Differences in Smooth Pursuit and Associated Saccades in Children With Two Common Types of Strabismus

Abstract

Purpose

The main characteristics of smooth pursuit are well-described, but few studies have analyzed performance in strabismic patients, especially regarding differences between types of strabismus. In this study, we analyzed smooth pursuit eye movements of children with intermittent exotropia (IXT) or accommodative esotropia (AET), compared with age-matched controls.

Methods

We included 58 children, aged between 4 and 13 years old (AET n = 24, IXT n = 18, and controls n = 16). After ophthalmological evaluation, eye movements were recorded using a Tobii Pro X3 eye tracker (120 Hz). Smooth pursuit was elicited by a single image moving sinusoidally at the horizontal axis of the screen, with a velocity of 11°/s. Smooth pursuit gain, latency, and frequency of saccades during pursuit, were analyzed.

Results

Smooth pursuit showed different characteristics in the three groups. Gain was significantly lower in IXT, compared with the control group (0.66 vs 0.69; P < 0.01). In contrast, AET patients showed a significantly greater number of anticipatory saccades during pursuit (1.84 [1.46/2.23] vs 0.76 [0.57/0.94]; P = 0.028) compared with controls. No differences were found in latency between the three groups, although we observed faster responses when a saccade occurred (259.2 vs 282.81 ms). A laterality analysis showed better performance for rightward than leftward movements and small differences between nasal and temporal pursuit directions.

Conclusions

Smooth pursuit is affected by strabismus, but performance differs between clinical subtypes. We hypothesized that the smooth pursuit system is primarily affected in IXT, whereas the saccade system might be more affected in AET.

Our eyes are always moving, gathering information about our environment, before the brain processes this information. Humans are able to perform different kinds of eye movements, depending on the characteristics of the task at hand. When the object of interest is moving, like when we are trying to catch a ball, smooth pursuit eye movements are generated. They are conjugated slow eye movements of the eyeballs that allow us to maintain the image of a moving target near the fovea.¹

Characteristics of smooth pursuit eye movements in adult populations are well-known.² For example, Rashbass³ showed that smooth pursuit has a latency of approximately 150 ms and a variable gain, depending on target velocity. Interestingly, smooth pursuit is not completely developed until adolescence.^4–6 The pursuit system is immature at birth⁷ and progressively improves over the first years of life, although there is some disagreement between studies regarding when adult performance is observed. For example, when estimating pursuit accuracy, Sinno et al.⁸ found a gain of 0.63 at 5 to 8 years old and of 0.86 at 15 to 17 years old, whereas Vinuela-Navarro et al.⁹ showed that at 7- to 8-year-olds’ pursuit performance was similar to adults. A review made by Karatekin¹⁰ concluded that lower gain in children was only found when using motion at high frequencies, but was similar to gain in adults at low frequencies. Beside this controversy on the age of complete maturation of the pursuit system, studies suggest that during childhood the characteristics of smooth pursuit gradually improve, leading to an increased gain and a decrease in latency and the number of saccades.^11,12

Oculomotor performance can be affected by multiple disorders. For example, in cases of schizophrenia¹³ or autism,¹⁴ smooth pursuit might show different parameter values than seen for age-matched controls. The same phenomenon occurs in cases of strabismus. Strabismus is a disorder in which the visual axes are not properly aligned, and this misalignment causes both aesthetic and visual impairments. Types of strabismus are usually classified depending on the direction of the deviation (e.g., exotropia or esotropia), and subdivided by the onset (congenital, early, etc.), whether the deviation in both eyes is comitant or incomitant, intermittent or constant, or if it is associated to problems in refraction or vergence.¹⁵ Two of the most prevalent types of strabismus are accommodative esotropia (AET) and intermittent exotropia (IXT).^16–18 AET is a constant esodeviation caused by accommodation in the presence of hypermetropia. It appears in early childhood, usually between 2 and 5 years of age.^19,20 IXT is an exodeviation that is not constant, elicited by stress or fatigue. It appears at 2 or 3 years of age, and the frequency of deviation may increase with age.²¹

Although strabismus might interfere with oculomotor performance, there are only a few studies that have analyzed smooth pursuit performance in strabismus.^22–30 Unfortunately, their results do not provide a clear picture owing to crucial differences between the studies: the use of different techniques to measure smooth pursuit (e.g., search coil or infrared eye tracker), different tasks (e.g., triangle, trapezoidal or sinusoidal trajectory, or Initiation with a ramp or step-ramp paradigm), and different procedures regarding data analysis. A consequence of this is the absence of a gold standard. A review of these problems was published by Smyrnis.³¹ Another major problem when comparing studies is the lack of inclusion criteria used for the strabismus participants: different subtypes of strabismus are not always distinguished, and subjects with amblyopia are sometimes included in the same group. This factor makes it difficult to reach conclusions regarding oculomotor performance for each subtype of strabismus and the possible differences between them.

In this context, the aim of this study was to characterize the parameters of smooth pursuit eye movements in children in the two mentioned subtypes of strabismus, AET and IXT. We hypothesize that their principal characteristics could be different, not only compared with controls, but also between both subtypes, owing to a different development of the sensorimotor system.

Methods

Participants

Fifty-eight participants (24 with AET, 18 with IXT, and 16 control participants) were included. Children's ages ranged between 4 and 13 years. All subjects were recruited in the ophthalmology clinic at Hospital Ramón y Cajal. The study was approved by the Ethics Committee of Hospital Ramón y Cajal and the Universidad Autónoma de Madrid. Our research adhered to the tenets of the Declaration of Helsinki. Informed consent was given by the children's parents (in Spain, only children >14 years old can give consent.).

A complete ophthalmological examination was performed before eye movements were recorded. This included visual acuity (VA) and refraction, simultaneous and alternate prism and cover test, anterior segment biomicroscopy, and fundoscopy. Inclusion criteria for AET participants were esotropia developed after 4 months of age, with at least +2.50 D of hyperopia and at least 20/30 of VA in the worst eye. Deviation might be fully corrected with spectacles (fully accommodative) or partially corrected (partially accommodative, whose hyperopia could be less than +2.50). Subjects with any other eye disease, amblyopia, another type of esotropia, or previous strabismus surgery were excluded. Inclusion criteria for IXT participants were basic type of IXT developed after 2 years old and 20/30 or better VA in the worst eye. Cases with myopia greater than −5.00 D, greater deviation at near than at distance, any other eye disease, amblyopia, another type of exotropia, or previous strabismus surgery were excluded. Age-matched controls were also studied. For inclusion in the control group, participants had to show no ocular deviation and spherical equivalent between −0.50 and +0.75. Participants with any eye disease were excluded. Stereopsis and contrast sensitivity were evaluated the day of the eye movement recording with Randot Test (Stereo Optical Company, Chicago, IL, USA) and Mars Letters Contrast Sensitivity Test (Mars Perceptrix, Chappaqua, NY, USA), respectively.

Stimuli, Task, and Instruments

A computer screen of 17.5” (1600 × 900 px) was settled in front of the participant, 55 cm away. A chinrest was used to reduce head movements, and participants performed the task wearing their refractive correction. The room had a dim light. Data were collected with a 120 Hz eyetracker, Tobii Pro X3-120 (Tobii, Stockholm, Sweden). We used a five-point calibration and recorded all tasks using binocular vision. The eye tracker obtained raw data separately for each eye.

Psychtoolbox^32,33 functions implemented using MATLAB (MATLAB and Statistics Toolbox, The MathWorks, Inc., Natick, MA, USA) were used to create the tasks. During the recording session, participants performed different tasks that required eye movements (saccades, steady fixation, and pursuit eye movements). To elicit pursuit eye movements, we presented a static stimulus that started to move. The target used was made up of colored small circles that together looked like a cartoon figure. The figure had a total size of 2 cm (2.8°) and was presented on a black background. The target moved following a ramp paradigm: it appeared at the center of the screen, jumped to one side and, immediately after that, it started moving horizontally in the opposite direction. The total movement trajectory corresponded with one and one-quarter sinusoidal cycles. The length of the horizontal trajectory, side to side, was 34.35°. The mean movement velocity was 11°/s. During each trial, sounds were also presented to keep attention in the stimulus. Each block consisted of 2 trials, each starting at opposite sides. Each participant performed a total of 5 blocks (10 trials). To keep interest and reduce fatigue, a visual search game was presented on the screen between consecutive blocks of trials.

PreProcessing of Eye Movement Data

Raw eye movement data were filtered and analyzed using custom-written scripts in MATLAB. After finding a positive correlation between the two eyes, the mean of x axis data was obtained, simulating the cyclopean eye. These position data were used for the rest of the analysis. First, outliers were eliminated using a velocity threshold of 750°/s. Saccades, which are often performed during pursuit, were identified by velocity, with a threshold of 30°/s and a minimum of two consecutive points above that value. When the amplitude of a saccade was less than 1°, it was classified as a microsaccade. Classification of the remaining saccades depended on eye movement direction and eye position with respect to the target.^6,34 If eye movement was in the same direction as target, we defined three different types of saccades: catch-up, overshoot, and anticipatory saccades. A catch-up saccade was defined as a saccade in which eye position before the saccade was behind the target, and the final eye position was at the location on the target. Overshoot saccades also started with eye position behind the target, but eye position after the saccade landed ahead of the target. When the initial eye position before the saccade coincided with the target location but the final eye position was ahead of the target, it was classified as an anticipatory saccade. If a saccade was performed in the opposite direction with respect to target motion, it was classified as a back-up saccade. A small group of saccades did not fit in any of the mentioned groups and were classified as other saccades. Information about the total number of saccades in each direction of movement (rightward/ leftward) was also collected.

To determine the characteristics of smooth pursuit, all those points classified as saccades, microsaccades, and outliers were eliminated and were replaced using lineal interpolation. Fixations and smooth pursuit periods were separated using a velocity and dispersion threshold identification algorithm.³⁵ A minimum duration of 65 ms was also required to define fixations. Smooth pursuit data were analyzed after removing those periods in which motion direction changed. For that, a smoothing velocity filter was applied and a 50- to 100-ms temporal window was used to calculate mean velocity for each section.³⁶ Gain was defined as the ratio between eye velocity and target velocity. Figure 1 shows data from a trial and the segmentation of the different eye movements.

Figure 1.

Trial of an AET participant. The x axis is time in seconds. The left y axis represents horizontal position, in pixels. The right y axis represents eye velocity in pixels/second. The red sinusoidal line shows target trajectory. The blue line represents right eye horizontal position over time, including the initial seconds before the start of smooth pursuit. The orange line represents the corresponding eye velocity. High-velocity peaks match jumps in eye position and correspond with saccades. The colored line on the top of the figure represents eye movement classification for this trial: green corresponds with fixations, magenta corresponds with smooth pursuit periods, and black corresponds with saccades.

Smooth pursuit latency was defined as the interval between the start of target movement and the beginning of eye motion. Smooth pursuit might start with or without a saccade. We considered that it started with a saccade when a saccade was detected up to 125 ms after target started moving sinusoidally.^37,38 When smooth pursuit started directly (without a saccade), onset was calculated as in other studies.^39,40 First, a velocity baseline was estimated as the mean and SD of the interval between 58 ms before and after target movement begins. Then, the point at which velocity exceeded 3 SD of the baseline was determined, and a second lineal regression was created using the mean velocity of the next 125 ms. The intersection point of this lineal regression and the baseline was considered the start of pursuit eye movement.

Statistical Analysis

Statistical analyses were carried out using SPSS (IBM SPSS Statistics for Windows, Version 22.0, IBM Corp., Armonk, NY, USA). An intereye correlation test was used to compare data of both eyes. The Saphiro–Wilk test was chosen for normality measures. Given that the distribution of data in most variables was not compatible with normality, we used nonparametric tests to compare performance between the three groups of participants. Specifically, the Kruskal–Wallis test was used for multiple comparisons, and, when significant, a post hoc Dunn test with Bonferroni correction was performed for repeated pairwise comparisons. A Mann–Whitney U test was used for comparisons between two independent groups, and the Wilcoxon test was used for two related samples. To analyze repetition effects over trials, the Friedman test was used for nonparametric repeated measures data, and repeated measures ANOVA for parametric data. The χ² test was used to compare the proportion of trials initiated with or without a saccade, and a Spearman correlation test was used to analyze the effects of age or VA. A P value of less than 0.05 was considered statistically significant.

Results

General Performance

Our main goal was to determine whether there were differences in smooth pursuit characteristics between the three groups: AET, IXT, and controls. Data from three AET and one IXT participants were eliminated owing to bad recordings. All participants had at least 20/25 of VA in the worst eye, except an exotropia participant, who had 20/30 in both eyes, so amblyopia was not observed. Ophthalmological examination showed that most esotropias were fully accommodative, and they did not have a high AC/A ratio. Spectacles were used in all cases requiring refractive correction.

To give an overview of the differences between groups, Table 1 summarizes the main variables in each of the groups. No differences in age were found between the three groups (P = 0.351).

Table 1.

Characteristics of Participants

	Control	AET	IXT
Age	7.87 (6.96/8.78)	8.61 (7.69/9.65)	7.64 (6.40/8.89)
RE VA	0.98 (0.95/1.01)	0.93 (0.89/0.97)	0.95 (0.89-1)
LE VA	1.00 (0.97/1.05)	0.94 (0.9/0.97)	0.92 (0.87/0.99)
RE spherical equivalent	−0.07 (−0.34/0.21)	+4.02 (+3.27/+4.78)	0,00 (−0.67/+0.67)
LE spherical equivalent	−0.12 (−0.35/+0.10)	+4.06 (+3.31/+4.81)	+0.13 (−0.37/+0.63)
Stereoacuity (arcsec)	32.12 (26.19/38.18)	195.33 (122.78/267.88)	77.29 (46.84/107.74)
Distance deviation		1.43 (−0,63/3.49)	−22.06 (−28.29/−15.82)
Near deviation		4.18 (0.63/8.98)	−15.59 (−22.56/−8.61)

Values are described as mean (95% CI). Distance and near deviation are measured with correction.

A correlation test was performed on the eye position raw data, comparing both eyes. Results reflected a very high correlation (>0.8) in most of the trials, which indicates that, during the experiment, eyes were not deviated. Only two trials of two different participants showed low correlation, so they were removed from other analysis. A linear mixed effects model including all relevant variables revealed no significant effects owing to age, so in the next sections we just report results for the different clinical groups. Despite the absence of a statistically significant effect, there was a trend in all parameters showing improvement with age.

Table 2 summarizes the characteristics of smooth pursuit eye movements for the three groups. We found significant differences between the groups in some of the measures, which are presented in the next sections.

Table 2.

Characteristics of Smooth Pursuit Eye Movement

	Control	Esotropia	Exotropia	P Value
Gain	0.69 (0.67/0.71)	0.7 (0.68/0.71)	0.66 (0.64/0.67) †	0*
Latency	286.99 (254.63/319.35)	290.05 (260.87/319.23)	271.39 (238.12/304.66)	0.513
Saccades
Catchup	0.27 (0.16/0.37)	0.28 (0.18/0.37)	0.29 (0.2/0.38)	0.646
Overshoot	1.06 (0.83/1.29)	0.98 (0.78/1.18)	1.23 (0.95/1.52)	0.775
Anticipatory	0.76 (0.57/0.94)	1.84 (1.46/2.23)†	1.04 (0.77/1.31)	0.016*
Backup	5.38 (4.39/6.36)	6.75 (5.68/7)	5.39 (4.61/6.18)	0.175

For saccades we report the number of saccades of each type in a cycle. Values are described as mean (95% CI). P values are calculated with the Kruskal–Wallis test.

^* P < 0.05.

^† P < 0.05 in pair-wise comparisons with controls.

Gain

Figure 2 shows gain values obtained in each trial and for each participant in the control, AET and IXT groups. Kruskal-Wallis test showed significant differences between groups (P < 0.001). A post hoc analysis showed that gain was significantly lower in IXT (0.66 [0.64/0.67]) compared with the other two groups (control, 0.66 [0.64/0.67], Z = −2.934, P = 0.01; AET, 0.66 [0.64/0.67], Z = −3.909. P < 0.001).

Figure 2.

Gain values in the three groups. Horizontal color lines indicate the median values for each group.

We also determined whether fatigue influenced gain. For that, we compared for each group independently the values obtained over the five trials recorded. A Friedman test showed no difference in the control group (P = 0.267), whereas both strabismic groups had a significant reduction of gain over time (AET, P = 0.04; IXT, P = 0.023).

Latency

As explained elsewhere in this article, pursuit onset in a ramp procedure can occur with a saccade or as pure pursuit. To calculate latency, these two onset types were analyzed separately. Data from two controls, one AET, and one IXT were excluded owing to problems with data timing. In the control group, 69.6% of valid trials started with a saccade, 74.4% in AET, and 67.8% in IXT. A χ² test showed these small differences were not significant (χ² = 1.601; P = 0.449). A Spearman test also showed that these proportions did not correlate with age (r = 0.107; P = 0.458) or VA (r = 0.043; P = 0.768). Considering all groups together, pursuit onset with a saccade showed significantly lower latencies than onset without it (saccade latency, 259.2 ms [246.99/271.41 ms]; only pursuit latency, 282.81 ms [265.06/300.56 ms], Z = −2.467; P = 0.014).

For those trials in which pursuit started directly, a Kruskal–Wallis test found no significant differences between the three groups (control, 286.99 ms [254.63/319.35 ms]; AET, 290.05 [260.87/319.23 ms]; IXT, 271.39 ms [238.12/304.66 ms]; P = 0.513), as shown in Figure 3. Repeated measures ANOVA showed no effect owing to task repetition in any group (control, P = 0.831; AET, P = 0.492; IXT, P = 0.824). Direction of movement could not be compared, because only 12 valid trials had a rightward pursuit onset vs. 99 trials with leftward onset. Using data of each eye separately, a Mann–Whitney U test showed differences in latency depending on whether the direction of motion for each eye was in the nasal/temporal direction only in AET participants (P = 0.05).

Figure 3.

Latency (ms) of smooth pursuit eye movement (when pursuit started without a saccade).

Those trials that started with a saccade were also analyzed, even though the following characteristics do not refer to pursuit part of the movement. Differences between groups in mean saccade latency were not significant (controls, 273.78 ms [250.35/397.21 ms]; AET, 254.76 ms [235.68/273. 85 ms]; IXT, 253.03 ms [230.93/275.13 ms]; P = 0.282). Because in most cases pursuit started with a saccade, it was possible to analyze how direction of motion affected performance. Both strabismic groups showed longer saccade latencies in leftward than rightward direction (Mann–Whitney U test; AET, Z = −2.066, P = 0.039; IXT, Z = −1.988, P = 0.047).

Frequency of Each Type of Saccade

We also analyzed the frequency (number of saccades per cycle) of each type of saccade in each of the three groups. This data is summarized in Figure 4. Catch-up saccades (top left part of Fig. 4) occurred very rarely: the mean frequency was 0.27 (0.16/0.37) in the control group, 0.28 (0.18/0.37) in AET, and 0.29 (0.2/0.38) in IXT. These values were not significantly different (Kruskal–Wallis test; P = 0.646). We only found an effect of repetition in AET group (Friedman test; P = 0.032).

Figure 4.

Boxplots, with median and 25% to 75% IQR, show number of saccades according to type for each group (y axis is scaled for each type): Catch-up (top left), overshoot (top right), anticipatory (bottom left), and back-up saccades (bottom right). X represents outliers.

Overshoot saccades were more common, with a mean frequency of 1.06 (0.83/1.29) in the control, 0.98 (0.78/1.18) in the AET, and 1.23 (0.95/1.52) in the IXT group (top right part in Fig. 4). No differences between the three groups were found (Kruskal–Wallis test; P = 0.775). Repetition did not show an effect in any of the groups (control, P = 0.336; AET, P = 0.852; IXT, P = 0.192).

Regarding anticipatory saccades (bottom left part of Fig. 4), we did find differences between the three groups (Kruskal–Wallis test; P = 0.016). Specifically, a post hoc test showed that the frequency of anticipatory saccades was higher for AET (1.84 [1.46/2.23]) than for controls (0.76 [0.57/0.94]; Z = −2.602; P = 0.028). The differences between IXT (1.04 [0.77/1.31]) and the other groups did not attain significance (control IXT, Z = −0.382, P = 1; AET-IXT, Z = −2.222, P = 0.079). We only found an effect linked to repetition in the control group (Friedman test; P = 0.042).

Back-up saccades (bottom right part of Fig. 4) occurred very often: the mean frequency in the control group was 5.38 (4.39/6.36), in the AET group was 6.75 (5.68/7), and in the IXT group was 5.39 (4.61/6.18). Still, there were significant differences between the three groups (Kruskal–Wallis test; P = 0.175). Interestingly, the frequency of backup saccades, measured with Friedman test, did increase significantly owing to repetition in the control (P = 0.008) and IXT (P = 0.044) groups.

We also performed a laterality analysis, combining all types of saccades. We found a significantly higher number of saccades in the leftward than the rightward direction in each group. These were separately analyzed with Wilcoxon test (control, rightward = 4.68 [3.94/5.41], leftward = 6.37 [5.51/7.23]; P < 0.001; AET, rightward = 5.83 [5.11/6.56], leftward = 8.14 [7.13/9.15]; P < 0.001; IXT, rightward = 4.69 [4.07/5.3], leftward = 6.34 [5.56/7.13]; P < 0.001). When we compared the three groups grouping data by motion direction, this difference in saccade frequency was still present [rightward motion direction: control = 4.85 [4.03/5.67], AET = 6.04 [5.28/6.79], IXT = 4.7 [4.05/5.35]; P = 0.046; leftward motion direction: control = 6.41 [5.45/7.37], AET = 8.43 [7.39/9.47], IXT = 6.28 [5.45/7.1]; P = 0.023). However, we found no significant difference in the frequency of saccades depending on whether eye motion was in the nasal/temporal direction (Wilcoxon test; control, P = 0.277; AET, P = 0.449; IXT, P = 0.504).

Discussion

The main goal of this study was to characterize pursuit eye movements in the two most common types of strabismus, determining whether impairments in performance, if present, were similar in both clinical groups, or differed (AET vs. IXT). Our results showed that different types of strabismus affect the oculomotor system in different ways. We found that pursuit gain was significantly lower in IXT participants than in the AET and control groups, which had equivalent values. In contrast, the AET group showed significantly more anticipatory saccades than the IXT and control groups. Finally, saccade frequency and latency showed an asymmetric effect, with a greater frequency of saccades when target motion was in the leftward direction. We now discuss each of these results in greater detail.

Gain

Gain is the most frequently used parameter to characterize smooth pursuit movement. In normal adults, it usually has a value around 1. However, our results showed that control children and preadolescents had lower gain values. This finding agrees with what has been reported in other studies^4,5,6,8 and suggests that the smooth pursuit system is not totally developed until at least late adolescence. There are still some discrepancies that need an explanation, because other studies have reported gain values similar to those of adults at lower ages,^9,41 although this could result from the use of different methods when determining gain.

Given the difficulties in comparing results from different studies, the present investigation, in which three different groups of participants were tested using the same task and procedures, can provide relevant information. Our results indicated that smooth pursuit gain was affected by strabismus. Specifically, the IXT group had lower gain than controls, while the AET group showed gain values similar to those of controls. This is in contrast with the results of Hepokur et al.,²⁵ who found lower gain in all three subgroups of AET than in controls. It also differs from the results reported by Lions et al.,²² who found similar gain values in a strabismic group than in controls, with participants in a similar range of ages. However, in that study the strabismic group was made of a mix of 10 patients, 4 with exotropia (3 IXT and one constant), and 6 with esotropia (4 early onset esotropia and 2 acquired). This heterogeneity could be the reason that no differences were found in gain. The authors also reported higher mean gain values, reaching 0.8 at 9 years old, but this could be due to differences in the calculation of gain.

Latency

We did not find differences in latency between the three groups, with values that differed by less than 20 ms (271–290 ms). Still, latencies deviated from the normal range of pursuit latencies found for adults,^2,42,43 which is approximately 100 to 180 ms. Our hypothesis is that this difference reveals the slow development of pursuit, which continues improving until late adolescence,^5,6 although, as we explained in the Introduction, different methods of measuring and analyzing data might result in differences when determining age effects.³¹

In contrast, the saccadic system reaches adult values at 10 to 12 years of age,^44,45 with latencies of approximately 200 ms.² Our results agree with this assumption, since we found saccadic latencies between 250 and 270 ms, in a range of ages between 4 and 13 years old. Significant differences were found in all groups depending on the type of pursuit onset, showing faster latencies when a saccade occurs. Importantly, the proportion of trials starting with each onset type was similar in all groups.

Saccades During Smooth Pursuit

Our results indicate that catch-up, overshoot, and back-up saccades are not affected by strabismus. Lions et al.²² and Ciuffreda et al.²³ also reached a similar conclusion regarding catch-up saccades. We found a combination of reduced gain and normal number of catch-up saccades in the IXT group. According to Hutton and Kennard,¹³ this could be the result of an impairment of the smooth pursuit system combined with an augmented position error tolerance. Regarding anticipatory saccades, we found that AET participants made significantly more of these saccades than the other two groups. According to van Gelder et al.,³⁴ catch-up saccades are corrective saccades, whereas anticipatory saccades are made by mistake when trying to improve task performance. However, other authors⁴⁶ have hypothesized that the increment in anticipatory saccades could be related to problems with attentional focus. Level of attention to our task is reflected in the variability across trials.¹² Our results revealed no significant effects of fatigue, owing to repetition, but more variability in anticipatory saccades in AET participants (also between trials). Therefore, attention could be a relevant factor, and possible impairments may be affecting the saccade system. Although saccades and pursuit eye movements are controlled by different neural networks, both systems share some paths, so an altered development that affects both types of movements is possible in strabismus.

We also found that the frequency of saccades during pursuit was influenced by the direction of smooth pursuit. A rightward direction of motion elicited fewer saccades than leftward. Quite the opposite result was found by Lions et al.,²² who argued that pursuit following reading direction could elicit more saccades. Unfortunately, in their study only catch-up saccades were analyzed. Moreover, we did not find any differences depending on whether eye motion was in the nasal/temporal direction, as has been previously reported.^24,29 Given that our data were collected binocularly, differences between both eyes depending on direction might have been minimized, following Hering's law.⁴⁷ In any case, the rightward bias owing to reading direction in our culture might explain why we obtained better performance in the control of pursuit (fewer saccades and lower latencies) for that direction.

Differences Between Strabismus Types

Results show that IXT and AET pursuit parameters are differently affected. Gain was the only parameter that was affected in IXT. In contrast, the AET group had normal gain and latency, but showed a significantly greater number of anticipatory saccades and more frequent (although not significantly) backup saccades. These saccades could result from a previous anticipatory saccade, producing a square wave jerk. Taken together, these results suggest that the smooth pursuit system is impaired in IXT, whereas in AET an impairment of the saccade system is more likely, with effects that also extend to the control of smooth pursuit. This means that both types of strabismus present specific characteristics, which questions the idea of a connection between these two subtypes of strabismus suggested by Brodsky and Jung.⁴⁸

The control of smooth pursuit is carried through a complex network involving many different areas, including the visual cortex, areas middle temporal, middle superior temporal, or frontal eye field.^1,49 It is a well-defined system, different from the one involved in saccadic generation. Still, both systems share some areas. One possible explanation of the different parameter values that we have obtained in our study for the different clinical groups is that impairments linked to strabismus affect the two systems differently. Considering all those areas involved in the smooth pursuit system, we can suggest some hypothesis on how some of them might misfunction in cases of a strabismic disorder. The superior colicullus (SC) is mostly involved in target selection.⁵⁰ Patients with AET could show an impairment in SC functioning, which would produce inaccurate pursuit. If this were the case, a misestimation in determining eye final position would increase the number of saccades produced during pursuit. This result is what our results show, given that we found a higher frequency of anticipatory saccades in AET group. Furthermore, it is known that omnipause (OPN) cells are controlled by SC. OPN cells are continuously firing, and it is a signal of SC that causes a pause in these cells, producing a saccade. If there is an impairment in SC activity, it could send inhibitory signals to OPN, causing the production of undesired saccades that might disturb smooth pursuit performance.⁵¹ There are also studies that suggest that SC is influenced by attention.^49,52 This finding supports a possible link between anticipatory saccades and problems with attentional focus.⁴⁶ Considering all these aspects, we believe it is reasonable to suggest that AET impairment might be related to SC functioning. It may be that the excess of hypermetropia at early ages (until refraction is corrected) that causes the esotropia creates blur in vision,¹⁷ which leads to problems in eye position control. This might be coherent with problems in the SC.

Regarding the pursuit characteristics obtained in the IXT group, our hypothesis is that the affected areas could be different. For example, the ventral paraflocculus, a cerebellar region of the floccular complex, is involved in the smooth pursuit control system.⁵³ This area has been linked to our ability to maintain fixation and with the vestibulo-ocular reflex,⁵⁴ but also seems important to track target velocity and position. Problems in the ventral paraflocculus produce impaired pursuit,^50,55 which is what we have seen for IXT (lower gain). Zee et al.⁵⁶ have also reported a reduced gain of 0.65 in a monkey with ablation of flocculus and paraflocculus areas. As we have explained, the pursuit system at those ages is not completely developed, so an interruption of binocularity, or even suppression in some cases,¹⁸ owing to exotropia, could disrupt or delay the proper development of these areas.

Overall, our results provide important insights into how different types of strabismus might affect the oculomotor system. Still, more research is necessary to understand the processes implied and how these types of strabismus develop. Most important, this study shows that the characteristics of smooth pursuit eye movements are differently affected in the two most common types of strabismus. This factor should be considered when comparing performance between clinical groups and controls and might be a relevant issue in the design of clinical interventions.