Effect of visual impairment on goal-directed aiming movements in children


  • Published online 11th August 2008

* Correspondence to first author at Bartiméus, P O Box 1003, 3700 BA Zeist, the Netherlands.
E-mail: A.Reimer@Bartimeus.nl


This study investigated potential differences in motor control between children with a visual impairment (diagnosed albinism; n=11, mean age 8y 4mo [SD 7mo]; seven males, four females,) and children with normal vision (n=11, mean age 8y 4mo [SD 7mo]; six males, five females). Mean near visual acuity in the albinism group was 0.19 (SD 0.07, Snellen: 20/104). Children performed two types of movements (discrete and cyclic) in two orientations (azimuthal and radial, i.e. along the viewing and lateral direction), and with two amplitudes (10 and 20cm). All movements were performed in two subsequent target conditions: with and without visual information on the target location. Overall, children with visual impairment displayed larger endpoint variability. Discrete movements and movements over large distances were less fluent in both groups, but especially in the children with visual impairment. Children with visual impairment seemed to have more difficulties with calibrating the sensory information. Specifically, they made larger errors along the lateral direction, when the target was not visible. Results suggest that children with visual impairment have specific differences in motor control compared with children with normal vision, which are not all directly related to their poorer vision.

Several studies have demonstrated the importance of visual guidance for achieving speed and accuracy in goal-directed movements. Although brief visual samples of the movement environment are sufficient for reasonably precise, closed-loop control,1 vision remains of significant importance for optimal accuracy, even after extended practice.2 Visual information for controlling goal-directed movements is used in different ways depending on the movement phase.3,4 In the initial (ballistic) phase of a rapid aiming movement, vision makes a significant contribution to determining the direction of the movement. One or more saccades direct the eyes to the target location. Because this usually precedes limb movement, the eyes focus on the target before the finger.5,6 In the final (homing-in) phase of the movement, vision is particularly important for movement accuracy. Visual feedback enables corrections in a later part of the trajectory so that the hand can reach the goal precisely. Therefore, fine eye–hand coordination requires setting up a temporary, task-specific synergy between the eyes and hand.7

Movement control in children aged 6 to 8 years becomes more dependent on visual information because the ability to use alternative perceptual strategies becomes limited.8 A straightforward hypothesis would be that visual impairment has a negative influence on children’s eye–hand coordination, as less visual information is available to guide the movement towards its target. Although differences in motor development between children with and without visual impairment have been observed,9–11 scant experimental research has been carried out to investigate potential differences in the kinematics of goal-directed movements between these groups under different movement conditions.

In the present study, children with visual impairment diagnosed with albinism, and children with normal vision performed repeated goal-directed aiming movements,12 first several times with and subsequently several times without visual information of the target location. As mentioned above, speed and accuracy of such movements depend on real-time information from the visual system. In addition, information coming from the proprioceptive system also plays an important role.13,14 Between the two phases of the task (i.e. with and without view of the target location), the influence of the visual and proprioceptive subsystems will be different. Accuracy is expected to be lower in the second phase and it is likely that the movements will be guided more on the basis of proprioception, even though there is feedback from the hand and visual memory of the experiment is set-up.15,16 However, in the first phase of the task the visual feedback might be used to calibrate the proprioceptive system.

In addition to the two target conditions, several different movement conditions were presented. By varying the orientation, mode, and amplitude of the movements it was possible to examine the interaction of visual and proprioceptive information over a wide range of naturally occurring movement conditions. The possible influences these variations might have on movement control in the two groups can be described in two ways.

First, children had to carry out the movements along the viewing direction (azimuthal) as well as the lateral direction (radial). The literature concerning asymmetries in goal-directed movements report that each hand is faster when moving in the ipsilateral hemispace,17 and with respect to accuracy, there are also reports of fewer errors towards the ipsilateral side of the body.18,19 The level of integration between visual and proprioceptive feedback also varies with movement orientation. Hand-position estimates rely more on proprioception along the lateral direction than along the viewing direction, and are also more accurate along the lateral direction.20 Based on these findings, the impact of visual impairment on movement speed and accuracy is expected to be different between these two movement conditions, as well as between the two vision conditions.

Second, children had to perform the (repeated) movements in a discrete and a cyclic mode, i.e. with a distinct start signal at each individual movement and in a continuous fashion respectively. Both kinds of movements are frequently performed in daily life, but are considered to rely on different control mechanisms.21 Discrete movements have a clear beginning and end and, therefore, are usually controlled more by feedback than cyclic movements. Visual guidance is the most important control input for these corrective movements. During the main part of the movement, as well as during the homing-in phase of a discrete movement, many corrective changes can be made based on visual feedback; however, this will lead to longer movement times.

Contrary to discrete movements, cyclic (or rhythmical) movements are more of a ballistic nature. These movements rely more on an open-loop control strategy in which less-corrective sub-movements are made and fewer changes in speed occur.21 It might be that children with a visual impairment, generally, rely more on open-loop control. However, it is unclear whether this leads to better performance. Moreover, using an open-loop control strategy, especially in the second phase of the task (without view of the target location) requires good calibration of the proprioceptive subsystem based on experience and fine-tuning in the first phase (with view of the target location). Little is known about such sensory calibration, and whether it is fully developed or accurate in children with visual impairment. The aim of this study was to investigate potential motor–control differences between children with a visual impairment and children with normal vision in repetitive aiming goal-directed movements.



A group of 11 children with a visual impairment diagnosed with albinism (seven males, four females), mean age 8 years 4 months (SD 7mo), and a group of 12 children with normal vision (seven males, five females), mean age of 8 years 6 months (SD 7mo) participated in this experiment. Albinism is a hereditary genetically determined disorder of the melanin synthesis within pigment cells that has widespread and variable effects on the eyes, visual system, and the skin, often accompanied by a misdirection of the optic nerve fibres.22 This study only included children who had no demonstrable neurological disorder, intellectual impairment, nor other pathology.

Nine children with a visual impairment attended regular primary education and two attended classes for special education at the Bartiméus, Zeist, the Netherlands. All children in this group were recruited from Bartiméus’ archives. Children with normal vision all attended regular primary education, and were contacted through the Ichthus and De Sluis School, Zeist. Permission for the study was obtained from the board of managing directors of the institute. After written invitation, informed consent was received from parents of the participants. The study was performed in accordance with the ethical standards of the Declaration of Helsinki (1964).

Mean visual acuity in children with visual impairment (measured in 10 children with the nystagmus) was 0.19 (SD 0.07) Snellen 20/104. In addition, they had varying degrees of nystagmus, which is typical for albinism.23 People with nystagmus reduce the negative effect of the condition by holding their heads at a tilted and/or rotated way (torticollis). Thus each child had a different head orientation to ensure that it was least disturbing.

Fine-motor skills were tested with the Manuvis test,24 which provides specific norms for poor-sighted children. All children who scored higher than the 15th centile were classified as having normal motor development. One child with normal vision who scored lower than the 15th centile was not included in the analyses, leaving 11 children in the comparison group. On average, the children with visual impairment scored 13% lower on the Manuvis test and performed the fine-motor tasks slower than the children with normal vision. Hand preference was indicated by the hand that children used to write.25 In each group there were two left-handers.

Material and procedure

Children had to move a small puppet over the surface of a digitizer (sample rate 206Hz; Wacom, Saitama, Japan; type Cintiq 18sx) which was positioned horizontally in front of the shoulder of their preferred hand (Fig. 1). The digitizer incorporated a high-luminance LCD monitor (SXGA 24-bit full colour), which was used to display the targets that served as the beginning and end-points for each movement. Targets were circles of 2.5cm in diameter.

Figure 1.

Top view of the experimental set-up for two orientation conditions: (a) azimuthal and (b) radial. In both conditions, movements were performed for target distances of 10cm and 20cm, and in a discrete and continuous fashion. In all conditions movements were performed 10 times with, followed by 10 times without, visual information of the target location.

Before the start of each trial, the child was asked to position the puppet in the starting circle on the digitizer, after which the experimenter started the experiment. A random period of about 0.5 to 1.5 seconds later, an acoustic signal was given and the target circle appeared on the digitizer. This was the indication for the child to move the puppet as fast and as accurately as possible towards the target location. Children performed the experiment with their preferred hand, and they could observe the puppet’s motion as well as that of their arm at all times. All children started with a practice session.

Several different movement and vision conditions were presented to each child in separate blocks. To determine whether amplitude has a differential effect, participants had to move the puppet over two distances: 10cm and 20cm between starting circle and target circle (Index of Difficulty is 3 and 4, respectively; with A the amplitude of the movement and W the target width12). Moreover, movements were performed in two orientations relative to the body: azimuthal and radial, i.e. along the viewing direction and along the lateral direction respectively. Furthermore, the mode of the movements was varied as discrete and cyclic. In the discrete mode, the procedure described above was repeated 10 times. In the cyclic mode, after the starting signal, children moved the puppet between the two targets for a period of 6 seconds. The order of presentation of the eight resulting movements-condition blocks (2 amplitudes × 2 orientations × 2 modes) was counterbalanced across participants.

Finally, within each of the movement conditions, visual information about the location of the target circle was varied (target condition). Movements were performed first with visual information of the target’s location, then as part of a series without visual information of the target’s location. In every block the without-target condition was always presented after the with-target condition.

Data analysis and dependent measures

Movements were recorded and analyzed using OASIS software.26 First, data from left-handed participants was switched between left and right to pool it with that of the right-handed participants. Second, the first movement of each condition was not used in the analyses, and the median value was used to filter out possible outliers in every condition. After this, a set of dependent variables was calculated for each child and each condition by averaging individual movements. Finally, each of these variables was entered in a repeated measures analysis of variance with three within participant factors (amplitude, orientation, and mode) and one between participants factor (vision group). Analyses of variance (ANOVAs) were performed separately for each of the two vision conditions. The assumptions underlying ANOVA were checked by spread-versus-level plots and residual plots, and by homogeneity of variance tests (Levene test). Significance was set at p<0.05 (two-tailed).

Four dependent variables were derived from the movement data: endpoint variability, reaction time, movement time, and peak-over-mean velocity. As a measure of movement accuracy, the endpoint variability was calculated. Endpoint variability was defined as the distance (centimeters) between the centre of the puppet and the centre of the target circle. Reaction time was defined as the temporal delay (seconds) between the acoustic signal and the start of the movement. Reaction time was only determined for the discrete conditions. Movement time was defined as the single-movement duration (seconds), i.e. the time it took the child to move the puppet once from the starting circle to the target circle. Measurement of movement time started at the first displacement of the puppet and ended when its speed dropped below 0.2cm/s in the target circle. Finally, as a measure of the ballistic nature of the movement, and to quantify the extent to which movements were produced by open-loop or closed-loop control, the ratio of peak-over-mean velocity was calculated.


The experimental design required participants to perform two types of movements: movements towards the body or midline (i.e. retraction or adduction) and movements away from the body or midline (i.e. protraction or abduction). For three of the dependent variables (movement time, reaction time, and peak-over-mean velocity), preliminary analysis revealed no difference between these two types of aiming movements, and data were pooled within each condition; however, differences were found for endpoint variability in both target conditions.

Tables I and II display group averages (pooled) of the four dependent variables as a function of the amplitude, orientation, and mode of the movements, in the conditions where visual information on the target’s location was present (with-target condition) or absent (without-target condition) respectively. There were no missing data.

Table I.   Dependent variables for vision groups as a function of amplitude, orientation, and mode in the with-target conditiona
 Visually-impaired group (n=11)Normal-vision group (n=11)
Amplitude (ID)OrientationModeAmplitude (ID)OrientationMode
10cm (3)20cm (4)AzimuthalRadialDiscreteCyclic10cm (3)20cm (4)AzimuthalRadialDiscreteCyclic
  1. aValues in table are means. bThis variable is only relevant to discrete mode.

  2. ID, Index of Difficulty; EV, endpoint variability in centimetres; MT, movement time in seconds; RT, reaction time in seconds; PoM, peak-over-mean velocity.

EV (cm)0.4780.5520.4950.5340.3930.6360.3960.4520.4190.4290.3280.520
MT (s)0.7550.9880.8840.8600.9780.7650.7440.9820.8440.8830.9090.817
RTb (s)0.5770.5030.4980.5820.5400.4170.4320.4110.4380.425
Table II.   Dependent variables for vision groups as a function of amplitude, orientation, and mode in the without-target conditiona
 Visually-impaired group (n=11)Normal-vision group (n=11)
Amplitude (ID)OrientationModeAmplitude (ID)OrientationMode
10cm (3)20cm (4)AzimuthalRadialDiscreteCyclic10cm (3)20cm (4)AzimuthalRadialDiscreteCyclic
  1. aValues in table are means. bThis variable is only relevant to discrete mode.

  2. ID, Index of Difficulty; EV, endpoint variability in centimetres; MT, movement time in seconds; RT, reaction time in seconds; PoM, peak over-mean velocity.

EV (cm)1.1691.3251.1271.3671.2671.2260.9590.9250.9700.9150.9930.892
MT (s)0.7491.0210.8930.8771.0440.7250.7631.0180.8600.9211.0100.770
RTb (s)0.5110.4860.4940.5020.4980.4250.4170.4220.04210.422

Endpoint variability

In the without-target condition (see Table II) a significant overall difference was found between the two vision groups with respect to endpoint variability (F[1,21]=10.71, p=0.004). In the with-target condition (see Table I) the overall difference in endpoint variability was not significant (F[1,21]=3.99, p=0.059), which might be due to insufficient power. Children with visual impairment were less accurate than children with normal vision: mean error sizes were 1.25cm versus 0.94cm in the without-target condition, and 0.51cm versus 0.42cm in the with-target condition respectively. An interaction effect of orientation and vision group in the without-target condition (F[1,21]=5.20, p=0.033) revealed that movement accuracy in the visually-impaired group was poorer in the radial orientation than in the azimuthal orientation (1.37cm vs 1.13cm), while it was about the same in the normal-vision group (0.92cm vs 0.97cm).

When the target was visible, differences in endpoint variability were not equal in size for different amplitudes (F[1,21]=10.13, p=0.004) and for different modes (F[1,21]=56.92, p<0.001). Wider movements (20cm/ID, Index of Difficulty=4) and cyclic movements resulted in more error in both vision groups. These differences between movement conditions were not present when the target was no longer visible, and performance was generally poorer.

As mentioned above, there were differences for endpoint variability with respect to the direction of the movements (i.e. between movements to proximal targets and distal targets) in both target conditions. In Table III, endpoint variability in both target conditions is separated for proximal and distal movements. Differences in accuracy are quite large in the without-target condition and oppositely directed to those in the with-target condition. Without visual feedback of the target, movements away from the body are significantly more accurate than movements towards the body (F[1,21]=10.66, p=0.004). With visual feedback of the target, the (opposite) effect proved not to be significant (F[1,21]=4.15, p=0.055).

Table III.   Endpoint variability in both target conditions separated for movements to proximal (left/down) and distal (right/up) targetsa
   Visually-impaired group (n=11)Normal-vision group (n=11)
Proximal movementDistal movementProximal movementDistal movement
  1. aValues in table are means. bThe two orientation conditions perfectly separate movements into orthogonal directions: In the azimuthal orientation, movements were in the down-up direction. In the radial orientation, movements were in the left-right direction. ID, Index of Difficulty.

With targetAmplitude (ID)10cm (3)0.4480.5070.3920.400
20cm (4)0.5110.5920.4220.481
Without targetAmplitude (ID)10cm (3)1.3740.9631.0720.846
20cm (4)1.5191.1301.0390.811

Movement time

Overall, movement time was not significantly different between children with visual impairment and children with normal vision. In both target conditions movement time varied as a function of amplitude, in such a way that movements over a larger distance took more time to perform (F[1,21]=194.48, p<0.001, in the with-target condition; and F[1,21]=120.73, p<0.001, in the without-target condition). A similar difference was present for mode, such that discrete movements took more time than cyclic movements (F[1,21]=13.84, p=0.001, in the with-target condition; and F[1,21]=35.85, p<0.001, in the without-target condition).

Reaction time

No effect of the different movement and target conditions or between the vision groups was found for reaction time, although a trend can be detected from the results. Averaged over all conditions, reaction times for the children with visual impairment were 0.54 seconds in the with-target condition and 0.50 seconds in the without-target condition. For the children with normal vision this was 0.42 seconds in both target conditions.

Peak-over-mean velocity

Peak-over-mean velocity varied as a function of the amplitude and the mode of the movement in both target conditions. Peak-over-mean velocity was higher for movements over a larger distance (F[1,21]=14.97, p=0.001, in the with-target condition; and F[1,21]=9.66, p=0.005, in the without-target condition). With respect to mode, peak-over-mean velocity was higher for discrete movements than for cyclic movements (F[1,21]=55.67, p<0.001, in the with-target condition; and F[1,21]=88.24, p<0.001, in the without-target condition).

A significant interaction between mode and vision group in the with-target condition (F[1,21]=5.93, p=0.024) revealed that for the discrete movements, children with a visual impairment had a higher peak-over-mean velocity than children with normal vision (2.13 vs 1.97). For the cyclic movements this difference did not appear to be present (1.84 vs 1.82). No further group effect was found.


The purpose of this study was to investigate potential motor-control differences between children with a visual impairment and children with normal vision in repetitive aiming movements. It is the first study concerning children with visual impairment in this respect, especially those diagnosed with albinism who have varying degrees of nystagmus.23 Results demonstrated that children with visual impairment were less proficient in performing the movements under the various conditions presented. The study compared children’s performance under various movement conditions (amplitude, orientation, and mode) and target conditions (i.e. visibility of the target location). Arguably, these conditions give rise to different types of motor control.21

Accuracy of movement

The hypothesis that children with visual impairment are less accurate in the execution of aiming movements was confirmed in this experiment. In particular, and quite surprisingly, movement accuracy suffered more in this group in the phase with no visual information on the location of the target. A possible explanation for this might be found in calibration processes between sensory subsystems. This argument of calibration has two aspects: developmental and task specific.

It is known from studies of early motor development that the vestibular and proprioceptive subsystems are calibrated using visual feedback.11 This calibration is necessary for postural control, but also for fine-tuning of prehension movements. Prechtl et al.11 reported that in infants with profound visual impairment a delay is present in the development of proprioception caused by the lack of visual integration, and that it is still unclear if this delay is fully recovered later in life. It is reasonable to suspect that proprioception is also less adept in children with visual impairment, albeit less severe. Less optimally calibrated sensory subsystems might have led to less accurate movement control when the target was not visible.

Calibration plays a crucial role in adaptive action control, under changing task constraints or when sensory information about these constraints changes.27,28 Even in the absence of a visible target, children still had some global visual information of the task setting (e.g. edges of the digitizer) and of the movements of their own arm. This information might have been used to continue to guide the movements visually (or at least partially), and if so it is likely that the children with poorer vision benefited less from this information. As a result, under these less than optimal feedback conditions, their ‘drift’ away from the optimal movement trajectory would be larger, as was observed.

For both groups, when the target was no longer visible, accuracy deteriorated more for movements towards the body or midline. This suggests that the calibration process is different for proximally and distally directed movements. Perhaps related to this, for the children with normal vision, endpoint variability was about the same for the two orientations. However, the children with a visual impairment made larger errors in directing the puppet when moving along the lateral direction. All the children with visual impairment who participated in this study had albinism with nystagmus to some degree. The natural coping strategy for this is to hold the head at a certain angle, called ocular torticollis, to reduce the effects of this nystagmus. The result of this is a limited gaze, which had a negative influence on accuracy in the radial orientation. Another factor could be that radial movements rely more on visual information than on proprioceptive information.13 Both factors may be responsible for the larger impact of poor vision in these tasks.

Movement onset

Visually-guided arm movements, such as reaching or pointing, are accompanied by saccadic eye movements that typically begin prior to movement initiation of the arm.6 The group of children with visual impairment all had nystagmus. The type of head position, to ensure that the nystagmus was the least disturbing, varied for each child. It is possible that this influenced movement onset time because focusing on the target is more difficult, explaining the trend in reaction times between the two groups.

Fluency of movement

In the cyclic condition, vision is less crucial for guiding the movements. It is known that cyclic movements are performed under open-loop control, which relies less on visual control.19 Results showed that peak-over-mean velocity, as a measure of the fluency of the movements, was lower during cyclic tasks. This suggests that fewer corrective movements were made in this condition, which makes the overall movement more fluent. When accuracy demands are higher, more changes in velocity peaks are expected.6,29 Results of the current study concur with these previous findings and suggest an explanation for the differences in fluency between the two groups in the discrete tasks. Having less visual information seems to influence the fluency of discrete movements more than the cyclic movements.

The main limitations of the present study are concerning methodological issues and utilization. Methodologically, a more thorough research set-up is needed, for instance, using motion-capture and eye-tracking devices and/or electromyogram, to scrutinize the control systems involved in the various movement conditions and the calibration processes in the two target conditions. With respect to utilization, the research group is quite small, and although aetiological variations were limited, the group is still heterogenic in many respects. This makes an easy translation of the results to a medical or therapeutic setting quite difficult, because many more factors are involved, the influence of which we know very little about.


This study shows that there are several differences between children with visual impairment and children with normal vision with respect to the control of aiming movements. Results emphasize that more research is needed regarding the influence of degraded visual information on movement in natural viewing and movement conditions. Follow-up studies might focus on the influence of nystagmus and eye movements. There is also scope for study into the differential role of developmental and task-related changes in the calibration and integration of vision and proprioception. A careful analysis could provide valuable insight into the development of motor control in children with visual impairment.