Effects of Anisometropic Amblyopia on Visuomotor Behavior, III: Temporal Eye-Hand Coordination during Reaching

Abstract

Purpose.: To examine the effects of anisometropic amblyopia on the temporal pattern of eye-hand coordination during visually-guided reaching.

Methods.: Eighteen patients with anisometropic amblyopia and 18 control subjects were recruited. Participants executed reach-to-touch movements toward visual targets under three viewing conditions: binocular, monocular amblyopic eye, and monocular fellow eye viewing. Temporal coordination between eye and hand movements was examined during reach planning (interval between the initiation of saccade and reaching) and reach execution (interval between the initiation of saccade and reach peak velocity). The frequency and dynamics of secondary saccades were also examined.

Results.: Patients with severe amblyopia spent a longer time planning the reaching response after fixating the target in comparison with control subjects and patients with mild amblyopia (P = 0.029). In comparison with control subjects, all patients extended the acceleration phase of the reach after target fixation (P = 0.018). Secondary (reach-related) saccades were initiated during the acceleration phase of the reach and patients executed these saccades with greater frequency than control subjects (P < 0.0001). The amplitude and peak velocity of reach-related saccades were higher when patients viewed with the amblyopic eye in comparison with the other viewing conditions (P < 0.001).

Conclusions.: This is the first study to show that patients with anisometropic amblyopia modified the temporal dynamics of eye-hand coordination during visually-guided reaching. They extended the planning and execution intervals after target fixation and increased the frequency of secondary, reach-related saccades. These may represent visuomotor strategies to compensate for the spatiotemporal visual deficits to achieve good reaching accuracy and precision.

Eye-hand coordination involves a complex interaction between the visual, ocular, and manual motor systems. The accuracy and precision of fine motor skills, such as reaching, grasping, and manipulating, depend on the ability to extract information from the environment and to coordinate the appropriate ocular and manual motor responses. ^1,2 Seemingly simple actions consist of several submovements that are integrated optimally in space and time into seamless goal-directed behavior. ³ For example, to pick up a cup, vision is used to detect and localize the cup. Saccadic eye movements are then executed to bring the image of the cup onto the fovea. The reach movement is initiated while maintaining fixation on the cup. ⁴ During the reaching movement, visual feedback is used to monitor the reach trajectory and to update the motor command when errors are detected. ^{5 –8}  

Spatiotemporal eye-hand coordination has been studied extensively in visually-normal people (for review see Bekkering and Sailer, 2002 ⁴ ). Saccades typically precede hand movement by 50 to 100 ms during variety of manual tasks. ^{9 –12} Directing the eyes to the target before initiation of hand movement allows the central nervous system (CNS) to obtain a high-resolution image of the target before the reach is initiated, which can facilitate programming of the reaching movement. In addition, when the eyes fixate on the target early during the reach trajectory (i.e., before the hand reaches peak velocity), visual feedback can be used to update the initial motor plan. Visual information can also be used during the deceleration phase in the latter part of the movement to fine tune the hand trajectory and to improve the performance (accuracy and precision) of the reach. ¹³ Indeed, Prablanc and colleagues reported that reaching performance improved substantially when the hand movement was initiated at least 40 ms after the eyes fixated on the target. ¹² Taken together, these findings suggest that the temporal delay between eye and hand movement initiation is not primarily a result of the smaller inertia of the eyeball than the arm, rather this delay serves to facilitate the planning and execution of the reaching movement. 

Amblyopia is a visual impairment of one eye caused by inadequate stimulation during early childhood that cannot be corrected by optical means. ¹⁴ Despite extensive evidence of spatiotemporal visual deficits in amblyopia, ^{15 –19} it is surprising that very few studies have examined how impaired spatiotemporal vision in amblyopia affects motor functions such as visually-guided eye and arm movements. Although the effects of amblyopia on eye movements ^{20 –22} and some aspects of manual motor control ^{23 –25} have been studied separately, the coordination between the visual, ocular, and manual motor systems has not been investigated previously in amblyopia. 

We have previously reported that patients with anisometropic amblyopia had delayed saccade initiation when viewing with the amblyopic eye in comparison with binocular or fellow eye viewing and in comparison with control subjects. ²¹ We also found that the manual reaction time and end-point performance (accuracy and precision) of the reach-to-touch movements in patients were comparable with control subjects in all viewing conditions. However, patients exhibited reduced peak acceleration and a prolonged acceleration phase during reaching. ²³ We postulated that patients with amblyopia optimized their reaching performance by adopting different kinematic strategies during planning (feedforward stage) and execution (feedback stage) of visually-guided reaching movements to compensate for their degraded spatiotemporal vision. In our previous studies, we examined the dynamics of the saccadic eye movements ²¹ and the kinematics of reaching movements ²³ separately. Because tight temporal coupling of eye and hand movements is important for optimal reaching performance, in this article, we aimed to investigate further whether patients with amblyopia adopt a different eye-hand temporal strategy to optimize their performance. We examined the effects of anisometropic amblyopia on the temporal pattern of eye-hand coordination during the planning stage (defined as the interval from target onset to the initiation of the reaching movement) and execution stage (defined as the interval from reach initiation to the end of movement) of visually-guided reach-to-touch movements (referred to as reaching movements from here onward). Specifically, we investigated whether patients spent longer time planning the reaching movement and whether they extended the acceleration phase of the reach after the eyes fixated the target. 

Materials and Methods

Participants

Eighteen patients with anisometropic amblyopia were recruited (6 males, age 26.4 ± 10 years). The clinical details of all patients are shown in Table 1. All participants underwent a complete orthoptic assessment by an unmasked certified orthoptist with 17 years clinical experience. The assessment included visual acuity testing using the Snellen chart (recorded as the last row in which a participant can correctly read all letters), measurement of eye alignment using the prism cover test, measurement of refractive errors, and stereoacuity testing using the Titmus test. The fusion ability of patients who lacked stereopsis (negative Titmus test) was tested using the Worth four dot test and the Bagolini test. Anisometropic amblyopia was defined as amblyopia in the presence of a difference in refractive error between the two eyes of ≥1 diopter (D) of spherical or cylindrical power. ^{25 –28} All patients had visual acuity between 20/30 and 5/400 in the amblyopic eye, 20/20 or better in the fellow eye, and an interocular acuity difference ≥2 lines. Six patients were orthophoric and 12 had monofixation syndrome associated with their anisometropic amblyopia ²⁹ ; that is, they had a microtropia ≤8 prism diopters (as a result of a foveal scotoma arising from the anisometropia; it is not the cause of the amblyopia), inability to bifixate, and presence of fusional vergence. Twelve patients had mild amblyopia (amblyopic eye acuity 20/60 or better) and 6 had severe amblyopia (amblyopic eye acuity 20/100 or worse). Eighteen visually normal participants (7 males, age 30.7 ± 11 years) served as control subjects. They had normal or corrected-to-normal visual acuity of 20/20 or better in each eye, and stereoacuity ≤40 seconds of arc. Exclusion criteria were any ocular cause for reduced visual acuity, prior intraocular surgery, or any neurologic disease. The study was approved by the Research Ethics Board at The Hospital for Sick Children and all protocols adhered to the guidelines of the Declaration of Helsinki. Informed consent was obtained from each participant. The saccade data from 12 patients ²¹ and the reaching data of 13 patients ²³ have been reported separately in two previous reports. These two sets of data, however, have not been analyzed together previously to investigate the temporal relationship between saccades and reaching movements, which was the purpose of the present study. 

Table 1.

 

Clinical Characteristics of Patients with Anisometropic Amblyopia

Patient	Age (y)	Visual Acuity (Snellen Chart)		Refractive Error		Stereoacuity (sec arc)	Fusion	Alignment
		RE	LE	RE	LE
1	33	20/15	20/30	−0.75	+2.00	140	NA	Orthophoria
2	29	20/50	20/15	+2.50 + 0.75 × 50	+0.25	3000	NA	Monofixation RE
3	17	20/20	20/40	pl +0.25 × 94	−1.00 + 1.00 × 92	80	NA	Orthophoria
4	14	20/50	20/15	+3.25 + 1.25 × 90	+2.00	50	NA	Orthophoria
5	18	20/15	20/40	Plano	+2.00 + 0.25 × 130	60	NA	Orthophoria
6	20	20/15	20/50	Plano	+1.50	120	NA	Orthophoria
7	35	20/15	20/60	−4.25	−0.75	3000	NA	Monofixation LE
8	36	20/15	20/400	−5.25	−12.00	Negative	Suppress	Monofixation LE
9	25	20/40	20/15	+1.00 + 0.25 × 22	Plano	400	NA	Monofixation RE
10	28	20/20	20/400	+4.00	+6.00 + 1.75 × 90	Negative	Fused*	Monofixation LE
11	21	20/30	20/15	+1.50	Plano	3000	NA	Monofixation RE
12	20	5/400	20/20	−2.00	−3.00 + 0.75 × 15	Negative	Fused†	Monofixation RE
13	19	20/20	20/40	−3.50 + 1.50 × 90	−3.50 + 2.50 × 102	200	NA	Monofixation LE
14	36	20/15	20/40	−1.50	+1.50 + 1.00 × 15	200	NA	Monofixation LE
15	56	20/100	20/20	+4.00	−2.25	Negative	Fused*	Monofixation RE
16	16	20/200	20/15	none	−2.75	Negative	Fused*	Monofixation RE
17	25	20/50	20/20	−1.50 + 1.50 × 80	−3.00 + 2.50 × 80	120	NA	Orthophoria
18	26	20/100	20/15	+2.00	Plano	3000	NA	Monofixation RE

 LE, left eye; NA, not tested (Titmus stereo was positive); RE, right eye.

*  Fused using Worth 4-dot test.

†  Fused using Bagolini test.

Apparatus

Eye movements were recorded at 200 Hz using an infrared video-based binocular pupil/iris tracking system (Chronos Vision, Berlin, Germany). The system has a maximum resolution of 6 minutes of arc. Linearity is <0.5% for both horizontal and vertical eye movements, over a range of ±20°. Before each experiment, a horizontal and vertical calibration was performed for each eye using fixation targets at five locations: 0° and ±10° horizontally and vertically. 

Reach-to-touch movements of the upper right limb were recorded (Optotrak Certus 3020 system; Northern Digital, Waterloo, Canada), an infrared illumination-based motion capture system. This system is noninvasive and allows for precise 3-D motion tracking of the limb (spatial accuracy 0.1 mm, resolution 0.01 mm, sampling frequency 200 Hz). The coordinate system was defined as follows: x-axis, horizontal plane; y-axis, vertical plane; z-axis, median plane. The system was calibrated before starting the experiment by using a four-marker digitizing probe to define the coordinate frame for the reaching movement. Two infrared markers (4 mm diameter) were affixed to the index fingertip and wrist joint of the participant's right (dominant) hand. A 15-mm diameter force-sensitive resistor (FSR; Tekscan, Boston, MA), was placed on the table at the participant's midline 28 cm from the computer screen and 17 cm from the participant. The FSR was used to trigger the initiation of each trial and to control when the visual target was switched off during a trial. 

Experimental Conditions and Procedure

The visual stimulus was a white circle (visual angle 0.25°) presented on black background generated by a custom-written program (in MATLAB, version 7.6.0; The MathWorks, Natick, MA) and presented on a 20-inch computer screen (Diamond Pro 2070SB; resolution 1600 × 1200 at 85 Hz; NEC-Mitsubishi, Itasca, IL) located 42 cm from the subject using a visual stimulus generator (ViSaGe; Cambridge Research Systems, Kent, UK). The distance from the starting position of the index finger to the computer screen was 43 cm in 3-D space. Testing was conducted in a dimly lit room. The target was presented at four eccentricities: ±5°or ±10°, all along the horizontal axis at eye level and in random order. 

Participants were seated at a table with their heads stabilized with a chin rest. There were three viewing conditions: binocular (BE), monocular amblyopic eye (AE), and monocular fellow eye (FE) viewing, and data were collected in blocks. For control participants, viewing was binocular, monocular left eye, and monocular right eye. Participants wore a black patch during monocular viewing. The order of viewing conditions was randomized among the participants. At the start of each trial, the right hand was placed on the table and the index finger was placed on the FSR. Participants fixated on a cross presented at midline. After a variable delay (1.5 to 3 seconds), the target was presented in the horizontal plane and participants were instructed to look and touch the target located on the computer screen 28 cm (from the starting position of the hand) in front of the subject as fast and as accurately as possible. On 50% of the trials, the target remained visible throughout the trial (target ON condition). On the remaining 50% of the trials, the target was switched off at the onset of hand movement, i.e., as soon as the finger was lifted off the FSR (target OFF condition). On these trials, participants were instructed to touch the location where they had seen the target. The target ON and OFF conditions were randomly interleaved. 

Participants completed 10 trials in each combination of the experimental conditions for a total of 240 trials. The intertrial interval varied among trials and it was at least 5 seconds. Practice trials were completed before starting the experiment to familiarize the subjects with the experimental procedure. We did not perform a separate test to ascertain specifically how well each patient could see the target. However, all patients, including those with severe amblyopia, executed spatially and temporally appropriate eye and hand responses during practice trials, indicating that they were able to detect the target, although patients with severe amblyopia might not have seen the target clearly. 

Data Analysis

Eye position data were filtered using a second-order, dual-pass Butterworth filter with a cutoff frequency of 50 Hz. Hand position data were also filtered using a second-order dual-pass Butterworth filter with a cutoff frequency of 7.5 Hz. Eye and hand velocity were obtained using a two-point differentiation method. A custom-written program (in Matlab) was used to identify saccades using a velocity threshold of 20° per second, while the initiation of hand movement was defined as when the velocity of the finger y-coordinate (i.e., vertical axis) exceeded 30 mm per second. The end of the reaching movement was defined as when the finger reached the computer screen and the velocity of the finger z-coordinate fell and stayed below 30 mm per second. All trials were inspected visually to ensure that the saccades and reaching movements were identified correctly by the program. 

Temporal coordination between eye and hand movements was examined in two stages of the reaching movement: the planning stage (i.e., from target onset to reach initiation) and the execution stage (i.e., from reach onset to the end of reach movement). Typical trials illustrating eye-hand coordination when reaching to a 10° target during binocular viewing for a control subject and two patients (one with mild and one with severe amblyopia) are shown in Figure 1. Eye-hand coordination during the planning stage of the reaching response was assessed by calculating the time interval after the eyes fixated on the target (i.e., at the end of saccade) and the initiation of the reaching response (from here on, we will refer to this interval as the saccade-to-reach planning interval, as depicted by the red arrow in Fig. 1). The saccade-to-reach planning interval was calculated on a trial-by-trial basis by subtracting saccade reaction time (latency) and saccade duration from the reach reaction time (see Fig. 1). It reflects the time that was available for planning of the reaching response after the saccade had been completed and the eyes were in the vicinity of the target. 

Figure 1.

 

Representative data showing the temporal relationship between eye and hand movements on a typical trial during binocular viewing of a 10° target for (a) a control subject, (b) a patient with mild amblyopia, and (c) a patient with severe amblyopia. The eye tracings represent the right eye of the control subject and the fellow eye of patients. The dotted gray vertical line depicts the time when the primary saccade was completed, whereas the solid gray vertical line depicts the time when the hand reached peak velocity. Saccade-to-reach planning interval (indicated by the red arrow) reflects the duration of time that was available for planning of the reaching response after the primary saccade had been completed and the eyes were in the vicinity of the target. Saccade-to-reach-peak velocity interval (indicated by the blue arrow) reflects the duration of time after the eyes fixated the target during the early part of reach execution. Patients were more likely to have extended planning and execution intervals after fixating the target in comparison with the control subjects.

The planning stage of the reaching response was also assessed by examining the frequency of trials when the hand movement was initiated before the eye movement. The accuracy and precision were compared between trials when reaching was initiated before or after the saccade. 

Eye-hand coordination during the execution stage of the reaching response was assessed by calculating the time interval between the end of the saccade and the hand reaching peak velocity (PV; from here on, we will refer to this interval as the saccade-to-reach PV interval, as depicted by the blue arrow in Fig. 1). Saccade-to-reach PV interval was calculated on a trial-by-trial basis by subtracting saccade reaction time and saccade duration from the time the hand reached PV (from target onset; see Fig. 1). It reflects the duration of time after the eyes fixated on the target during the early part of reach execution. Visual information acquired during this interval, which includes the acceleration phase of the reach, can be used to make compensatory adjustments to the reach trajectory in the later part of the movement. ³⁰  

Eye-hand coordination during the execution stage of the reaching response was also assessed by examining the frequency of secondary saccades. In this article, secondary saccades which occurred during the reach and which occurred >250 ms after the primary saccades were defined as reach-related saccades. We reasoned that these saccades were reach-related and were not secondary “corrective” saccades after the primary saccades under- or overshot because secondary “corrective” saccades typically occur with a latency of 100 to 250 ms. ^{31 –33} The amplitude, peak velocity, and latency (with respect to the initiation of the reaching response) of the reach-related saccades were calculated. Secondary corrective saccades after primary saccades have been described in our previous report. ²¹  

Statistical Analysis

All continuous dependent variables (saccade-to-reach planning interval, saccade-to-reach PV interval, latency, amplitude, and peak velocity of reach-related saccades) were submitted to a repeated-measures mixed ANOVA with group (three levels: control subjects, patients with mild amblyopia, patients with severe amblyopia) as a between-subjects factor and viewing condition (three levels: binocular, monocular fellow eye, monocular amblyopic eye; for control subjects, binocular, monocular left eye, monocular right eye) as a within-subjects factor. 

The frequency of reach-related saccades was compared between patients and control subjects using Pearson's χ² statistic. The effect of viewing condition was then examined within each group using Pearson's χ² statistic. 

All statistical analyses were performed using a statistical software package (SAS 9.2, Cary, NC). Descriptive statistics were reported as the mean and corresponding SD. Any main effects and interactions were analyzed further using Tukey-Kramer post hoc tests to adjust for multiple comparisons. The significance level was set at P < 0.05. Preliminary analysis showed that visual feedback of target had no significant effect on any outcome measures, therefore, data with or without visual feedback were collapsed for analysis and reporting. 

Results

Figure 1 shows representative eye and hand velocity tracing from a control subject, a patient with mild amblyopia, and a patient with severe amblyopia during binocular viewing of a 10° target. The control subject showed a stereotypical pattern of eye-hand coordination where the saccade was initiated ∼100 ms before reaching (Fig. 1a). In contrast, both patients showed longer saccade-to-reach planning intervals and longer saccade-to-reach PV intervals, which were more evident in the patient with severe amblyopia (Fig. 1c) than the one with mild amblyopia (Fig. 1b). 

Eye-Hand Temporal Coordination during the Planning Stage of Reaching

There was a significant interaction between group and viewing condition (F _(4,66) = 2.88; P = 0.029) for mean saccade-to-reach planning interval. As shown in Figure 2a, control subjects initiated reaching approximately 110 ± 65 ms after the eyes fixated on the target, regardless of viewing condition. Post hoc tests showed no significant difference for saccade-to-reach planning intervals between control subjects and patients with mild amblyopia (binocular, 120 ± 84 ms; fellow eye, 137 ± 84 ms; amblyopic eye, 102 ± 91 ms). In contrast, patients with severe amblyopia had longer saccade-to-reach planning intervals during all three viewing conditions (binocular, 197 ± 43 ms; fellow eye, 181 ± 51 ms; amblyopic eye, 152 ± 70 ms) in comparison with control subjects and patients with mild amblyopia. 

Figure 2.

 

(a) The saccade-to-reach planning interval across the three viewing conditions in patients with mild and severe amblyopia, and in normal participants. The saccade-to-reach planning interval is related to the planning stage of the reaching response. It reflects the time that was available for planning of the reaching response after the saccade had been completed and the eyes were in the vicinity of the target. Patients with severe amblyopia had a significantly longer saccade-to-reach planning interval in comparison with patients with mild amblyopia, and with control subjects (P = 0.02). Error bars indicate ±1 SE. (b) Cumulative frequency distributions for saccade-to-reach planning interval illustrating the temporal relation between the eye and hand movements during the planning stage of the reaching movement. A negative saccade-to-reach planning interval (left tail of the distribution) indicates that the reach was initiated before the saccade. Each curve represents data from one viewing condition for control subjects, patients with mild amblyopia, and patients with severe amblyopia.

Figure 2b shows the cumulative frequency distributions of saccade-to-reach planning intervals for control subjects and patients with mild and severe amblyopia for all viewing conditions. The saccade-to-reach planning interval was positive when the eyes fixated on the target prior to reach initiation and negative when the eyes fixated on the target after reach initiation. As illustrated in Figure 2b, saccades were initiated prior to reaching in the majority of trials in control subjects (98.8%) and patients (96.7%; χ_{(df 1)} = 34.07; P < 0.0001). For control subjects, there were no significant differences among viewing conditions (binocular, 98.5%; monocular left eye, 98.7%; monocular right eye, 99.1%; χ_{(df 2)} = 1.45; P = 0.482). In contrast, a significant difference among viewing conditions was found for patients with mild amblyopia (χ_{(df 2)} = 56.79; P < 0.0001); they initiated saccades before reaching with reduced frequency when viewing with the amblyopic eye (93.4%), in comparison with binocular (97.3%) and fellow eye viewing (98.9%). Similarly, patients with severe amblyopia initiated saccades before reaching with reduced frequency when viewing with the amblyopic eye (97.5%), compared with binocular (99.9%) and fellow eye viewing (99.8%; χ_{(df 2)} = 15.31; P < 0.001). 

Because patients initiated saccades before reaching on a reduced number of trials when viewing with the amblyopic eye, that is, they initiated reaching before saccades more often than usual, we examined whether this reverse eye-hand temporal coupling affected end-point accuracy or precision of reaching. Regardless of severity level, performance in patients was comparable whether saccades were initiated before or after the reaching movement. For patients with mild amblyopia, there was no difference in performance when saccades were initiated before (accuracy 0.29 ± 4.29 mm; precision 3.73 ± 1.93 mm) or after (accuracy 0.28 ± 4.89 mm; precision 3.65 ± 2.88 mm) the reach. Similarly, for patients with severe amblyopia, there was again no difference in performance when saccades were initiated before (accuracy −1.59 ± 11.09 mm; precision 4.97 ± 5.68 mm) or after the reach (accuracy 3.03 ± 4.82 mm; precision 5.61 ± 4.08 mm). 

Eye-Hand Temporal Coordination during the Execution Stage of Reaching

Saccade-to-Reach PV Interval.

Cumulative frequency distributions of the saccade-to-reach PV intervals for control subjects and patients with mild and severe amblyopia were plotted for all viewing conditions in Figure 3a. The frequency distributions for both groups of patients were shifted toward longer intervals in all viewing conditions in comparison with control subjects, suggesting that patients extended the acceleration phase of the reach once the eyes fixated on the target during the execution stage. There was a significant interaction between group and viewing condition (F _(4,66) = 3.21; P = 0.018). As shown in Figure 3b, control subjects had comparable saccade-to-reach PV intervals in all viewing conditions (binocular, 306 ± 114 ms; right eye, 314 ± 105 ms; left eye, 300 ± 101 ms). In comparison with control subjects and regardless of severity level, patients had significantly longer saccade-to-reach PV intervals when viewing with the amblyopic eye (mild amblyopia, 359 ± 127 ms; severe amblyopia, 363 ± 61 ms) and when viewing with the fellow eye (mild amblyopia, 386 ± 111 ms; severe amblyopia, 388 ± 51 ms). Post hoc tests showed that during binocular viewing patients with severe amblyopia (409 ± 44 ms) had significantly longer saccade-to-reach PV in comparison with patients with mild amblyopia (344 ± 104 ms) and control subjects (306 ± 114 ms). 

Figure 3.

 

(a) Cumulative frequency distributions for saccade-to-reach-peak velocity interval by viewing eye condition illustrating the temporal relation between the eye and hand movements during the execution stage of the reaching movement. Each curve represents data from one viewing condition for control subjects, patients with mild amblyopia, and patients with severe amblyopia. The frequency distributions were shifted toward longer intervals for patients with mild and severe amblyopia compared with control subjects. (b) Mean saccade-to-reach peak velocity intervals were significantly longer for patients with mild and severe amblyopia in comparison with control subjects in all viewing conditions. During binocular viewing patients with severe amblyopia also had significantly longer saccade-to-reach PV intervals in comparison with patients with mild amblyopia (P = 0.018). Error bars indicate ±1 SE.

Frequency of Reach-Related Saccades.

The frequency of reach-related saccades was significantly greater in patients (mild amblyopia, 14.1%; severe amblyopia, 21.8%) than in control subjects (11.8%) (χ² _{(df 2)} = 85.14; P < 0.0001). As shown in Figure 4a, control subjects executed reach-related saccades with greater frequency when viewing monocularly (right eye, 13.2%; left eye, 13.6%) in comparison with binocular viewing (8.6%; χ² _{(df 2)} = 21.09; P < 0.0001). In contrast, patients with amblyopia executed reach-related saccades with comparable frequency across all three viewing conditions. In patients with mild amblyopia, the frequency was 14.6% during binocular viewing, 13.7% during fellow eye viewing, and 13.8% during amblyopic eye (χ² _{(df 2)} = 0.31; P = 0.854). Similarly, in patients with severe amblyopia, the frequency was 20.2% during binocular viewing, 23.7% during fellow eye viewing, and 21.6% during amblyopic eye viewing (χ² _{(df 2)} = 1.69; P = 0.423). 

Figure 4.

 

Mean metrics of reach-related saccades. (a) Patients initiated reach-related saccades more frequently in comparison with control subjects. (b) Patients with mild amblyopia had longer reach-related saccade latencies of in all viewing conditions in comparison with control subjects. For patients with severe amblyopia, the reach-related saccade latencies were longer during amblyopic eye viewing in comparison with binocular and fellow eye viewing (P = 0.003). (c) Patients with mild and severe amblyopia had higher reach-related saccade amplitudes during amblyopic eye viewing in comparison with other viewing conditions (P < 0.0001). (d) Patients with mild and severe amblyopia had higher reach-related saccade peak velocities during amblyopic eye viewing in comparison with other viewing conditions (P < 0.001).

Latency of Reach-Related Saccades.

There was a significant interaction between group and viewing condition for the latency of reach-related saccades (F _(4,62) = 4.62; P = 0.003; Fig. 4b). For control subjects, the latency was not different between viewing conditions (binocular, 146 ± 59 ms; right eye, 158 ± 52 ms; left eye, 155 ± 59 ms). For patients with mild amblyopia, the reach-related saccade latencies were longer in comparison with control subjects but not different between viewing conditions (binocular, 184 ± 74 ms; fellow eye, 201 ± 45 ms; amblyopic eye, 207 ± 71 ms). In contrast, patients with severe amblyopia had significantly longer reach-related saccade latencies when viewing with the amblyopic eye (252 ± 124 ms) in comparison with binocular viewing (113 ± 22 ms) and fellow eye viewing (129 ± 48 ms). 

Amplitude of Reach-Related Saccades.

The interaction between group and viewing condition was significant for amplitude (Fig. 4c) of reach-related saccades (F _(4,62) = 9.04; P < 0.0001). For control subjects, amplitude was not different between viewing conditions (binocular, 0.90° ± 0.28°; right eye, 0.98° ± 0.39°; left eye, 1.00° ± 0.41°). Patients with mild amblyopia had comparable amplitudes during binocular viewing (0.90° ± 0.28°), fellow eye viewing (0.85° ± 0.33°), and amblyopic eye viewing (1.09° ± 0.37°). Patients with severe amblyopia had higher reach-related saccade amplitudes during amblyopic eye viewing (1.49° ± 0.28°) in comparison with viewing with the fellow eye (0.68° ± 0.31°) or binocularly (0.79° ± 0.16°). 

Peak Velocity of Reach-Related Saccades.

The interaction between group and viewing condition was also significant for peak velocity (Fig. 4d) of reach-related saccades (F _(4,62) = 6.53; P < 0.001). For control subjects, peak velocity was not different between viewing conditions (binocular, 70° ± 19° per second; right eye, 73° ± 19° per second; left eye, 76° ± 28° per second). Patients with mild amblyopia had comparable peak velocities during binocular viewing (76° ± 25° per second) and during fellow eye viewing (73° ± 27° per second). Post hoc testing indicated that peak velocity was higher during amblyopic eye viewing (87° ± 21° per second) in comparison with the other viewing conditions and with control subjects. Patients with severe amblyopia also had higher peak velocities during amblyopic eye viewing (97° ± 24° per second). However, peak velocities were lower during fellow eye (58° ± 20° per second) and binocular (59° ± 15° per second) viewing. 

Discussion

This study examined the effects of impaired spatiotemporal vision on the temporal pattern of eye-hand coordination during the planning and execution stages of visually-guided reaching movements in patients with mild and severe anisometropic amblyopia. The major findings are: (1) patients with severe amblyopia spent a longer time planning the reach after the eyes fixated on the target before they initiated the hand movement; (2) all patients extended the acceleration phase of the reaching movement after the eyes fixated on the target; (3) patients executed reach-related saccades with greater frequency; and (4) the reach-related saccades in patients had longer latency, higher amplitude, and higher peak velocity during amblyopic eye viewing. 

Eye-Hand Coordination during the Planning Stage of Reaching

Directing the fovea of the eyes to the target before reach initiation allows the use of high-resolution visual feedback to plan the reaching response before the limb movement begins. Indeed, visually normal participants typically move their eyes to the target before initiating hand movement even during experiments when they are not specifically instructed to move their eyes. ^10,11 In addition, the central nervous system may also use extraretinal eye position signals (i.e., efference copy or proprioception) generated by the orienting saccades to update the early motor plan to improve reach performance. ^4,34  

Regardless of viewing condition, control subjects in our study initiated reaching approximately 110 ms after the eyes fixated on the target. In contrast, patients with severe amblyopia spent a significantly longer time planning the reaching movement after fixating on the target before the hand movement was initiated, regardless of viewing condition. Patients may have extended the planning interval to compensate for their poor acuity during amblyopic eye viewing to improve reaching performance. Interestingly, the extended planning interval was also evident during fellow eye and binocular viewing. This finding might be surprising at first glance because the fellow eye had acuity of at least 20/20, and because binocular acuity in 10 patients was comparable with fellow eye acuity (we did not collect binocular acuity in the other 8 patients because this was not part of our original protocol). However, despite normal acuity, deficits in the fellow eye have been well-documented in people with amblyopia, including second-order spatial loss, ^35,36 impaired motion processing, ^37,38 deficient contour integration, ^39,40 and impaired perception of images of natural scenes. ⁴¹ It has been hypothesized that these higher-order deficits exist because second-order neurons are binocular and they require normal binocular input during development. Thus, anomalous binocular vision during early development leads to higher-order cortical deficits, which can be detected during monocular viewing with either the amblyopic or fellow eye. This hypothesis is supported by anatomic and neurophysiologic studies which showed that early-onset monocular deprivation leads to a reduced proportion of functionally binocular neurons in primary visual cortex ^{42 –44} and extrastriate cortex. ⁴⁵ In addition, suppression of the fellow eye by the amblyopic eye has been documented in cats, ^46,47 monkeys, ⁴² and in humans. ⁴⁸ Furthermore, binocular suppression is also evident in monkeys ⁴² and humans, ⁴⁸ suggesting that while amblyopia disrupts predominantly the excitatory interactions between the two eyes, cortical inhibitory binocular connections are less susceptible to abnormal visual experience. Our results thus provide additional support to the growing body of evidence that abnormal visual processing ^{35 –38,41,49} and altered motor behavior ²³ are also present during fellow eye and binocular viewing in patients. 

Patients with mild amblyopia and control subjects had comparable reach planning intervals after fixating on the target. In agreement with previous studies, ^12,50,51 control subjects initiated saccades before reaching on > 98% of trials. In contrast, patients with mild or severe amblyopia initiated reaching before saccades on significantly more trials when viewing with the amblyopic eye. More importantly, despite this reversal in eye-hand coupling in patients, reaching accuracy and precision were comparable between trials when saccades were initiated prior to the reach and those when saccades were initiated after the reach. Three explanations are possible. One possibility for a lack of difference in reaching performance between these two types of trials may be due to the relatively small number of trials in which reaching was initiated before saccades. Another possibility is that our subjects did not have to extract any fine details from the visual target when they performed our relatively simple motor task. It remains to be seen whether patients would show altered eye-hand coupling in more difficult visuomotor tasks. A third, and more likely, possibility is that good spatial reaching performance was achieved due to the substantial difference between saccade and reaching duration. In this study, the mean saccade duration was approximately 40 ms, whereas the mean reaching duration was approximately 550 ms for control subjects and approximately 650 ms for patients. This substantial difference between saccade and reaching duration meant that the eyes were able to fixate on the target well in advance of the hand reaching the target. Thus, both patients and control subjects had enough time to update the target's location by using retinal and/or extraretinal feedback to adjust the hand approach trajectory and to modify the landing position of the hand. The ample time allowed them to achieve good reaching accuracy and precision, even in trials when the hand movement was initiated before the saccade. 

Eye-Hand Coordination during the Execution Stage of Reaching

Tight coupling between saccades and early kinematic markers of the reaching response has been reported by Helsen and colleagues. ¹⁰ They found a strong positive correlation between the time when the eyes fixated on the target and the time when the hand reached peak velocity. They suggested that this temporal sequence might facilitate optimal extraction of visual information for guiding movement. When the eyes fixate on the target at a time when the hand is in the early phase of movement, high-resolution visual feedback of the target can be acquired to update the reach plan and used to make compensatory adjustments to the hand reach trajectory in the final approach phase. Indeed, Helsen and colleagues ⁵⁰ showed that the tight eye-hand temporal coupling might be responsible for reducing the spatial variability of the hand trajectory in the interval between peak velocity and the end of the primary submovement (i.e., before the corrective movement began) by a factor of five. We have previously shown that patients extend the acceleration phase of the reaching movement ²³ ; in this study, we examined specifically whether the extended acceleration phase is related to the time of target fixation. 

Consistent with the results in our previous study with a smaller sample size, ²³ we found that the duration of acceleration phase after target fixation was extended in both patient groups and under all viewing conditions in comparison with control subjects. Despite a considerable difference in visual acuity between patients with mild (range 20/30 to 20/60) and those with severe amblyopia (acuity 20/100 or worse), we found that the duration of acceleration phase was extended comparably in all patients after fixating on the target during monocular fellow eye and amblyopic eye viewing. 

In contrast, the duration of the acceleration phase after target fixation was affected differentially by the severity of amblyopia during binocular viewing. Specifically, the acceleration interval after target fixation was shorter in patients with mild amblyopia (but still longer than in control subjects) compared with patients with severe amblyopia. One possible explanation is that binocular vision provides important information for both movement planning and online control, ^{52 –55} and that even residual binocularity may provide some advantage during the execution of reaching movements. In this study, all patients with mild amblyopia had residual stereopsis, whereas the majority of patients (five out of six) with severe amblyopia had no stereopsis. The reduced acceleration interval in patients with mild amblyopia might be due to their ability to use residual binocular information to program a more precise initial motor plan or to make online compensatory adjustments during the reaching movement. 

Reach-Related Saccades

Primary saccades to visual targets usually undershoot the target's location by approximately 10%. ³¹ Numerous studies have documented that primary saccades are usually followed by secondary corrective saccades with a latency of 100 to 250 ms. ^32,33,56 In the present study, we examined secondary saccades that occurred after the onset of the reaching movement (which we defined as reach-related saccades) and that were not temporally associated with primary saccades (i.e., latency >250 ms after the primary saccade). We reasoned that the function of secondary saccades initiated during reach execution is to facilitate the online control of the reaching response. To the best of our knowledge, reach-related saccades have not been reported previously even in visually-normal individuals. Our study is the first to show that visually-normal people made reach-related saccades more frequently when viewing monocularly compared with binocular viewing, indicating that binocular vision provides important input for programming and execution of reaching movements. ^57,58  

The frequency of reach-related saccades was greater in patients with amblyopia and did not differ among viewing conditions. Overall, patients with severe amblyopia initiated reach-related saccades most frequently, while patients with mild amblyopia had frequency of reach-related saccades comparable with control subjects when control subjects viewed monocularly. Although we found some differences between patients and controls for the latency of reach-related saccades, these saccades were initiated during the acceleration phase of the reaching movement on the majority of trials. Because visual information acquired early in the trajectory can be used to make compensatory adjustments later in the trajectory, the reach-related saccades that were executed relatively early during the reaching movement may play a functional role in facilitating reaching performance. The frequency pattern and the latency of reach-related saccades suggest that they might be an adaptive strategy and/or compensation that patients developed to maintain good reaching accuracy and precision in face of their spatiotemporal visual deficits. 

Although patients had a similar frequency of reach-related saccades across viewing conditions, the peak velocity of these reach-related saccades were higher during amblyopic eye viewing in comparison with the other viewing conditions and with control subjects. In addition, patients with severe amblyopia had higher amplitude and peak velocity of reach-related saccades during amblyopic eye viewing (but not during binocular or fellow eye viewing) in comparison with patients with mild amblyopia. These results suggest that reach-related saccades are most likely initiated based on a retinal error signal, which may be impaired due to visual positional uncertainty when patients viewed with the amblyopic eye. ^59,60 During amblyopic eye viewing, patients with severe acuity impairment might have less reliable retinal position error signals of the target/hand image such that the visual error must be larger to be detected and before a reach-related saccade is initiated. This, in turn, leads to higher amplitude and peak velocity of reach-related saccades when viewing with the amblyopic eye. 

In conclusion, we demonstrated that patients with anisometropic amblyopia modified the temporal dynamics of eye-hand coordination during visually-guided reaching and that the severity of amblyopia affected the strategies that patients used. Patients with severe amblyopia extended the planning stage before reach initiation after the target was acquired by the primary saccade. All patients extended the acceleration phase of the reaching movement after the eyes fixated on the target and executed more reach-related saccades during the acceleration phase of the reach, irrespective of their level of visual acuity. We propose that the extended planning and execution intervals after target fixation and the additional reach-related saccades are strategies that patients developed to compensate for their spatiotemporal visual deficits, allowing them to achieve good accuracy and precision during visually-guided reaching. 

Footnotes

 Supported by Grants MOP 106663 from the Canadian Institutes of Health Research (CIHR), Leaders Opportunity Fund from the Canadian Foundation for Innovation (CFI), and the Department of Ophthalmology and Vision Sciences and Research Training Centre at The Hospital for Sick Children.

Footnotes

 Disclosure: E. Niechwiej-Szwedo, None; H.C. Goltz, None; M. Chandrakumar, None; Z. Hirji, None; A.M.F. Wong, None