The role of binocular vision in the control and development of visually guided upper limb movements

Abstract

Vision provides a key sensory input for the performance of fine motor skills, which are fundamentally important to daily life activities, as well as skilled occupational and recreational performance. Binocular visual function is a crucial aspect of vision that requires the ability to combine inputs from both eyes into a unified percept. Summation and fusion are two aspects of binocular processing associated with performance advantages, including more efficient visuomotor control of upper limb movements. This paper uses the multiple processes model of limb control to explore how binocular viewing could facilitate the planning and execution of prehension movements in adults and typically developing children. Insight into the contribution of binocularity to visuomotor control also comes from examining motor performance in individuals with amblyopia, a condition characterized by reduced visual acuity and poor binocular function. Overall, research in this field has advanced our understanding of the role of binocular vision in the development and performance of visuomotor skills, the first step towards developing assessment tools and targeted rehabilitation for children with neurodevelopment disorders at risk of poor visuomotor outcomes.

This article is part of a discussion meeting issue ‘New approaches to 3D vision’.

1. Introduction

Binocularity is a key aspect of visual function that requires the ability to combine inputs from each eye into a unified percept. Because each eye receives identical or similar visual input from an overlapping visual field, separate retinal images must be fused to achieve binocular single vision—a pinnacle of visual processing that engages an extensive neural network in the primary visual cortex, extrastriate cortices and association areas [1]. Accumulating research demonstrates that compared to monocular viewing, binocular viewing is associated with more efficient performance of visually guided movements in children and adults [2–5]. Likewise, individuals with abnormal binocular vision perform more poorly on motor tasks than those with normal vision [6–11]. This paper aims to evaluate how binocular visual processing contributes to the control of upper limb reaching and grasping movements. The paper begins with a brief summary of binocular processing in typical and atypical development. Next, the contribution of binocular vision to upper limb movement control is examined using the multiple processes model of limb control [12–14]. This approach reveals that vergence and stereopsis provide distinct advantages for optimal sensorimotor control and the development of manual dexterity. Understanding the contribution of binocular vision to the control and performance of fine motor skills could reveal fundamental mechanisms underlying sensorimotor control and provide some guidance for assessing visuomotor functions in children and adults with visual impairments.

2. Binocular visual function: summation and fusion

In this paper, binocular visual function is an umbrella term that refers to processes involved in combining the inputs from both eyes resulting in improved performance. Figure 1 depicts the two domains of binocular visual function—summation and fusion—that have been studied extensively [15].

Figure 1.

Binocular visual function refers to processes involved in combining the inputs from both eyes. Summation and fusion are two aspects of binocularity. Binocular summation refers to enhancement in performance when viewing with both eyes (compared to one eye) for tasks that do not require depth perception. For example, lower thresholds have been documented for visual acuity and contrast sensitivity during binocular viewing. Binocular fusion refers to achieving and maintaining eye alignment (motor fusion via vergence eye movements), which allows integration of the retinal images from the left and right eyes into a single unified percept (sensory fusion). Stereopsis is the hallmark of normal binocular vision and relies on bifoveal fixation (blue outline). Amblyopia results from decorrelated binocular experience during development as a result of strabismus (eye deviation), anisometropia (unequal refractive errors between eyes) or ocular media opacity (deprivation) (red outline). Amblyopia is associated with reduced or absent sensory fusion. Binocular summation may be reduced or absent, dependent on individual clinical profile and the method used for visual function assessment. (Online version in colour.)

Binocular summation refers to an enhancement in performance when viewing with both eyes (compared to one eye) for tasks that do not require depth perception [16,17]. For example, lower thresholds have been documented for visual acuity [18], contrast sensitivity [19] and global motion during binocular viewing [20]. Typically, the improvement ranges from 1.4 to 2 and is dependent on the properties of the stimulus, such as spatial and temporal frequency [21]. Binocular summation can be explained by several mechanisms, including probability summation, increase in signal-to-noise ratio or neural summation [17].

Binocular fusion, on the other hand, refers to achieving and maintaining eye alignment (motor fusion via vergence eye movements), which allows integration of the retinal images from both eyes into a single unified percept (sensory fusion). Both sensory and motor fusion are, to some extent, co-dependent processes driven by binocular disparity [22,23]. When the eyes are aligned (i.e. motor fusion), images of an object falling on the horopter (i.e. the location of objects that can be fused binocularly [24]) stimulate corresponding retinal points in both eyes resulting in a unified percept (i.e. sensory fusion). Binocular single vision is maintained when images in the two eyes activate non-corresponding retinal points within a narrow region around the horopter, referred to as Panum's fusional area [25]. The ability to fuse images with small binocular disparity within Panum's area is the basis of stereopsis (i.e. perception of depth that arises from the cortical fusion of disparity cues), and bifoveal fixation is required to achieve fine stereoacuity (i.e. a measure of quantum of depth that can be detected and recorded in arcsec). Images falling outside Panum's region are perceived as double (i.e. physiologic diplopia) and provide the stimulus that initiates fusional vergence eye movements.

Inputs from each eye are transmitted along the optic nerve such that axons carrying information from the nasal retina decussate at the optic chiasm. This partial decussation provides the structural foundation for binocularity: neurons in the primary visual cortex processing input from the contralateral visual field receive converging inputs from both eyes [26]. Neurophysiological studies demonstrate that binocular neural responses contribute to two distinct processes: summation and stereopsis. Neurons receiving input from corresponding retinal points in each eye provide the neurophysiological basis for binocular summation. Activation of neurons tuned to binocular disparity within Panum's area supports stereoscopic depth perception [27,28]. It has been proposed that disparities in the diplopic range could provide useful information about depth (coarse stereopsis) [29]. Thus, binocular neurons in the primary visual cortex and multiple extrastriate areas where they project form the basis of binocular single vision—the ability to integrate information from both eyes into a unified percept with a vivid impression of depth.

Development of binocularity occurs postnatally, with binocular summation and fusion emerging in the first few months after birth. They continue to improve throughout childhood provided that each eye has normal visual acuity and the eyes are aligned (i.e. orthophoria). The onset of binocular summation of luminance [30–32] and gross stereopsis [33,34] occurs between two and four months of age in humans, which is highly dependent on visual experience [35,36]. After its emergence, stereoacuity improves rapidly to approximately 300 arcsec in infants at 12–18 months [37]. Further improvements are more gradual, with thresholds ranging between 200 and 100 arcsec in pre-school children [38–40]. Using clinical tests, the age of maturation has been estimated to be between 6 and 9 years, with stereoacuity threshold values reaching the adult level of better than 40 arcsec [41,42]; however, the maturation age is somewhat dependent on the test used for assessment [43]. Notably, in contrast to clinical tests, which have a fixed, coarse step sizes between measured targets and are usually based on a single trial, psychophysical testing with repeated trials reveals that the processing of small disparities is still immature at the age of 14 years [44]. Finally, it is important to note that adult thresholds have been estimated to be as low as 2–5 arcsec using psychophysical laboratory tests [15]. Therefore, the current clinical tests grossly underestimate the sensitivity of the human stereoscopic system.

Evaluation of motor fusion includes assessment of eye alignment and fusional vergence. Research demonstrates reliable vergence eye movements in one-month-old infants in response to static [45] or moving targets [46,47] presented at different distances or with a prism placed in front of one eye [48], with the precision of these movements improving over the first four months. Thus, vergence responses appear to emerge earlier than stereopsis. A recent study that examined the concurrent maturation of sensory and motor fusion in pre-school children supports tight coordination between reflex vergence response and a percept of a single fused image. Notably, the fusional (i.e. Panum's) area was larger in younger children than in adults; thus, disparities perceived as diplopic by adults were perceived as single by children [49]. Maturation of some aspects of the vergence system continues into middle childhood, with the latency of vergence reaching adult levels at around 10–12 years [50].

Normal vision development is severely disrupted by amblyopia—a common paediatric condition affecting 3–5% of children and adults [9,51]. Amblyopia or ‘lazy eye' is a neurodevelopmental disorder caused by decorrelated binocular visual experience in infants and young children. It results from strabismus (i.e. eye deviation), anisometropia (i.e. unequal refractive error between the two eyes) or, more rarely, ocular media opacity (e.g. congenital cataracts) [52] (figure 1). The hallmark of amblyopia is reduced visual acuity in one or rarely both eyes, which cannot be immediately corrected using prescription lenses. In addition to spatial acuity deficits, amblyopia is associated with reduced contrast sensitivity (particularly at high spatial frequencies) as well as deficits in processing of global form and motion [53–55]; these deficits may also be present when viewing with the unaffected (i.e. fellow) eye [56,57]. Notably, higher-level deficits in attention, visual search or executive function were demonstrated in adults [58–64] and children [65] with amblyopia. However, not all aspects of attentional processing are affected, such as for example covert attention [66–68]. The pattern of visual deficits varies with the type of amblyopia [69], but in general, individuals with strabismic amblyopia tend to have poorer binocular vision than those with anisometropic amblyopia [69,70]. For example, patients with anisometropic amblyopia are more likely to have measurable stereopsis when compared to strabismic amblyopia [51,71]. It is important to note that patients with no measurable stereopsis may still have some degree of sensory fusion, which can be assessed using the Worth 4 Dot test [72,73]. Complete sensory suppression of the amblyopic eye is more likely in individuals with strabismus of large magnitude, which prevents motor fusion, and thus, sensory fusion is precluded. Studies examining binocular outcomes after congenital cataracts indicate that motor and sensory fusion are possible in cases with early surgical intervention [74,75].

Although amblyopia has traditionally been considered a monocular disorder, recent research highlights reduced binocularity as a core deficit [9,51,76,77]. Reduced or absent binocular summation of acuity [18,78], contrast sensitivity [79–81] and global motion [82] have been well documented [83]. The apparent deficit in contrast summation can be ameliorated during dichoptic viewing when the stimulus presented to the fellow eye has lower contrast than the stimulus shown to the amblyopic eye [84]. Thus, binocular summation is possible in most individuals with amblyopia when the two eyes are aligned and receive balanced inputs [85]. Indeed this finding is the foundation of recently developed binocular treatments for amblyopia [86,87]. To summarize, deficits in sensory and perceptual processing in amblyopia have been examined extensively over the past 40 years [76]; in contrast, the consequences of reduced binocularity on the performance of visually guided movements have been explored with less vigour [6].

3. Visual control of upper limb movements

Vision provides a key sensory input for the planning and executing of goal-directed actions [88]. This section starts with an overview of a conceptual model that explains visuomotor control and then considers how binocular viewing may contribute to movement planning and execution. The role of visual feedback in upper limb reaching performance has been studied extensively over the pst 120 years [89]. Woodworth [90] proposed a two-component model of limb control that includes an initial ballistic phase when the limb moves towards the target (i.e. the acceleration phase) followed by a feedback error-correcting phase (i.e. the deceleration phase) to ensure accuracy as the limb approaches the target. More recently, Elliott et al. [13,14] extended the model by outlining the multiple processes involved in optimizing visually guided movements. Similar to Woodworth's model, the two main phases are impulse (i.e. feedforward) and limb-target (i.e. feedback) control; however, Elliott and colleagues emphasize two aspects of impulse control. First, the impulse phase is not entirely ballistic. It is now widely accepted that movement programming is associated with the formation of an internal representation of the movement (i.e. efference copy) and the corresponding predicted sensory consequences (i.e. corollary discharge) via an internal model—a mental simulation of the planned action [91]. Since efference copy and corollary discharge are available prior to movement initiation, they are part of an internal feedback loop that can be used to correct errors shortly after movement initiation. Second, Elliott's model emphasizes the importance of initial conditions for impulse control, including the availability of current and expected sensory information, as well as the stochastic properties of sensory noise and potential error related to force specification (i.e. motor noise). In essence, faster movements require higher initial force, which is associated with increased limb trajectory variability [92].

Moreover, initial conditions such as the quality and availability of visual input directly impact movement planning and the formation of the internal model, which affect the initial phase of movement execution (i.e. feedforward/impulse control). As the movement progresses, afferent proprioceptive and visual inputs about limb and target are continuously monitored to ensure that the movement goal is achieved successfully (i.e. feedback/limb-target control). When visual feedback is restricted, accuracy and precision are reduced [93]. Conversely, when visual feedback is available, errors associated with motor planning are corrected in the latter part of the movement trajectory [94,95]. Critically, these corrections are dependent on the quality of feedback—if visual or kinaesthetic input is limited, errors are less likely to be detected and amended. In all, impulse and limb-target control are tightly regulated on a trial-by-trial basis, tuned to biomechanical constraints and experimental context, and influenced by practice and learning [13,89,96].

To summarize, executing fast, accurate and precise upper limb movements requires an interplay between feedforward (i.e. impulse) and feedback (i.e. limb-target) control. Woodworth's original two-phase model has been updated based on extensive research that highlights the complexity of sensorimotor control of upper limb movements.

4. Contribution of binocular viewing to upper limb movement control

The contribution of binocular viewing to motor performance has been studied extensively over the past three decades in visually normal adults and children [2,3,97–101]. Elliott's model provides a useful framework for evaluating these results. Here, we consider how different aspects of binocular function contribute to impulse and limb-target control. Importantly, the framework's utility is that differences in sensorimotor control are evaluated via objective measures of limb kinematics where events that occur early in the trajectory during the acceleration phase reflect impulse control; in contrast, events that occur in the deceleration phase reflect limb-target control (figure 2). Elliott's model of visuomotor control was developed and tested extensively using aiming tasks. Here, we extend the application of the model to prehension, which is an upper limb movement requiring spatiotemporal coordination between limb transport and formation of grip aperture [102]. In this case, the internal model formation will include grasp programming. Research demonstrates that peak grip aperture, which occurs during reach deceleration, is scaled precisely with object size [102]. Conversely, the initial load and grip force magnitude are scaled to the expected object weight and material properties [103]. Thus, grip aperture scaling and initial force application reflect impulse control of grasp, while the latter part of grasp application reflects limb-target adjustments using visuo-haptic input.

Figure 2.

Performing visually guided prehension movements relies on multiple processes to optimize movement speed, accuracy and energy. Kinematic analysis of velocity (illustrated here) and acceleration trajectory provides insight into sensorimotor control where criterion thresholds (shown as red ellipses on the velocity plot) are used to identify distinct processes. Movement planning is directly influenced by the available and expected sensory information and the stochastic properties of neural-motor noise. Movement execution involves impulse control, which accelerates the limb towards the target and limb-target control, which relies on afferent feedback to reduce endpoint error. The initial limb direction during reach acceleration and peak velocity is influenced by the state of the vergence system. The deceleration part of the reach trajectory, in particular reach-to-grasp coordination, is influenced by stereopsis. Grasp planning and execution are more efficient with fine stereopsis. (Online version in colour.)

Binocularity affords unique advantages: summation, ocular vergence and stereopsis [104]. Viewing with two eyes provides the brain with independent but concordant signals, which can be combined to improve the signal-to-noise ratio [105]. Accordingly, binocular viewing provides a significant advantage for the performance of motor tasks, particularly in dim lighting when visibility and contrast are low [106,107]. The main benefit of binocular summation for motor control would be expected during movement planning, specifically stimulus detection. Indeed, saccades are initiated more quickly [108,109] and the ocular following response is more robust [110] during binocular compared to monocular viewing. In contrast, not all studies report a significant improvement in manual reaction time [5,111,112]. It is possible that binocular advantage for manual responses is more difficult to detect in statistical analysis due to a larger within- and between-participant variance.

After target detection, localization is the next crucial stage of sensorimotor processing that requires encoding the target's direction and distance. For visually guided movements, the retinal position of a target must be combined with an eye position signal to compute the target location in egocentric coordinates [113]. According to Hering's law of visual direction, the apparent location of an object falls at the intersection of the visual axis passing through the fovea of each eye. Thus, the brain uses the position of the two eyes to derive the cyclopean direction of the target [114]. Disparity vergence drives motor fusion to ensure that the image of a fixated object falls on the fovea of both eyes; hence, the main benefit of ocular vergence is for impulse control; specifically, the formation of an internal model for reach direction during movement planning. Restricting visual input by covering one eye results in phoria, a temporary eye deviation that occurs when binocular vision is disrupted. Because phoria affects the vergence angle, perceptual judgements of direction [115] and the initial direction of limb movement are impacted [116]. During open-loop pointing, the magnitude and direction of phoria are correlated with endpoint error [117]. Crucially, errors in the initial reach trajectory are quickly and seamlessly corrected if visual feedback is available during movement execution [118].

The muscular effort associated with a particular vergence angle may be one of the cues contributing to estimating the target's absolute (metric) distance [119]. Placing a prism in front of the eye(s) alters the vergence effort, which has been associated with changes in reach peak velocity [120,121]. Specifically, using a base-out prism increases the effort of convergence which is associated with lower reach peak velocity, consistent with an interpretation that a target is perceived as being closer. In contrast, using a base-in prism decreases vergence demand, which is associated with higher reach peak velocity, consistent with a target localized further away. Notably, these results were obtained using 5–8 prism diopter (PD) prisms. Research reports mixed findings regarding changes in reach peak velocity during monocular viewing: some studies found lower peak velocity whereas others reported no difference [97,101,111,122,123].

On the one hand, reduction in reach peak velocity during monocular viewing could reflect an underestimation of target distance if the covered eye experiences esophoria (convergent misalignment). However, studies show that exophoria (mean range 3–5 PD) is more common in adults at near viewing distances [124,125]. Thus, given the divergent misalignment, the target location would be overestimated and a higher peak velocity would be expected. On the other hand, it is possible that the magnitude of phoria during monocular viewing is not large enough to have a reliable effect on reach peak velocity. Instead, reduced peak velocity may reflect the perceived difficulty of a task [126,127] where a higher safety margin is adapted due to increased sensory uncertainty of the target's direction and depth when viewing with one eye.

Finally, binocular stereopsis provides the most precise relative depth information, with adult threshold values below 10 arcsec [104]. Binocular advantage for grasping has been well documented [98,127–131]: monocular viewing is associated with increased peak grip aperture, prolonged reach deceleration when approaching the object and poor grasp execution. In the context of Elliott's model, stereopsis provides important input for both impulse and limb-target control of prehension movements. First, terminal reach culminates in finger placement on the object. The disparity between the approaching finger and object could contribute to limb-target control and improve the reach-to-grasp coordination [132,133]. In this case, stereopsis contribution would be expected to be greater for smaller objects where the margin of error for placing the fingers precisely is small [134]. Second, fine stereopsis affords the formation of an accurate internal representation of an object's intrinsic properties, including its 3D size, curvature and texture [135,136]. Thus, stereopsis may be critical for impulse control of grasping, including the initial application of grip and load force, which are programmed predictively based on visual information [137]. A recent study showed that reducing stereopsis in the range of 100–200 arcsec was associated with lower initial grip force, consistent with the idea that impulse control of grasping was compromised, leading to prolonged grasp duration suggestive of increased reliance on haptic feedback [129]. Thus, it is plausible that fine stereoacuity provides a critical sensory input about object features to facilitate grasp planning and execution. Indeed, these findings are consistent with the idea that the main function of stereopsis is to guide efficient prehension [138].

5. Role of binocular vision during the development of upper limb control

Visuomotor coordination is a key aspect of manual dexterity—a hallmark of human evolution. Improvements in fine motor skills continue into adolescence, which is associated with the maturation of the corticospinal tract [139]. Similarly, maturation of the superior longitudinal fasciculus—a major axonal tract connecting the posterior (i.e. occipitoparietal) and frontal (i.e. motor) cortical areas—continues into adolescence [140]. Thus, cortical maturation of visual, somatosensory and motor processing provides the neural substrate for the refinement of fine motor skills. Behavioural studies demonstrate that motor control development is non-monotonic [141,142]. Typically developing children younger than 7 years tend to perform aiming movements ballistically, suggesting impulse control matures first. A major change in the control strategy occurs between the ages of 7 and 8 years when children begin to integrate a feedback-based (i.e. limb-target) form of control. This change is paradoxically associated with a temporary shift to prolonged movements and less accurate aiming when visual feedback is restricted [141]. Refinement and optimization of visuomotor control extend into adolescence [143] and most likely depend on good vision, including normal acuity as well as sensory and motor fusion [100].

Binocular vision disorders are common in children [144,145]; therefore, it is crucial to understand the role of binocularity in the development of motor proficiency [8]. In contrast to the extensive literature in adults, relatively fewer studies examined the contribution of binocular viewing to upper limb reaching and grasping in typically developing children. The first study conducted with children compared prehension performance during binocular and monocular viewing in two age groups (5–6 and 10–11 years old) [2]. Both groups of children had a shorter reaction time when viewing binocularly, suggesting an advantage for movement planning. However, there was no significant difference between viewing conditions for peak velocity or movement duration. Contrasting with the adult studies, peak grip aperture was not different between the viewing conditions, although aperture scaling was reduced in the younger children viewing monocularly. Consistent with the idea that limb-target control develops after the age of 7–8 years, binocular viewing was associated with a shorter reach deceleration phase only in the older group—suggesting that disparity facilitated reach-to-grasp coupling.

A subsequent study compared performance across tasks that varied in difficulty in a larger cohort of children and adolescents (5–13 years old) [99]. Results demonstrated that binocular viewing provides a greater advantage for tasks that require higher precision. Specifically, grasping a small bead and threading it onto a needle benefited from binocular viewing to a greater extent compared to a pegboard task. The bead threading task consists of reaching, grasping, transporting and placing a small bead on a vertical needle; therefore, a lack of binocular cues could disrupt any of these task components. A detailed kinematic analysis in a subset of 8–14 years old children revealed that different aspects of binocular vision contributed to the control of distinct phases of the task: better fusional vergence was associated with a higher peak velocity of reaching, and better stereoacuity was associated with shorter grasp execution [100].

To summarize, research findings with typically developing children highlight the role of normal binocular vision in the development of upper limb control. Interpreting the results in the context of Elliott's model of visuomotor control leads to a similar conclusion to the inference from the adult studies: ocular vergence is associated with impulse control of reaching whereas stereopsis is associated with reach-to-grasp coupling and grasp execution.

6. The effect of amblyopia on upper limb movement control

Unsurprisingly, performance on standardized clinical tests of fine motor skills in children with amblyopia tends to be poorer than age-matched controls, with the greatest deficits found on tests that require accuracy, precision and speed [8,10,146,147]. Clinical tests performed with a stopwatch provide important information about performance deficits but no insight about the underlying control issue. Detailed kinematic analysis for a prehension task in children with amblyopia revealed deficits in motor control that varied with age and the extent of residual binocular vision [148,149]. Children of 5–6 years old performed very poorly: reaching movement duration was 30% longer with significantly more reach and grasp errors. Consistent with previous findings [2], the control group's performance was ballistic with no difference in deceleration interval between binocular and monocular viewing. In contrast, children with amblyopia had longer acceleration and deceleration phases, suggesting a different control mode. As noted by the authors, younger children had more difficulty localizing the object and planning their movement, which impacted their impulse control. Consequently, errors in movement planning had to be corrected during movement execution (i.e. prolonged deceleration), which is consistent with the four-fold increase in spatial path corrections reported in that study. In contrast, older children with amblyopia (7–9 years old) exhibited relatively fewer deficits. Compared to matched controls, their movement duration was only approximately 8% longer, with fewer errors in reaching and grasping. Crucially, children with residual stereoacuity performed similarly to the control group on the reaching component, but grasp errors were still evident. Analogous results were found in a study with a different cohort of children (5–14 years old) who showed persistent deficits in grasping that were correlated with stereoacuity [150]. Finally, improvement in manual dexterity was found in children who completed a treatment that targeted binocular function [151]. Taken together, two main findings emerge from these kinematic studies: (i) young children with amblyopia may lag behind their peers in the development of fine motor skills and (ii) older children with residual stereoacuity develop better grasping skills. More research is needed in the paediatric population to establish how maturation of manual dexterity is influenced by different treatments and recovery of binocularity.

An important question is whether upper limb control becomes normal in amblyopia during adulthood after years of practice with everyday actions involving prehension movements. Using the same task and kinematic approach as the study with children [149], persistent deficits during binocular viewing were demonstrated in adults with amblyopia, including a longer terminal approach towards the object, more grasping errors and prolonged grasp execution [152]. Hence, normal binocular vision seems essential for efficient reach-to-grasp coupling and grasping. In contrast, reach duration and peak velocity were comparable to the adult control group. Thus, some aspects of upper limb movement control attain normal values while others do not.

An intriguing insight into the effect of amblyopia on the control of visually guided aiming in adults was revealed in an experiment in which the instructions stressed both speed and accuracy [112,153]. Adults with amblyopia had longer movement time (approx. 100 ms), which allowed them to attain endpoint accuracy and precision that were comparable to the control group when viewing binocularly or with the fellow eye. However, endpoint error remained significantly higher during amblyopic eye viewing. In general, movements that are performed fast tend to be less accurate, which is referred to as speed–accuracy trade-off or Fitts' Law in motor control [154]. Faster movements have higher peak acceleration (i.e. impulse control), which is associated with lower endpoint precision unless the potential motor error is corrected during the deceleration phase (i.e. efficient limb-target control). Indeed, high spatial variability in limb trajectory at peak acceleration is progressively reduced as the movement proceeds when visual feedback is available [94]. Restricting visual feedback during the movement affects limb-target control such that errors in motor planning, including motor error due to high peak acceleration, are not amended, resulting in a greater endpoint error. Hence, slowing the movement to achieve good endpoint spatial precision involves two distinct changes in control: (i) adjusting impulse control by lowering peak acceleration, thereby reducing the potential motor error and/or (ii) modifying limb-target control by extending the deceleration phase to correct trajectory errors in the latter part of the movement.

Adults with amblyopia and strabismus had significantly lower peak acceleration, indicating that they adjusted their impulse control [155]. On the other hand, there was no significant difference in the duration of the deceleration phase between the amblyopia and control groups. One interpretation of these findings is that normal binocular vision facilitates the development of optimal integration of impulse and limb-target control: higher impulse requires good vision for efficient feedback control. When vision is less reliable, the feedback control loop is less efficient. Thus, lowering the speed of the movement by reducing peak acceleration decreases the potential motor error and reduces the need for online feedback control. An interesting question to consider is why patients lowered peak acceleration rather than extended the deceleration phase? Both these strategies would be expected to reduce endpoint error; however, it is possible that their costs may be different. For example, correcting errors in the terminal approach phase could be more time-consuming or require more attentional resources.

Finally, it is important to consider whether the aetiology of amblyopia impacts upper limb movement control. Strabismus is associated with poorer binocular outcomes compared to anisometropia [69], which could lead to lower motor proficiency. However, research in this area is still inconclusive. One study with 8-year-old children reported significantly larger deficits on a clinical test of fine motor skill in those with strabismus [146]. In contrast, a study that examined prehension kinematics found a larger deficit in children with anisometropic amblyopia [149]. Studies with adults have not revealed any significant effects due to amblyopia aetiology [152,155]. Conflicting results may be due to relatively small cohorts tested and the heterogeneity of patients' clinical features.

7. Implications for assessment and rehabilitation

Broadly speaking, binocular viewing is associated with a higher proficiency of fine motor skills. However, the extent to which binocularity facilitates motor performance is both age and task dependent. Given the developmental changes in impulse and limb-target control, it would be expected that normal binocular vision plays an important role in optimizing impulse control in younger children (i.e. under 8 years old). If vergence provides important input for calibrating impulse control, one might expect larger motor deficits in young children with strabismus and strabismic amblyopia [146,156,157]. Typical maturation involves a shift towards limb-target control, followed by an integration of impulse and limb-target control. Efficient use of visual feedback is the hallmark of limb-target control, and binocular stereopsis may be crucial to optimize this mode of control. Because limb-target control matures later, it might be expected that reduced or absent sensory fusion would be associated with deficits in reach-to-grasp coordination in older children and adolescents [149,150]. Crucially, current clinical assessments of fine motor skills do not provide insight into which aspects of control are affected; therefore, it is important to develop novel tools that would allow better understanding of impairments in the sensorimotor control mechanisms.

Selection of tasks used for assessment is also critical. For example, maturation of impulse control could be assessed using tasks that emphasize speed and accuracy such as the classical Fitts' task [158]. In contrast, using a prehension task may provide more insight into the maturation of limb-target control [143]. Deficits in binocular visual function could be associated with a prolonged maturation trajectory and may involve reweighting of sensory inputs to achieve adequate performance. Future studies need to establish if this is the case. In the light of the recent studies showing impairments in performance of more complex behaviours such as filling out scantron cards [10] and a trail making test [65], it may be important to assess limb kinematics during tasks that require higher-level executive processing. This area is ripe for investigation. In all, such knowledge is crucial for developing individualized rehabilitation approaches for children with impaired vision.

Undoubtedly, vision is a key sensory input for the development and control of motor behaviours. Research supports that normal binocular vision provides distinct advantages related to movement planning and execution; however, these advantages vary across development and tasks used for assessment. While binocular visual function emerges in the first few months after birth, maturation continues into adolescence [44,50]. Significant changes in visuomotor control have been documented in typically developing school-aged children where adult-like proficiency is associated with a shift towards an integrated impulse and limb-target control [14]. Examining the role of fusional vergence and stereoacuity in the performance of visually guided movements reveals that they have a distinct influence on the control of the reach and grasp components. In line with these findings, a temporary eye occlusion in adults disrupts the grasping component of prehension to a greater extent compared to the reach, suggesting that stereopsis during binocular viewing facilitates limb-target control as the hand approaches the object and during grasp execution. Amblyopia disrupts the normal development of binocular vision, which has been associated with lower motor proficiency on some tasks—highlighting the importance of binocularity for optimal motor performance. Importantly, individuals with amblyopia develop compensatory strategies that allow sufficient function in many everyday tasks. Nonetheless some deficits remain, specifically for tasks that require speed, accuracy and coordination, which may impact athletic and/or academic performance [10], self-esteem [159] and career choices [160]. Given the paucity of kinematic research in children and adolescents with amblyopia across different tasks, more research is needed to provide better insight into the development of sensorimotor control. Such knowledge could be used to guide the development of targeted rehabilitation regimens to improve visuomotor control and function.

Data accessibility

This manuscript does not contain any primary research data.

Authors' contributions

E.N.-S.: conceptualization, funding acquisition, investigation, writing—original draft, writing—review and editing; L.C.: conceptualization, writing—review and editing; A.W.: conceptualization, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We have no competing interests.

Funding

This research was supported by National Sciences and Engineering Council of Canada (grant no. ENS: NSERC RGPIN-2016-04356)