the computational approach (Wolpert & Ghahramani, 2000), internal model representation in the brain (Gomi & Kawato, 1996), and direct programming of EMG patterns or forces (Todorov & Jordan, 2002). Grouped within the indirect framework are concepts of the structure of coordination and control (Bernstein, 1967; Latash, 2020; Scholz & Schöner, 1999), the equilibrium-point hypothesis extended to the theory of spatial threshold control and referent control (Feldman, 1966, 2019), dynamical action systems (Haken et al., 1985; Kelso, 1984), and ecological approaches (Gibson, 1966).
The direct framework is based on the idea that the sensorimotor system needs to produce a certain mechanical output and does so through direct programming of movement output (i.e., velocity, trajectories, EMG patterns, and forces). It was elaborated after Hollerbach (1982), inspired by computational (cybernetic) control schemes for robotic motion, suggested that the brain uses laws of mechanics and neural emulators of neuromuscular properties (i.e., hypothetical internal models) to precompute motor commands to muscles (EMG patterns) that elicit the desired behaviors (e.g., movement trajectory). In this approach, the brain should compute all the necessary forces with the requisite time profiles for each muscle to produce each specific movement. The assumption is that neurons can perform computations and then convert these computations into physical variables by unknown actuators or brain pathways. In order to do so, the brain should have a representation of all possible combinations of muscle activation patterns. This assumes that the brain has immense computational power, huge storage capacity, and can instantaneously produce a unique force output for each motor action from the multitude of possible outputs. Motor planning and control rely on feedforward or predictive mechanisms of the desired action (Wolpert & Jessell, 2021). How and where the brain “computes” these processes is unclear, but some computational processes have been identified in cortical pyramidal cells (Herz, 2006), and some actions have been modeled using neural networks (Mathis & Mathis, 2020). Actions have also been modeled using biomechanically based inverse dynamic computations and/or paired inverse and forward models (Wolpert & Ghahramani, 2000). Multiple internal models have to be produced that mimic not only motor but also sensory and cognitive processes and allow the system to rapidly adapt to the changing dynamical limb properties as it interacts with the environment. Such descriptions are specific to the movement studied, such as reaching into different arm workspace areas with and without the application of external force fields (e.g., Todorov & Jordan, 2002).
Although this notion is attractive, on closer examination, it has several limitations. Most importantly, the fact that neurons have threshold properties is overlooked. Notably, computational theories are physiologically incompatible with the nonlinear properties of motoneurons (MNs) since motoneuronal input/output functions are irreversible (Feldman, 2019). This means that the CNS cannot deliver input signals that can elicit the required output in terms of EMG patterns or muscle forces, rendering even the production of a simple, isometric force inexplicable. Moreover, physical actions of living systems are inevitably consistent with the laws of mechanics, and the idea of using internal models of mechanical laws for computation of motor outcomes is misleading.
Another significant drawback of the direct approach is the inability to solve the “posture-movement” problem to explain how movement can be produced from one position to another without evoking posture-stabilizing mechanisms (Latash & Zatsiorsky, 2016; reviewed in Feldman, 2019). The commonly accepted view of von Holst (1954) that reflexes are suppressed by efference copy when movements are produced in order to prevent reflex resistance to deflections from an initially stabilized posture (Cluff & Scott, 2016) has been criticized (Latash & Zatsiorsky, 2016, p. 140; Feldman, 2019). One major criticism is that by suppressing reflexes, the ability of the system to resist external perturbations during intentional movement would be lost, which is only observed in the pathological case of deafferentation (Blouin et al., 1993).
The direct approach also does not solve the “motor redundancy” problem (Bernstein, 1967), which describes the situation in the motor system in which the production of any particular movement requires the CNS to control a large number of elements (i.e., muscles and joints) at multiple levels. Take for example, the simple task of reaching the hand from one three-dimensional point to another in a particular time frame (Figure 1). This task involves the coordination of movement in at least three joints (i.e., shoulder, elbow, wrist) that comprise 7 degrees of freedom (DF)—three at the shoulder (abduction/adduction, flexion/extension, and rotation), one at the elbow (flexion/extension), and three at the wrist (flexion/extension, medial/lateral deviation, and pronation/supination), not to mention those at the scapula and the hand. From a computational perspective, the system would have to find a seven-dimensional vector describing the arm position at the final position that is a three-dimensional space, as well as the time functions for each transformation. This problem is not trivial and has no general solution. The direct programming approach attempts to simplify the problem by finding a solution that optimizes a cost function such as one that produces the smoothest movement (Flash & Hogan, 1985), uses a minimal amount of energy (Hasan, 1986), or depends on optimal feedback control (Todorov & Jordan, 2002). Note that the notion of “optimal movement” can refer to different motor outcomes depending on the task goal. For example, a movement that uses less energy may be considered more optimal from the standpoint of minimizing fatigue. A drawback however is that when trying to optimize speed, greater energy would need to be expended. Movements can also be optimized for endpoint smoothness that is considered to reflect better coordination or control (i.e., Saes et al., 2021). Thus, movement optimization is related to the task goal and the CNS needs to be flexible enough to quickly adapt motor output according to task needs. The question arises as to whether the CNS distinguishes between different optimality goals and how it manages to prioritize and “control” the multitude of elements involved in any given task. Some direct approaches acknowledge redundancy and provide a mechanism to reduce redundancy by neural organization of kinematic and kinetic variable into groups. They recognize that the problem of redundancy cannot be solved by a direct programming approach in which the plan of the CNS is to resolve the redundancy inherent in the musculoskeletal system by replacing the behavioral goal (achievable via infinitely many movement trajectories) with a specific desired trajectory. The problem is solved by the construction of synergies that serve to stabilize the task goal while permitting a certain variability in the individual DFs. Such schemes have been elaborated in various ways in the direct approach as synergetic coupling schemes (i.e., Santello & Soechting, 2000) and optimal feedback control (i.e., Todorov & Jordan, 2002).
—Schematic diagram of a subject making a reaching movement from position (a, left) to position (b, right) with the seven degree of freedom arm.
Citation: Motor Control 26, 4; 10.1123/mc.2022-0026
In contrast, the indirect framework assumes that the CNS does not need to seek an optimal solution to restrict the system’s redundancy. Instead, the CNS takes advantage of the system’s redundancy and produces multiple acceptable solutions (i.e., principle of abundance; Latash, 2012). Families of acceptable actions from this redundant set emerge as a consequence of the interaction of the body with the environment, rather than being directly programmed by the CNS (Feldman, 2015). The families of solutions may show optimal features, for example, Analytical Inverse Optimization cost function (Park et al., 2010) or end-state comfort effect hypothesis (Solnik et al., 2013).
Neurophysiological studies suggest that the CNS organizes elements (e.g., muscle activity, joint angles, finger forces, etc.) into task-specific ensembles capable of performing the desired action. Rather than finding a single solution, the CNS produces families of solutions that are equally able to accomplish the task. For example, in his classic experiment in blacksmiths, Bernstein (1967) observed that during repetitive hammer strikes, the hammer orientation with respect to the target was preserved (i.e., stabilized) despite a high spatial variability of the hammer trajectory during the swing. As long as the stability of the task goal was maintained, the actual hammer trajectory could vary while the general structure of the movement was produced according to a learned representation of the movement called a movement topology. Bernstein’s concept of a movement topology refers to a pattern of a neural variable stored in memory, related to the properties describing the structure (based on task parameters) of a learned movement. Note that Bernstein never postulated that topologies lead directly to the production of EMG outputs, forces, or desired trajectories. Côté et al. (2008) repeated the hammer experiment and showed that despite a change in both hammer trajectory and shoulder–elbow joint rotations in the presence of fatigue, the system continued to preserve the movement topology and the hammer orientation when striking the object (Figure 2). In other words, the system used the abundant possibilities of combining different DFs to stabilize the task goal, which was to hit the object with the hammer. Many other examples of trajectory and interjoint coordination variability suggest that the system is primarily concerned with stabilizing endpoint performance instead of directly specifying the details of force output and multiple joint rotations, which are defined by external factors such as biomechanical and task constraints (Newell, 1996; Scholz et al., 2000; Tomita et al., 2017). In the indirect approach, solutions to the redundancy issue have been proposed on the basis of the principle of motor equivalence (Bernstein, 1967; Lashley, 1951; Wing, 2000) in which the CNS finds motor solutions consistent with the natural constraints imposed by the body’s biomechanics to reduce the task space, while maintaining movement stability (i.e., uncontrolled manifold (UCM) approach, Scholz & Schöner, 1999).
—Trajectories of individual hammer strikes produced by elbow and shoulder rotations of the upper limb in the unfatigued state (gray traces) and the same action after fatigue of the shoulder extensors (black traces). Note that regardless of the fatigue state, the movement is organized so as to preserve the orientation of the hammer at the end of the strike (black rectangle). Reproduced with permission from Côté et al. (2008).
Citation: Motor Control 26, 4; 10.1123/mc.2022-0026
Only one indirect approach has proposed a physiological mechanism by which control is exerted at the effector level. In the referent control theory (originally called the equilibrium-point hypothesis), the CNS sets parameters that define the area of joint space in which a goal-directed movement can emerge. Control signals modify motoneuronal thresholds within the parameter space to produce movement. What sets this apart from other control schemes is that it describes how electrical neuronal activity (i.e., changes in the membrane potential [ΔV] of MNs) is converted to muscle length (i.e., spatial/position-dimensional quantity) variables based on setting the parameter, λ (Feldman, 1966). Muscle activation emerges as a function of changes in motoneuronal ΔV via descending and reflex inputs (Figure 3a and 3b; Feldman & Levin, 1995). The threshold of muscle activation is controlled by descending inputs directly and indirectly influencing MNs or other elements of the reflex loop (Matthews, 1959) as well as cutaneous afferents and reflex intermuscular interactions (Feldman & Orlovsky, 1972; Nichols & Steeves, 1986). Segmental mechanisms and descending influences combine to regulate motoneuronal thresholds in a multi-muscle system according to specific task demands (e.g., to specify different postures; Feldman & Levin, 1995; McClelland et al., 2001).
—(a) Muscle stretch (diagonal line with thick arrow) increases the ΔV of the motoneuron. When ΔV reaches the Vth, the motoneuron is recruited at a specific angle (open circle), called the TSRT1. Descending input increases ΔV (vertical arrow) such that the same stretch (upper diagonal dashed line with dotted arrow) results in a shift of TSRT to a smaller joint angle (black circle, TSRT2). The smaller joint angle is equivalent to a shorter muscle length. (b) The normal range of TSRT control (TSRT−, TSRT+) extends beyond the lower and upper limits of the biomechanical range (x−, x+). (c) In neurological pathology, the range of TSRT may be reduced. An example is shown of limitation in the upper limit of TSRT (TSRT+). In the joint range up to TSRT+, normal muscle activation is possible (black horizontal bar: active control zone). In the area beyond TSRT+, in which muscle lengths are longer, spasticity is present (gray horizontal bar: spasticity zone). ΔV = changes in membrane potential; Vth = threshold value; TSRT = tonic stretch reflex thresholds.
Citation: Motor Control 26, 4; 10.1123/mc.2022-0026
An example is shown in Figure 3 for a single muscle. In this approach, voluntary movements are produced by shifting spatial motoneuronal thresholds, here referred to as tonic stretch reflex threshold angles (TSRTs, λs). This is similar to the idea of an angular “set-point” for muscle activation, from which MNs are sequentially activated according to the size principle (Henneman et al., 1965). Extending the TSRT beyond the upper limit of the biomechanical joint range produces full muscle relaxation (TSRT+; Figure 3b). To produce physiologically possible combinations of voluntary muscle activity, torque, and position, the TSRT is shifted within the biomechanical joint range (to the left of TSRT+), defined by the physical joint limits.
Descending signals responsible for the setting of TSRTs are modifiable according to contextual (task) and environmental constraints and the associated sensory feedback. In this view, the CNS does not specify EMG output or force directly since physical, task, and environmental constraints make it impossible to predict the exact forces that will be needed to perform the task. For example, to initiate walking, the system shifts the referent configuration of the body (RC) from an initial (Figure 4a) to a final position (Figure 4b). After the shift, the actual body configuration is deflected from the new RC, and movement is produced to diminish the difference between the actual body configuration and the RC (see Feldman et al., 2021). In this way, balance and stability are transferred to a new position by the shift in the referent posture, thus solving the classical posture-movement problem. From the dynamical system’s perspective, the new RC becomes an attractor point that the system tries to reach without any loss of stability. This is in contrast to the popular direct approach view that taking a step results in the center of mass leaving the base of support, such that the body begins to fall and then needs to “catch” itself by taking a step (MacKinnon & Winter, 1993). The direct approach assumes that the COM is internally represented somewhere in the CNS and that the CNS can compute its motion based on laws of mechanics to determine when catching the body should be initiated to prevent falling. Such a falling-catching sequence is thought to occur in each gait cycle (Day & Bancroft, 2018; MacKinnon & Winter, 1993). However, this cumbersome computational task is avoided in the referent control theory by considering that the shift in posture is based on resetting threshold parameters and the tendency at any level (e.g., neurons and MNs) to minimize the difference between the actual body configuration and the RC. Therefore the system can perform the task using many different combinations of joint rotations—taking advantage of the system’s abundance, as has been demonstrated for head rotations in nonhuman primates (Lepelley et al., 2006), standing and leaning (Zhang et al., 2018), jumping (reviewed in Chan-Viquez et al., 2020; Feldman, 2015), and walking (Feldman et al., 2021) in humans.
—Schematic depiction of step initiation according to the referent control theory. (a) Prior to step initiation, the RC is in line with the AC. The center of pressure of the AC lies within the BOS in the dynamic stability area of both legs. (b) To initiate a step, the CNS projects the RC forward, so that the initial AC is now deflected from the RC. The shift in RC compels the resetting of the motoneuronal thresholds and the shifting of the body posture to the new position. This resetting results in movement of the body to the new position since the system acts to diminish the difference between the current AC and the specified RC. The intensity of motoneuronal activation (i.e., EMG output/force) depends on the difference between these two postures. RC = referent body configuration; AC = actual body configuration; BOS = base of support; CNS = central nervous system; EMG = electromyography.
Citation: Motor Control 26, 4; 10.1123/mc.2022-0026
Recommendations
We propose that sensorimotor rehabilitation should be predicated on one comprehensive theory of motor control and not on an eclectic group of noninterrelated notions of motor control processes as is popular in the rehabilitation literature (McLoughlin, 2020; Shumway-Cook & Woollacott, 2012). By adopting a unified theory, research would be better focused on producing evidence of the effectiveness of therapy that would fit into a logical structure of the understanding of motor control and disordered sensorimotor processes. Such an approach may lead to advancement in rehabilitation effectiveness. While the current predominant approach to therapy is predicated on the assumption that the CNS directly specifies motor output (i.e., force), we have described several compelling arguments why this may not be feasible. In this section, we suggest several pragmatic recommendations about the structure of therapy with the goal of improving the control of movement based on the idea of indirect, parametric control. The recommendations refer to common features of rehabilitation based on the principles of motor learning and adaptability (cf. Kleim & Jones, 2008).
Motor Learning
Motor learning is inextricably linked to the theory of the control of motor actions. Van Dijk et al. (2017) described two general approaches to motor learning based on the relationship between body functions and activity—a reductive (i.e., computational, internal model) approach in which the movement is broken down to its parts which are then practiced, and an emergent approach in which the task is practiced as a whole.
In the direct approach (i.e., Wolpert & Ghahramani, 2000) control is focused on the body functions themselves, and not on the task. The idea is that once the component parts of the task are learned, they are compiled into a representation of the movement that is stored in the brain. An example is the training of a reaching movement using a robotic device in individuals with stroke (reviewed in Marchal-Crespo & Reinkensmeyer, 2009). This approach is based on the assumption that the goal of the reaching movement is less important than its component parts, while the emphasis is placed on the underlying joint rotations needed to move the hand to the object and the problems encountered at the joint level, such as deficits in extending the elbow or aperture formation. However, since reaching and grasping are tightly spatially and temporally coupled (Jeannerod, 1984), training components separately may not result in improving the movement dynamics and task performance as a whole. Another assumption is that this same movement pattern can be applied across multiple reaching tasks, which has not, as yet, been supported.
Training based on the notion that actions are not directly programmed, but emergent, assumes that control is exerted at the task level. This concept has been widely implemented in the task-oriented approach to motor learning (Carr & Shepherd, 2010) and is similar to concepts described in the referent control theory (Feldman, 2015). From a referent control perspective, to produce a reaching movement, the CNS shifts the threshold of the referent body configuration from an initial to a final position and the movement emerges from the dynamics of the interactions between the joint rotations, the location and orientation of the object to be grasped and the environment (Feldman, 2015; Newell, 1985). This approach supports both part and whole training. In part training, the task is split into its component parts to focus on task subgoals and then on retraining the perception-action coupling essential to whole-task performance (Gibson, 1966; Reed, 1988). The focus is on adaptive tuning of the activation of multiple muscles and joints together during variations in task performance (Levin & Demers, 2020; van Dijk & Bongers, 2014). In this approach, the parts of an activity are functional units at the activity level, rather than individual joint movements, and have also been described as “task-specific synergies” (Latash et al., 2007; Tomita et al., 2017).
Adaptability and Redundancy
A movement may be considered impaired if it lacks speed or smoothness. However, from a control perspective, it may be insufficient to set up an exercise program based on the idea that faster and smoother movements are “better.” Indeed, studies have shown that reaching movements in poststroke individuals can be either faster, smoother, or more accurate when compensatory trunk and/or shoulder movements not normally recruited for the task are used (Hammerbeck et al., 2017; Levin et al., 2015; Michaelsen et al., 2001). Increasing movement speed and smoothness of the endpoint trajectory often occurs at the expense of movement stability, evidenced by a higher degree of variability either in the trajectory or in its component joint rotations (Hammerbeck et al., 2017). Rather than prescribing specific joint rotations, the referent control theory stresses that a reaching movement is produced by shifting the upper limb RC threshold from an initial to a final position. Once shifted, the actual arm position is seen as a deviation from the desired final arm position, and EMG is produced as a function of the difference between these two positions to correct the force imbalance (Feldman, 2019). Essential to this process is the need for the CNS to ensure movement stability. Scholz et al. (2000) measured the stability of a multi-joint movement produced by a redundant kinematic system, by calculating the variability in all upper limb joint rotations during a mock pistol-shooting task. The calculation, using the UCM approach, accounted for how much one joint moved with respect to its most proximal adjacent joint from the beginning to the end of the movement. A dynamic profile of multi-joint stability throughout the performance of the whole task was expressed as the ratio between how much movement in each joint contributed to stabilizing the endpoint (i.e., position or orientation of the pistol at a certain position “good” variability) compared with how much joint movement contributed to moving the endpoint away from that postion (i.e., “bad” variability). A dynamically stable movement is one in which there is a manifold of solutions described by a high ratio of good to bad variability. Scholz et al.’s (2000) results support the assertion that there is no single, “optimal,” way of performing a movement but rather there are many different combinations of joint rotations that lead to a movement solution. The UCM approach has been used to describe kinematic redundancy in individuals with stroke. The UCM provides a quantitative measure (synergy index, ΔV) about how flexible solutions are used to stabilize a particular performance variable (endpoint trajectory). ΔV is computed as the difference of the variance of the elemental variables (e.g., joint rotations) that does not affect the performance and the variance orthogonal to the UCM (that affects the performance) divided by the amount of the total variance (all variance is corrected per degree of freedom). Reisman and Scholz (2003) found that individuals with mild to moderate stroke could maintain a similar synergy index during a reaching task but that they used different patterns of joint coupling to produce stable trajectories. Investigating the coordination of trunk muscle activity during upward and downward seated reaches, Gera et al. (2016) found that individuals with stroke had deficits in flexibly combining trunk muscle activity associated with a higher level of bad variability (variance orthogonal), especially for reaching upward. Several studies have shown that kinematic redundancy is diminished in neurological populations, leading to a reduction in the number of possible combinations of muscle and/or joint rotations needed to perform a task (i.e., Shaikh et al., 2014; Tomita et al., 2017). This may impact the ability of such individuals to adapt movement to different environmental conditions and decrease how much they use their arm in everyday activities, often observed in stroke survivors despite clinical motor and functional improvement (Rand & Eng, 2015).
Motor recovery after stroke may be improved by helping patients make better use of their kinematic redundancy by setting up training activities that allow the learner to find their own multiple stable kinematic solutions to accomplish different tasks (i.e., the principle of motor equivalence; Bernstein, 1967; Lashley, 1951; Tomita et al., 2017). An increase in redundancy may represent a greater range of regulation of the TSRT in specific joints (Figure 3b), but this has yet to be demonstrated. All such solutions should be regarded as being acceptable, and indeed within the range of joint combinations that should be considered as normal, as long as they do not include the use of joint rotations that may lead to undesirable consequences, such as pain, contractures, or learned nonuse.
Several studies have shown that the injured or damaged nervous system uses different motor patterns to compensate for a lost or altered DF. When the compensatory joint rotation falls within the range of acceptable motor solutions, such adaptations may be tolerated. However, if motor compensations such as excessive trunk flexion or rotation during reaching movement leads to undesirable consequences such as elbow flexor shortening or contracture (Ada et al., 1994), they should be avoided. In addition, brain connectivity may reorganize to reinforce the substitutive pattern so that later recovery of the original pattern may be more difficult or even impossible (Jones, 2017). In stroke survivors, to encourage the use of desirable joint movement patterns and avoid compensations, physical restraints of movements of certain segments or joints can be used, such as a trunk restraint or a wrist extension support. Although this may lead to a certain decrease in the amount of motor abundance in the system, a greater range of joint variability may be encouraged in other joints such as the elbow and shoulder, within the range of acceptability. Feedback should be focused on endpoint performance to allow the system to find the best set of joint rotations to perform the task. Feedback on the use of specific joints should only be given to limit the use of undesirable movements (compensations) that may decrease the possibility of improvement or recovery in that joint. The decision to limit compensations should be based on the patient’s potential to recover voluntary movement, and may even be discouraged if the potential to recover is low (Connell et al., 2021). Once a stable synergy is learned, it should be incorporated into whole-task practice. Recently, Krakauer et al. (2021) found that training in patients with subacute stroke that focused on movement quality, that is, aimed at restitution of premorbid movement patterns, led to motor performance improvements when compared with usual care. This seems to be a promising approach.
Future research should identify and translate principles of motor control into clinical practice and test their effectiveness in ameliorating movement deficits. One approach would be to use the UCM (Solnik et al., 2020) or other measures of motor abundance and adaptability to track motor recovery in neurological populations. Sensorimotor rehabilitation research and practice based on a comprehensive theory of motor control should drive the research agenda of the scientific community for the next decades. We propose that the indirect (i.e., physical) theoretical approach stemming from laws of nature that characterize motor action may be more relevant for designing effective interventions.
Concluding Comments
The purpose of the present paper was to discuss existing frameworks for understanding normal and abnormal movement. Motor control and rehabilitation are broad and growing disciplines, from preclinical to epidemiological studies, including several domains of science (e.g., neuroscience, psychology). These are exciting times for their scientific advance. Using a coherent terminology and theory-based reasoning to investigate motor actions may promote our understanding of motor processes and how to ameliorate sensorimotor impairments.
Acknowledgments
Levin is a distinguished James McGill Professor. Piscitelli was supported by the Fonds de la Recherche en Santé du Québec.
References
Ada, L., Canning, C., Carr, J.H., Kilbreath, S.L., & Shepherd, R.B. (1994). Task specific training of reaching and manipulation. In K.M.B. Bennet & U. Castiello (Eds.), Insights in reach and grasp movement (pp. 239–265). Elsevier-Biosoft.
Aldabe, D., Milosavljevic, S., & Bussey, M.D. (2012). Is pregnancy related pelvic girdle pain associated with altered kinematic, kinetic and motor control of the pelvis? A systematic review. European Spine Journal, 21(9), 1777–1787. https://doi.org/10.1007/s00586-012-2401-1
Bernhardt, J., Hayward, K.S., Kwakkel, G., et al. (2017). Agreed definitions and a shared vision for new standards in stroke recovery research: The stroke recovery and rehabilitation roundtable taskforce. International Journal of Stroke, 12, 444–450. https://doi.org/10.1177/1747493017711816
Bernstein, N.A. (1967). The co-ordination and regulation of movements. Pergamon Press.
Blouin, J., Bard, C., Teasdale, N., Paillard, J., et al. (1993). Reference systems for coding spatial information in normal subjects and a deafferented patient. Expermental Brain Research, 93, 324–331. https://doi.org/10.1007/BF00228401
Burke, R.E. (1981). Motor units: Anatomy, physiology, and functional organization. In V.B. Brooks (Ed.), Handbook of physiology. Section 1: The nervous system (Vol. II, Part I, Ch. 10, pp. 345–422).
Carr, J.H., & Shepherd, R.B. (2010). Neurological rehabilitation (2nd ed.). Churchill-Livingstone.
Chan-Viquez, D., Hasanbarani, F., Zhang, L., et al. (2020). Development of vertical and forward jumping skills in typically developing children in the context of referent control of motor actions. Developmental Psychobiology, 62(6), 711–722. https://doi.org/10.1002/dev.21949
Churchland, M.M., Cunningham, J.P., Kaufman, M.T., et al. (2012). Neural population dynamics during reaching. Nature, 487, 51–56. https://doi.org/10.1038/nature11129
Cluff, T., & Scott, S.H. (2016). Online corrections are faster because movement initiation must disengage postural control. Motor Control, 20, 162–170. https://doi.org/10.1123/mc.2015-0027
Connell, L.A., Chesworth, B., Ackerley, S., Smith, M.C., & Stinear, C.M. (2021). Implementing the PREP2 algorithm to predict upper limb recovery potential after stroke in clinical practice: A qualitative study. Physical Therapy, 101(5), Article pzab040. https://doi.org/10.1093/ptj/pzab040
Côté, J.N., Feldman, A.G., Mathieu, P.A., & Levin, M.F. (2008). Effects of fatigue on intermuscular coordination during repetitive hammering. Motor Control, 12(2), 79–92. https://doi.org/10.1123/mcj.12.2.79
Day, B.L., & Bancroft, M.J. (2018). Voluntary steps and gait initiations. Handbook of Clinical Neurology, 159, 107–118. https://doi.org/10.1016/B978-0-444-63916-5.00006-9
Einstein, A. (1916). Relativity: The special and general theory (Translation 1920). H Holt & Company.
Feldman, A.G. (1966). Functional tuning of the nervous system with control of movement or maintenance of a steady posture. II. Controllable parameters of the muscle. Biophysics, 11, 565–578.
Feldman, A.G. (2015). Referent control of action and perception. Springer-Nature.
Feldman, A.G. (2019). Indirect, referent control of motor actions underlies directional tuning of neurons. Journal of Neurophysiology, 121(3), 823–841. https://doi.org/10.1152/jn.00575.2018
Feldman, A.G., & Levin, M.F. (1995). The origin and use of positional frames of reference in motor control. Behavioral and Brain Sciences, 18, 723–806. https://doi.org/10.1017/S0140525X0004070X
Feldman, A.G., Levin, M.F., Garofolini, A., Piscitelli, D., & Zhang, L. (2021). Central pattern generator and human locomotion in the context of referent control of motor actions. Clinical Neurophysiology, 132(11), 2870–2889. https://doi.org/10.1016/j.clinph.2021.08.016
Feldman, A.G., & Orlovsky, G.N. (1972). The influence of different descending systems on the tonic stretch reflex in the cat. Experimental Neurology, 37, 481–494. https://doi.org/10.1016/0014-4886(72)90091-X
Flash, T., & Hogan, N. (1985). The coordination of arm movements: An experimentally confirmed mathematical model. Journal of Neuroscience, 5(7), 1688–1703. https://doi.org/10.1523/JNEUROSCI.05-07-01688.198529
Foerster, O. (1977). The motor cortex in man in the light of Hughlings Jackson’s Doctrines. In I.D. Payton, S. Hirt, & R. Newman (Eds.), Scientific bases for neurophysiological approaches to therapeutic exercise (pp. 13–18). FA Davis.
Gera, G., McGlade, K.E., Reisman, D.S., & Scholz, J.P. (2016). Trunk muscle coordination during upward and downward reaching in stroke survivors. Motor Control, 20(1), 50–69. https://doi.org/10.1123/mc.2014-0038
Gibson, J.J. (1966). The senses considered as perceptual systems. Houghton Mifflin.
Gomi, H., & Kawato, M. (1996). Equilibrium point control hypothesis examined by measured arm stiffness during multi joint movement. Science, 272(5258), 117–120. https://doi.org/10.1126/science.272.5258.117
Groenewegen, H.J. (2003). The basal ganglia and motor control. Neural Plasticity, 10(1–2), 107–120. https://doi.org/10.1155/NP.2003.107
Haken, H., Kelso, J.A., & Bunz, H. (1985). A theoretical model of phase transitions in human hand movements. Biological Cybernetics, 51(5), 347–356. https://doi.org/10.1007/BF00336922
Hammerbeck, U., Yousif, N., & Hoad, D., et al. (2017). Chronic stroke survivors improve reaching accuracy by reducing movement variability at the trained movement speed. Neurorehabilitation and Neural Repair, 31(6), 499–508. https://doi.org/10.1177/1545968317693112
Hasan, Z. (1986). Optimized movement trajectories and joint stiffness in unperturbed, inertially loaded movements. Biological Cybernetics, 53(6), 373–382. https://doi.org/10.1007/BF00318203
Henneman, E., Somjen, G., & Carpenter, D.O. (1965). Functional significance of cell size in spinal motoneurons. Journal of Neurophysiology, 28, 560–580. https://doi.org/10.1152/jn.1965.28.3.560
Herz, A.V.M. (2006). Modeling single-neuron dynamics and computations: A balance of detail and abstraction. Science, 314(5796), 80–85. https://doi.org/10.1126/science.1127240
Hodges, P.W., & Richardson, C.A. (1996). Inefficient muscular stabilization of the lumbar spine associated with low back pain. A motor control evaluation of transversus abdominis. Spine, 21, 2640–2650.
Hollerbach, J.M. (1982). Computers, brains, and the control of movement. Trends in Neuroscience, 5, 189–192. https://doi.org/10.1016/0166-2236(82)90111-4
Jeannerod, M. (1984). The timing of natural prehension movements. Journal of Motor Behavior, 16, 235–254. https://doi.org/10.1080/00222895.1984.10735319
Jones, T.A. (2017). Motor compensation and its effects on neural reorganization after stroke. Nature Review Neuroscience, 18(5), 267–280. https://doi.org/10.1038/nrn.2017.26
Kawato, M. (1999). Internal models for motor control and trajectory planning. Current Opinion in Neurobiology, 9(6), 718–727. https://doi.org/10.1016/s0959-4388(99)00028-8
Kelso, J.A. (1984). Phase transitions and critical behavior in human bimanual coordination. American Journal of Physiology, 246(6), Article R1000. https://doi.org/10.1152/ajpregu.1984.246.6.R1000
Kitago, T., & Krakauer, J.W. (2013). Motor learning principles for neurorehabilitation. Handbook of Clinical Neurology, 110, 93–103. https://doi.org/10.1016/B978-0-444-52901-5.00008-3
Kleim, J.A., & Jones, T.A. (2008). Principles of experience-dependent neural plasticity: Implications for rehabilitation after brain damage. Journal of Speech, Language and Hearing Research, 51(Suppl. 1), S225–S239. https://doi.org/10.1044/1092-4388(2008/018
Krakauer, J.W., Kitago, T., Goldsmith, J., et al. (2021). Comparing a novel neuroanimation experience to conventional therapy for high-dose intensive upper-limb training in subacute stroke: The smarts2 randomized trial. Neurorehabilitation and Neural Repair, 35(5), 393–405. https://doi.org/10.1177/15459683211000730
Labanca, L., Mosca, M., & Ghislieri, M., et al. (2021). Muscle activations during functional tasks in individuals with chronic ankle instability: A systematic review of electromyographical studies. Gait & Posture, 90, 340–373. https://doi.org/10.1016/j.gaitpost.2021.09.182
Lashley, K.S. (1951). The problem of serial order in behavior. In L.A. Jeffress (Ed.), Cerebral mechanisms in behavior. Wiley.
Latash, M.L. (2012). The bliss of motor abundance. Experimental Brain Research, 217(1), 1–5. https://doi.org/10.1007/s00221-012-3000-4
Latash, M.L. (2020). On primitives of motor control. Motor Control, 24(2), 318–346. https://doi.org/10.1123/mc.2019-0099
Latash, M.L., Scholz, J.P., & Schöner, G. (2007). Toward a new theory of motor synergies. Motor Control, 11(3), 276–308. https://doi.org/10.1123/mcj.11.3.276
Latash, M.L., & Zatsiorsky, V.M. (2016). Biomechanics and motor control: Defining central concepts. Elsevier.
Leiras, R., Cregg, J.M., & Kiehn, O. (2022). Brainstem circuits for locomotion. Annual Review of Neuroscience, 45. https://doi.org/10.1146/annurev-neuro-082321-025137
Lepelley, M.C., Thullier, F., Koral, J., & Lestienne, F.G. (2006). Muscle coordination in complex movements during Jeté in skilled ballet dancers. Experimental Brain Research, 175, 321–331. https://doi.org/10.1007/s00221-006-0552-1
Levin, M.F., & Demers, M. (2020). Motor learning in neurological rehabilitation. Disability and Rehabilitation, 22, 1–9. https://doi.org/10.1080/09638288.2020.1752317
Levin, M.F., Liebermann, D.G., Parmet, Y., & Berman, S. (2015). Compensatory versus non-compensatory shoulder movements used for reaching in stroke. Neurorehabilitation and Neural Repair, 30(7), 635–646. https://doi.org/10.1177/1545968315613863
Low, M.A. (2018). Time to reflect on motor control in musculoskeletal physical therapy. Journal of Orthopedics and Sports Physical Therapy, 48(11), 833–836. https://doi.org/10.2519/jospt.2018.0614
MacKinnon, C.D., & Winter, D.A. (1993). Control of whole body balance in the frontal plane during human walking. Journal of Biomechanics, 26(6), 633–644. https://doi.org/10.1016/0021-9290(93)90027-c
Marchal-Crespo, L., & Reinkensmeyer, D.J. (2009). Review of control strategies for robotic movement training after neurologic injury. Journal of NeuroEngineering and Rehabilitation, 6, 20. https://doi.org/10.1186/1743-0003-6-20
Mathis, M.W., & Mathis, A. (2020). Deep learning tools for the measurement of animal behavior in neuroscience. Current Opinion in Neurobiology, 60, 1–11. https://doi.org/10.1016/j.conb.2019.10.008
Matthews, P.B.C. (1959). A study of certain factors influencing the stretch reflex of the decerebrate cat. Journal of Physiology, 147, 547–564. https://doi.org/10.1113/jphysiol.1959.sp006261
McClelland, V.M., Miller, S., & Eyre, J.A. (2001). Short latency heteronymous excitatory and inhibitory reflexes between antagonist and heteronymous muscles of the human shoulder and upper limb. Brain Research, 899(1–2), 82–93. https://doi.org/10.1016/S0006-8993(01)02097-2
McLoughlin, J. (2020). Ten guiding principles for movement training in neurorehabilitation. OpenPhysio Journal. https://doi.org/10.14426/art/1260
Michaelsen, S., Luta, A., Roby-Brami, A., & Levin, M.F. (2001). Effect of trunk restraint on the recovery of reaching movements in hemiparetic patients. Stroke, 32, 1875–1883.
Newell, K.M. (1985). Constraints on the development of coordination. In M.G. Wade & H.T.A. Whiting (Eds.), Motor development in children: Aspects of coordination and control (pp. 341–360). Martinus Nijhoff. https://doi.org/10.1007/978-94-009-4460-2_19
Newell, K.M. (1996). Change in movement and skill: Learning, retention, and transfer. In M.L. Latash & M. Turvey (Eds.), Dexerity and its development. Erlbaum.
Nichols, T.R., & Steeves, J.D. (1986). Resetting of resultant stiffness in ankle flexor and extensor muscles in the decerebrated cat. Experimental Brain Research, 62, 401–410. https://doi.org/10.1007/BF00238859
Park, J., Zatsiorsky, V.M., & Latash, M.L. (2010). Optimality vs variability: An example of multi-finger redundant tasks. Experimental Brain Research, 207(1–2), 119–132. https://doi.org/10.1007/s00221-010-2440-y
Pinto, S.M., Boghra, S.B., & Macedo, L.G., et al. (2021). Does motor control exercise restore normal morphology of lumbar multifidus muscle in people with low back pain? — A systematic review. Journal of Pain Research, 14, 2543–2562. https://doi.org/10.2147/JPR.S314971
Rand, D., & Eng, J.J. (2015). Predicting daily use of the affected upper extremity 1 year after stroke. Journal of Stroke and Cerebrovascular Diseases, 24(2), 274–283. https://doi.org/10.1016/j.jstrokecerebrovasdis.2014.07.039
Reed, E.S. (1988). Applying the theory of action systems to the study of motor skills. In O.G. Meijer & K. Roth (Eds.), Complex movement behaviour: The motor-action controversy (pp. 45–86). Elsevier.
Reisman, D.S., & Scholz, J.P. (2003). Aspects of joint coordination are preserved during poiting in persons with post-stroke hemiparesis. Brain, 126(11), 2510–2527. https://doi.org/10.1093/brain/awg246
Roland, P.E., Larsen, B., Lassen, N.A., & Skinhøj, E. (1980). Supplementary motor area and other cortical areas in organization of voluntary movements in man. Journal of Neurophysiology, 43(1), 118–136. https://doi.org/10.1152/jn.1980.43.1.118
Saes, M., Refai, M.I.M., van Kordelaar, J., et al. (2021). Smoothness metric during reach-to-grasp after stroke: Part 2. Longitudinal association with motor impairment. Journal of NeuroEngineering and Rehabilitation, 8(1), 144. https://doi.org/10.1186/s12984-021-00937-w
Santello, M., & Soechting, J.F. (2000). Force synergies for multifingered grasping. Experimental Brain Research, 133, 457–467. https://doi.org/10.1007/s002210000420
Scholz, J.P., & Schöner, G. (1999). The uncontrolled manifold concept: Identifying control variables for a functional task. Experimental Brain Research, 126, 289–306. https://doi.org/10.1007/s002210050738
Scholz, J.P., Schöner, G., & Latash, M.L. (2000). Identifying the control structure of multijoint coordination during pistol shooting. Experimental Brain Research, 135(3), 382–404. https://doi.org/10.1007/s002210000540
Shaikh, T., Goussev, V., Feldman, A.G., & Levin, M.F. (2014). Arm-trunk coordination for beyond the reach movements in adults with hemiparesis. Neurorehabilitation and Neural Repair, 28(4), 355–366. https://doi.org/10.1177/1545968313510973
Sherrington, C. (1947). The integrative action of the nervous system (2nd ed.). Yale University.
Shumway-Cook, A., & Woollacott, M.H. (2012). Motor control: Translating research into clinical practice (6th ed.). Wolters Kluwer Health, Lippincott Williams & Wilkins.
Sobinov, A.R., & Bensmaia, S.J. (2021). The neural mechanisms of manual dexterity. Nature Reviews Neuroscience, 22(12), 741–757. https://doi.org/10.1038/s41583-021-00528-7
Solnik, S., Furmanek, M.P., & Piscitelli, D. (2020). Movement quality: A novel biomarker based on principles of neuroscience. Neurorehabilitation and Neural Repair, 34(12), 1067–1077. https://doi.org/10.1177/1545968320969936
Solnik, S., Pazin, N., Coelho, C.J., Rosenbaum, D.A., Scholz, J.P., Zatsiorsky, V.M., & Latash, M.L. (2013). End-state comfort and joint configuration variance during reaching. Experimental Brain Research, 225(3), 431–442. https://doi.org/10.1007/s00221-012-3383-2
Suresh, A.K., Goodman, J.M., Okorokova, E.V., et al. (2020). Neural population dynamics in motor cortex are different for reach and grasp. eLife, 9, Article e58848. https://doi.org/10.7554/eLife.58848
Svoboda, K., & Li, N. (2018). Neural mechanisms of movement planning: Motor cortex and beyond. Current Opinion in Neurobiology, 49, 33–41. https://doi.org/10.1016/j.conb.2017.10.023
Todorov, E. (2000). Direct cortical control of muscle activation in voluntary arm movements: A model. Nature Neuroscience, 3, 391–398. https://doi.org/10.1038/73964
Todorov, E., & Jordan, M.I. (2002). Optimal feedback control as a theory of motor coordination. Nature Neuroscience, 5(11), 1226–1235. https://doi.org/10.1038/nn963
Tomita, Y., Feldman, A.G., & Levin, M.F. (2017). Referent control and motor equivalence of reaching from standing. Journal of Neurophysiology, 117(1), 303–315. https://doi.org/10.1152/jn.00292.2016
van Dijk, L., & Bongers, R.M. (2014). The emergence of an action system: The organization of gaze in creating novel tools. Ecological Psychology, 14(26), 177–197. https://doi.org/10.1080/10407413.2014.929476
van Dijk, L., van der Sluis, C., & Bongers, R.M. (2017). Reductive and emergent views on motor learning in rehabilitation practice. Journal of Motor Behavior, 49(3), 244–254. https://doi.org/10.1080/00222895.2016.1191418
von Holst, E. (1954). Relations between the central nervous system and the peripheral organs. British Journal of Animal Behavior, 2, 89–94. https://doi.org/10.1016/S0950-5601(54)80044-X
Weinrich, M., & Wise, S.P. (1982). The premotor cortex of the monkey. Journal of Neuroscience, 2(9), 1329–1345. https://doi.org/10.1523/JNEUROSCI.02-09-01329.1982
Wing, A.M. (2000). Motor control: Mechanisms of motor equivalence of handwriting. Current Biology, 10(6), R245–R248. https://doi.org/10.1016/S0960-9822(00)00375-4
Wolpert, D.M., & Ghahramani, Z. (2000). Computational principles of movement neuroscience. Nature Neuroscience, 3, 1212–1217. https://doi.org/10.1038/81497
Wolpert, D.M., & Jessell, T.M. (2021). Principles of sensorimotor control. In E.R. Kandel, J.D. Koester, S.H. Mach, & S.A. Siegelbaum (Eds.), Principles of neuroscience (6th ed.). McGraw-Hill.
World Health Organization. (2001). International classification of functioning, disability and health: ICF. World Health Organization.
Worsley, P., Warner, M., Mottram, S., et al. (2013). Motor control retraining exercises for shoulder impingement: Effects on function, muscle activation, and biomechanics in young adults. Journal of Shoulder and Elbow Surgery, 22, e11–e19. https://doi.org/10.1016/j.jse.2012.06.010
Zarzycki, R., Malloy, P., Eckenrode, B.J., Fagan, J., Malloy, M., & Mangione, K.K. (2022). Application of the 4-element movement system model to sports physical therapy practice and education. International Journal of Sports Physical Therapy, 17(1), 18–26. https://doi.org/10.26603/001c.30173
Zhang, L., Feldman, A.G., & Levin, M.F. (2018). Vestibular and corticospinal control of human body orientation in the gravitational field. Journal of Neurophysiology, 120(6), 3026–3041. https://doi.org/10.1152/jn.00483.2018
