Gap effect and express saccades generation in amblyopia

Abstract

Amblyopia is a neurodevelopmental vision disorder that is associated with abnormal visual stimulation during early childhood. Although our knowledge regarding spatial vision deficits in amblyopic subjects is well established, the neural control of eye movements in amblyopia is yet to be explored. In the present study we have evaluated the gap effect, and for the first time (to our best knowledge), express saccades generation in amblyopic (strabismic as well as anisometropic) and age-matched control subjects. We have compared the saccadic latency under different gap conditions (“no gap,” 50 ms gap, and 200 ms gap), between the amblyopic and control groups. Our results have shown that saccadic latency was reduced during the gap paradigms both for amblyopic and control groups for all viewing conditions. Furthermore, the size of the gap effect was comparable for all groups and viewing conditions (both for short and long gap durations). In addition, consistent with previous results, the amblyopic eye has manifested an increased saccadic latency as compared to the nondominant eye in the control group. Regarding the occurrence of express saccades, the 200 ms gap condition was associated with an increased number of express saccades as compared to 50 ms gap and “no gap” conditions, both for amblyopic and control subjects. We did not observe any significant difference in terms of express saccades production between the control and amblyopic subjects. Our findings may suggest that amblyopia does not alter physiological mechanisms related to the efficiency of visual attention/fixation disengagement as supported by the observation that the gap effect and express saccades production was comparable between the normal and amblyopic subjects.

Introduction

Amblyopia is a neurodevelopmental vision disorder that is associated with abnormal visual stimulation (typically strabismus and/or anisometropia) during early childhood. Amblyopia is commonly defined as a unilateral or bilateral (though not so often) decrease of visual acuity. The literature suggests that amblyopia affects approximately 3% of population and is a major cause of vision loss in children (Webber & Wood, 2005; Levi, Knill, & Bavelier, 2015). In addition, reduced activation in the number of cortical areas as well as in lateral geniculate nucleus (LGN; Anderson & Swettenham, 2006; Bonhomme et al., 2006; Hess, Thompson, Gole, & Mullen, 2009) observed in amblyopic individuals, provides the evidence that abnormal visual experience during early childhood may affect physiological development of central nervous system (CNS). Although our knowledge regarding spatial vision deficits in amblyopia is rather well established, the data regarding neural control of eye movements in amblyopic subjects is yet to be explored. Several previous studies have shown that amblyopia may affect smooth pursuit, fixational eye movements, or fixation stability (Ciuffreda, Kenyon, & Stark, 1979; Gonzalez, Wong, Niechwiej-Szwedo, Tarita-Nistor, & Steinbach, 2012; Subramanian et al., 2013; Chung, Kumar, Li, R. & Levi, 2015; Ghasia, 2015; Shaikh, Otero-Millan, Kumar, & Ghasia, 2016; Kelly, Cheng-Patel, Jost, Wang, & Birch, 2018). In terms of saccadic eye movements, reduced precision of saccade amplitude or increased saccadic reaction time (latency) may be present in amblyopic subjects (Niechwiej-Szwedo, Chandrakumar, Goltz, & Wong, 2012; Perdziak, Witkowska, Gryncewicz, Przekoracka-Krawczyk, & Ober, 2014; McKee, Levi, Schor, & Movshon, 2016; Perdziak, Witkowska, Gryncewicz, & Ober, 2016). However, in the vast majority of previous research, saccadic latency in amblyopic subjects was measured during the classic paradigm in which the peripheral target is switched on simultaneously with an offset of fixation target—there is no gap condition. In the 60s, Saslow (1967) described interesting phenomena related to the pause between the offset of fixation target and onset of peripheral saccadic target. Namely, if the fixation target is removed shortly before the new target appears (the gap paradigm or gap condition), the saccadic reaction time is considerably (usually ∼20–40 ms) reduced (gap effect) as compared to the “no gap” condition (Saslow, 1967; Dorris & Munoz, 1995; Coubard, Daunys, & Kapoula, 2004). Thus, the gap effect refers to the saccadic latency advantage resulting from removing the fixation target before the onset of a new peripheral target (Dorris & Munoz, 1995). It is believed that the offset of the centrally fixated target, reduces the activity of fixation neurons in SC (superior colliculus), releasing the visual attention and permitting the saccadic system to respond more quickly to new stimuli (Everling, Paré, Dorris, & Munoz, 1998). Gap durations ∼200–300 ms were found to be optimal for demonstrating the gap effect (manifested both as reduced saccadic reaction times and as the increased percentage/number of express saccades; Saslow, 1967; Mayfrank, Mobashery, Kimmig, & Fischer, 1986; Dorris & Munoz, 1995; Braun and Breitmeyer, 1998). It was observed that express saccades (ES, usually defined as saccades with a very short latencies between 80–20 ms; Kingstone & Klein, 1993; Coubard et al. 2004; Bucci, Pouvreau, Yang, & Kapoula, 2005) tend to occur more often in humans not only during the gap conditions, but also under conditions of fronto-executive loading (cognitive distraction) or frontal lobes lesions, where cortical control may be weakened (Halliday & Carpenter, 2010). Thus, the neural mechanisms involved in express saccades initiation are of particular interest, because after the afferent and efferent neural transportation delays are subtracted from the latency time, very little time remains to decide whether and how to respond to the new stimuli (Carpenter, 1981; Fischer & Weber, 1993; Munoz, Dorris, Paré, & Everling, 2000). To our best knowledge, neither of previous research investigated the impact of amblyopia on the generation of express saccades with the use long (200 ms) and short (50 ms) gap durations. 

In the discussion section of our recent work (Perdziak, Witkowska, Gryncewicz, & Ober, 2018), we have suggested that it will be interesting to explore the impact of amblyopia on saccades control during the gap condition. Recently Gambacorta, Ding, McKee, and Levi (2018), using a Gabor target, have reported in a very small group of unilateral amblyopes (two anisometropic amblyopes, three strabismic amblyopes, and two control subjects) that the gap effect is similar in both amblyopic and dominant eyes; however, there is still a lack of sufficient data to describe the express saccades production and the existence of gap effect in amblyopic subjects in a more exhaustive manner. We hypothesize that the gap effect will be present in both amblyopic and normal eyes since amblyopia affects rather higher cortical processing associated with saccade initiation than processes related to saccade programming or visual fixation disengagement. Several previous studies have shown that the introduction of a gap period between the offset of fixation target and onset of peripheral saccadic target reduce saccadic latency and favor express saccades occurrence not only in adult subjects but also in young children (Munoz, Broughton, Goldring, & Armstrong, 1998; Klein & Foerster, 2001; Bucci et al., 2005), a finding that may indicate that subcortical areas involved in oculomotor control, such as the superior colliculus, are already matured during early childhood (Fukushima, Hatta, & Fukushima, 2000). Braun and Breitmeyer (1988) concluded that for visual attention disengagement, at least a 50 ms gap period is needed. Thus, if amblyopia delay neural circuitry associated with visual fixation disengagement, the gap effect under extremely short 50 ms gap period should be reduced or even absent for amblyopic eye viewing when comparing to normal eyes. 

Thus, the aim of this study was to (a) further explore the impact of amblyopia on the gap effect using long (200 ms) and short (50 ms) gap periods and based on a larger group of subjects, including anisometropic and strabismic amblyopes as well as control individuals; and (b) investigate the influence of amblyopia on the population of express saccades under different gap conditions. 

Materials and methods

Eight anisometropic amblyopes (aniso; mean age 30 ± 12 years; six women, two men), eight strabismic amblyopes (strab; mean age 38 ± 13 years; three women, five men) and nine age-matched control participants (control; mean age 32 ± 11 years; six women, three men) were recruited. None of our amblyopic participants underwent active amblyopia treatment during early childhood, but some of them underwent a short period (usually lasting several months) of passive occlusion therapy. In addition, two of our strabismic amblyopes underwent (unsuccessful) strabismus surgery during childhood (No. 1 and No. 3). Although three of our strabismic amblyopes (No. 3, 4, and 8) manifested small anisometropia, we decided not to create a third group of participants with mixed type of amblyopia and to classify subjects with strabismus as only strabismic amblyopes, irrespectively of their refractive status (Gambacorta et al., 2018). The clinical characteristics of strabismic as well as anisometropic individuals is presented in Table 1 and 2. For this study, anisometropia was defined as a difference in refractive error between the eyes ≥ 1.00 D in one or both (in case of astigmatism) meridians. Participants were recruited through private optometry practice, and an experienced optometrist (first author of this paper) provided vision examination for each participant prior to enrolling. Based on a medical interview, all participants were healthy without any ophthalmological, neurological, or psychiatric/mental diseases, and none of them was receiving medications, known to affect their attention or reaction time. 

Table 1

 

Clinical characteristics of subjects with anisometropic amblyopia. Notes: RE = right eye; LE = left eye; Interocular suppression was evaluated with the Worth 4 dot test at the distance of 6 m. Stereoacuity = Stereo Fly test at 0.4 m.

 

Table 2

 

Clinical characteristics of strabismic subjects. Notes: RE = right eye; LE = left eye; CXT = constant exotropia; CST = constant esotropia. Interocular suppression was evaluated with the Worth 4 dot test at the distance of 6 m. Stereoacuity = Stereo Fly test at 0.4 m.

 

Optometric examination

In all participants the optometric examination included the measurement of visual acuity (Snellen letter chart) with and without optical correction, ocular dominance (hole in the hands test), refractive error examination (static retinoscopy and subjective refraction), Worth four-dot test at far and near, the stereopsis (Stereo Fly Test), vertical and horizontal phoria examination–Maddox rod test, Amsler Grid test, and the slit lamp (anterior segment) examination. Eye alignment was evaluated with the use of a prism cover test at far and near. 

The inclusion criteria were 

•  
  age 16 years or older
•  
  no latent or congenital nystagmus
•  
  no history of ophthalmological, neurological, or psychiatric/mental diseases
•  
  interocular visual acuity difference of at least two lines on Snellen letter chart

Both eye examination and the saccadic eye movement measurements were performed by the same optometrist. This study was approved by the Bioethics Committee of Medical University in Poznan, and was performed in accordance with the ethical standards laid down in the Declaration of Helsinki. 

Apparatus

The eye movements were recorded using a photoelectric method (based on the amplitude changes of the infrared reflection from the cornea; Ober, 1994), providing 1 ms temporal and 5 arcmin spatial resolution for saccades detection. The onsets of saccadic movements were detected automatically on line, based on local minimum eye velocity, followed by a movement exceeding 5°/s. The eye movement sensor—Saccadometer Plus device (Ober Consulting, Poland) was used. Saccadic latencies and saccadic trajectories were recorded and reviewed off-line using LatencyMeter software provided by Ober Consulting. The stimulus for saccades was projected using miniature lasers providing maximum contrast target. Laser projectors were mounted on the sensor forehead plate. 

Experiment procedure

Saccadic latencies were measured during the “no gap” (gap 0 ms) condition as well as two gap periods (one short, gap 50 ms; and one standard, gap 200 ms). Saccadic trials were performed in silence, and the targets were displayed at the distance of 3 m. Before recording eye movements, all participants underwent the practice trials. During the experimental trials, the participants used their optimal optical correction. In the beginning, the subject was instructed to look at the central target and to make a saccade to the peripheral target as quickly and accurately as possible. The stimuli for saccadic refixation appeared randomly either 10° to the right or 10° to the left of the central fixation spot. Different gap conditions were presented in blocks (each block contains 50 saccadic trials), randomly. All participants completed 150 (3 × 50 trials for each gap duration) saccadic trials for each viewing condition: (a) binocular viewing (BV); (b) monocular viewing with the dominant eye (DE); and (c) monocular viewing with the nondominant (NDE)/amblyopic eye (AE). Thus, each subject examination required 450 saccadic responses and together with optometric examination, took about 90 minutes. The order of viewing conditions was randomized between the participants. 

Data analysis and statistics

We removed wrong directed saccadic responses from the analysis, as well as signal loss trials, too small and too large saccades (amplitude below/above half the distance to the target) and saccades overlapped by blinks. Moreover, saccades with latencies shorter than 80 ms and longer than 800 ms as well as with durations longer than 100 ms, were removed from the further analysis. Express saccades were defined as short latency saccades with latencies between 80 and 120 ms (Kingstone & Klein, 1993; Coubard et al., 2004; Bucci et al., 2005). All those criteria resulted in the elimination of 9.3% saccadic trials in the anisometropic group, 7.5% in the strabismic group, and 4.2% in the control group. Inside the anisometropic group, the percent of eliminated trials was the highest for amblyopic eye (NDE/AE: 12.1%; DE: 7.8; BV: 8.1%). A similar pattern related to the eliminated trials was observed for strabismic group (NDE/AE: 11.9%; DE: 4.7%; BV: 5.8%). With respect to the control group, the percent of eliminated trials was quite similar between the viewing conditions (NDE: 4%; DE: 5.6%; BV: 3%). Note that for most of our participants, viewing by the amblyopic eye was associated with discomfort or even headaches. In addition, we have observed intensive squinting, tearing, or blinking in some of our amblyopes, conditions which significantly affect oculomotor signal, increasing the percentage of eliminated trials. Furthermore, we have observed an increased number of extremely long reaction times (above 1000 ms) or abnormal saccades amplitude for amblyopic eye viewing (below/above half the distance to the target). Moreover, at the beginning of the experiment, some amblyopes (especially those with sever loss of visual acuity) showed more errors—wrongly directed saccades. All this contributed to the increased data excluded from the amblyopic eye viewing conditions. 

The gap effect was quantified for each subject by subtracting the average saccadic latency for the gap condition from the average saccadic latency for “no gap” condition (Matsuo et al., 2015). For main analysis, we used R statistical software in version 3.4.4 (R Core Team, 2018). Since the data was unbalanced (there were nine control, eight anisometropic, and eight strabismic participants), we ran two mixed-effect models using lme4 library (Bates, Mächler, Bolker, & Walker, 2015) with random intercept for participants. The library uses the maximum likelihood method for parameters estimation. 

For saccadic latency as a dependent variable, we used a mixed-effects linear model with main and interaction effects of paradigm, group, and eye. Note, that homogeneity of variance is not an important assumption for the mixed-effects linear model used for saccadic reaction time (SRT) data. Regarding normality, we note that the results are qualitatively the same when using natural logarithms of SRT. Thus, we report the analysis of saccadic latency on the original scale for easier interpretation. Violin plots, which show distributions of SRT in different subgroups of data, clearly indicate that saccadic latency is highly right-skewed. All effects were entered using orthogonal sum to zero contrasts. We used an analysis of the variance table of type III with Satterthwaite approximation for degrees of freedom in order to determine the significance of each term. We also report R² statistics for fixed effects and for the whole model (i.e., variance explained by fixed and random effects; Nakagawa & Schielzeth, 2012). 

Regarding the number of express saccades as a dependent variable, we used mixed-effects Poisson regression. However, Poisson distribution has the limitation in that its mean is equal to its variance, which is not usually observed in the data. To overcome this limitation we used an additional noise parameter to account for overdispersion (Gelman & Hill, 2006). All effects were entered using orthogonal sum to zero contrasts. Here, we used the analysis of deviance table with Type III Wald chi-square tests to test significance of terms entered to the model. As an intuitive measure of overall model fit we used Nagelkerke's R² (1991), statistic based on the log-likelihood ratio between full model and null model. As a full model, we used the model fitted with all reported fixed and random effects, whereas the null model contained only random effects. Thus, the statistics can be interpreted as “variance explained” by the fixed effects. 

For statistically significant main effects and interactions, we ran posthoc comparisons using emmeans library (Lenth, 2018) with Bonferroni correction for multiple comparisons. For both analyses, we reported analysis of variance/deviance tables with additional information regarding random effects, model fit statistics, and posthoc comparisons for significant effects with accompanying figures that presented raw data with central tendencies and dispersion measures. 

Results

Saccadic latency and the gap effect

Main results regarding saccadic reaction times are summarized in Table 3. Posthoc comparisons for significant main effects and the interaction are presented in Table 4, 5, and 6, respectively. Results in graphical form are depicted in Figures 1 and 2. We observed significant main effects of group, paradigm, eye, and significant interaction of group and eye. Fixed effects accounted for 38% of latency variance, whereas random intercept for participants accounted for additional 37% of the dependent variable variance. 

Table 3

 

Main results regarding saccadic latency. Results of mixed-effects linear regression. Notes: SS = sum of squares; Mean SS = mean sum of squares; df1, df2 = degrees of freedom.

 

Table 4

 

Results of posthoc analyses for significant main effects with saccadic latency as dependent variable. Notes: estimate = regression coefficient; SE = standard error of the estimate; lower CL = lower (2.5%) confidence level; upper CL = upper (97.5%) confidence level.

 

Table 5

 

Results of posthoc analyses for the significant interaction effect of eye and group factors with saccadic latency as dependent variable. Notes: estimate = regression coefficient; SE = standard error of the estimate; lower CL = lower (2.5%) confidence level; upper CL = upper (97.%) confidence level.

 

Table 6

 

Results of posthoc analyses for the significant interaction effect of group and viewing conditions. Notes: estimate = regression coefficient; SE = standard error of the estimate; lower CL = lower (2.5%) confidence level; upper CL = upper (97.5%) confidence level.

 

Figure 1

 

Summarized results of saccadic latency as a function of group, paradigm and viewing condition. Results of posthoc analyses with violin plots. Boxes show an interquartile range. Horizontal bold lines are medians. Gray dots show raw data, and violins present (mirrored) shape of distributions of the dependent variable. Each gray point represents the smallest unit of measurement (i.e., for one participant, within single watching condition for single experimental paradigm); ends of the violins present minimum and maximum values; blue points and horizontal lines show point estimates of the differences between means and 95% CIs of the difference. Red vertical lines show the position of zero relative to the difference estimate, and black horizontal lines indicate to which groups the comparison relates. Main effect of paradigm was significant: saccadic latencies were longest during “no gap” (stp) condition and shortest in the gap 200 ms condition. Amblyopic eye in strabismic as well as anisometropic group showed an increased saccadic latency as compared with nondominant eye in control group.

Figure 2

 

Results of posthoc analyses with violin plots. Boxes show an interquartile range. Horizontal bold lines are medians. Gray dots show raw data, and violins present (mirrored) shape of distributions of the dependent variable. Each point represents the smallest unit of measurement (i.e., for one participant, within single watching condition for single experimental paradigm); ends of the violins present minimum and maximum values; blue points and horizontal lines show point estimates of the differences between means and 95% CIs of the difference. Red vertical lines show the position of zero relative to the difference estimate, and black horizontal lines indicate to which groups the comparison relates. For all paradigms taken together, the amblyopic eye showed an increased saccadic latency when compared to the dominant eye or binocular condition.

As we expected, the amblyopic eyes (both for anisometropic and strabismic groups) showed an increased saccadic latency when comparing to nondominant eyes from the control group (Table 5). Furthermore, there were no differences in saccadic latency for the dominant eyes as well as binocular conditions between the all three groups. To our surprise, we did not find any statistical difference between the amblyopic eyes of anisometropic and strabismic subjects (Table 5). For a paradigm factor, all comparisons were significant, with longest saccadic latencies in the “no gap” condition and shortest latencies in the 200 ms gap condition (Table 4, Figure 1). To our best knowledge, neither of the previous studies used such a short (50 ms) gap period to verify whether amblyopias delay also the neural circuitry associated with visual fixation disengagement. Interestingly, in the present study, we have observed that the size of the gap effect was comparable between the groups and viewing conditions. That is, not only normal eyes, but also amblyopic eyes showed a saccadic latency advantage (gap effect) during the short (50 ms) and long (200 ms) gap durations, which may indicate that amblyopia does not delay the processes related to the visual fixation disengagement (see Figure 6 and Table 11). Summarized values of a mean gap effect size and mean saccadic latency for particular groups are presented (see Table 11, Table 12, and Figure 6). 

Figure 3

 

Number of express saccades as a function of paradigm and viewing condition. Results of posthoc analyses with violin plots. Boxes show an interquartile range. Horizontal bold lines are medians. Gray dots show raw data, and violins present (mirrored) shape of distributions of the dependent variable. Each point represents the smallest unit of measurement, that is the number of express saccades within a single trial (i.e., for one participant, within single watching condition for single experimental paradigm); ends of the violins present minimum and maximum values; blue points and horizontal lines show point estimates of the differences between means and 95% CIs of the difference. Red vertical lines show the position of zero relative to the difference estimate, and black horizontal lines indicate to which groups comparison relates.

Figure 4

 

Number of express saccades as a function of paradigm x group (top) and eye x group (bottom). Results of posthoc analyses with violin plots. Boxes show an interquartile range. Horizontal bold lines are medians. Gray dots show raw data, and violins present (mirrored) shape of distributions of the dependent variable. Each point represents the smallest unit of measurement—that is, the number of express saccades within a single trial (i.e., for one participant, within single watching condition for single experimental paradigm); ends of the violins present minimum and maximum values; blue points and horizontal lines show point estimates of the differences between means and 95% CIs of the difference. Red vertical lines show the position of zero relative to the difference estimate, and black horizontal lines indicate to which groups comparison relates. For all three groups, the 200 ms gap condition produced the highest number of express saccades compared with the 50 ms gap and “no gap” conditions. Moreover, for all three groups, nondominant (NDE) eye viewing was associated with a significantly less number of express saccades as compared to BV condition.

Figure 5

 

Number of express saccades for amblyopic/nondominant eye under the 200 ms gap condition. Horizontal bold lines are medians. Although no significant differences between the groups were observed, on a descriptive level one may note, that for an amblyopic eye viewing, the medians are slightly lower than for nondominant eye from the control group. In addition, no anisometropic amblyopes showed more than 14 express saccades, and no strabismic amblyopes showed more than eight express saccades. Thus, future research on a larger group of subjects and larger saccadic trials is needed to further explore the impact of amblyopia on the express saccades occurrence for the 200 ms gap duration.

Figure 6

 

Mean gap effect size (for 200 ms and 50 ms gap conditions) as a function of the group for particular viewing conditions. Note: The vertical bars indicate the standard error of the mean value; AE = amblyopic eye; NDE = nondominant eye; DE = dominant eye; BV = binocular viewing. Analysis did not show any differences in the size of the gap effect between the groups.

Table 7

 

Main results regarding express saccades. Results of mixed-effects Poisson regression.

 

Table 8

 

Results of posthoc analyses for significant main effects with express saccades number as dependent variable. Notes: estimate = regression coefficient; SE = standard error of the estimate; lower CL = lower (2.5%) confidence level; upper CL = upper (97.5%) confidence level. Note that ratio estimates of the number of express saccades are shown; thus, the effect is significant when Cis excludes 1.

 

Table 9

 

The results from posthoc analyses for the significant interaction effect of paradigm and group factors with express saccades number as dependent variable. Notes: estimate = regression coefficient; SE = standard error of the estimate; lower CL = lower (2.5%) confidence level; upper CL = upper (97.5%) confidence level. Note that ratio estimates of the number of saccades on a response scale are shown; thus, the effect is significant when Cis excludes 1.

 

Table 10

 

The results from posthoc analyses for the significant interaction effect of the eye and group factors with express saccades number as dependent variable. Notes: estimate = regression coefficient; SE = standard error of estimate; lower CL = lower (2.5%) confidence level; upper CL = upper (97.5%) confidence level. Note that ratio estimates of the number of saccades on a response scale are shown; thus, the effect is significant when Cis excludes 1.

 

Table 11

 

Summarized results of mean gap effect for each group and viewing conditions.

 

Table 12

 

Summarized results of mean saccadic latency for particular groups and viewing conditions.

 

Table 12

 

Extended

 

Number of express saccades

To our best knowledge, up until now, no previous research has investigated the impact of amblyopia on the express saccades generation. It is known that express saccades occur most readily when only a single target is presented after fixation offset and the introduction of a gap period facilitates express saccade production (Shiller et al., 2004). In our paradigm, we used two different gap periods (short, 50 ms; and long, 200 ms) and after fixation target disappearance, the single peripheral target appeared randomly either 10° to the right or 10° to the left of the central fixation spot. 

Main results regarding the number of express saccades are summarized in Table 7. We observed significant main effects of paradigm and eye and significant interactions of group and paradigm, as well as group and eye. Posthoc comparisons for significant main effects and the interaction are presented in Table 8, 9, and 10. Fixed effects in the model accounted for 62% of the express saccades “variance” (when comparing to null model with random effects only). The interactions of paradigms and groups, as well as eye and group were insignificant. 

We have found that the “no gap” condition was associated with the lowest number of express saccades, whereas the 200 ms gap condition had the highest number of express saccades. Further, the binocular viewing (BV) condition leads to a higher number of express saccades when compared to both dominant (DE) and nondominant (NDE) eye viewing conditions (see Table 8 and Figure 3). 

As we expected, the highest number of express saccades occurred during the 200 ms gap condition compared with 50 ms gap and “no gap” conditions, a result which in a good agreement with previous studies (Mayfrank et al., 1986; Braun & Breitmeyer, 1988). Posthoc analysis for the group and paradigm interaction revealed that all comparisons were statistically significant, but the differences between paradigms were most prominent within the control group (see Table 9 and upper panel of Figure 4). 

Although we did not observe any significant differences between the amblyopic and normal eyes related to the number of express saccades, posthoc analysis for the group and eye interaction revealed that for all three groups, nondominant (NDE) eye viewing was associated with significantly fewer express saccades as compared to the BV condition. (see Table 10 and lower panel of Figure 4). 

Discussion

In the present study, we have evaluated the gap effect, and for the first time (according to our knowledge), express saccades' generation in amblyopic (strabismic as well as anisometropic) and age-matched control subjects. The primary aim of this study was to compare the saccadic latency and number of express saccades under different gap conditions (“no gap,” 50 ms gap, and 200 ms gap), between the groups of amblyopic and control subjects. To be specific, this study explores two questions: (a) Does the amblyopic eye show a gap effect that is comparable to the dominant eye or normal eyes for both long and short gap periods? (b) Does the amblyopic eye show a number of express saccades that is comparable to those of dominant or normal eyes? 

We found that for all groups and viewing conditions (AE/NDE, DE, and BV), saccadic latency was reduced during the gap conditions, and there was no significant difference in the number of express saccades between the amblyopic and nondominant eyes from the control group. Furthermore, the size of the gap effect was quite similar both in the amblyopic and control groups for all viewing conditions: The analysis did not show any significant differences. That is, not only dominant eye viewing but also amblyopic eye viewing was associated with saccadic latency advantage during the gap conditions (for both the 50 ms and 200 ms gap durations) in strabismic as well as anisometropic subjects (see Figure 6, Table 11). Furthermore, as we expected, the amblyopic eye in the strabismic as well as the anisometropic group showed an increased saccadic latency as compared to the nondominant eye in the control group (Figure 1). Additionally, for both DE and BV conditions, saccadic latency did not differ among all three groups. However, to our surprise, the amblyopic eye of strabismic subjects did not differ from the amblyopic eye of anisometropic subjects in terms of saccadic reaction time. Regarding the express saccades, we expected that 200 ms gap condition will favor the express saccades production as compared to gap 50 ms and “no gap” conditions, and that was indeed what we found. Furthermore, for all paradigms taken together, NDE did not differ from DE in terms of the number of express saccades—for both the control and amblyopic groups. However, surprisingly, we have observed that for all three groups, the binocular condition was associated with an increased number of express saccades as compared to the nondominant eye viewing condition. 

Saccadic latency in amblyopia

It is commonly known that saccadic reaction time in response to amblyopic eye stimulation is substantially increased as compared to the dominant eye or normal eyes (Ciuffreda, Kenyon, & Stark, 1978; Niechwiej-Szwedo et al., 2012; Perdziak et al., 2014; McKee et al., 2016; Perdziak et al., 2016). In the present study, we have also observed an increased saccadic latency for amblyopic eye viewing that is in agreement with previous research. However, the nature and the neural correlates of saccadic latency increase in amblyopia are still under debate. In the 1970s Ciuffreda et al. (1978) suggested that abnormal saccadic latency in amblyopia should be interpreted in terms of processing delay in the sensory pathways. Similarly, Niechwiej-Szwedo, Goltz, Chandrakumar, Hirji, & Wong (2010) concluded that the saccadic latency increase in amblyopia may be associated with slower visual processing in the afferent pathway rather than with a deficit in the efferent (motor) pathway of the saccadic system. Thus, amblyopia may affect more complex cortical processing which is required to decide whether and how to respond to the new visual stimuli. Interestingly, considering saccadic reaction time as a decision time, we have been recently shown that amblyopia may affect cortical decision processes related to the initiation of saccadic response (Perdziak et al., 2014; Perdziak et al., 2018). Furthermore, McKee et al. (2016) speculated that abnormalities in fixational eye movements (frequent microsaccades) and the accompanying attentional shifts may significantly affect saccadic reaction time in amblyopic eyes. Interestingly, several behavioral studies of normal observers showed that faster reaction times were associated with a reduction in the frequency of microsaccades, i.e., responses were faster on trials without microsaccades (Betta & Turatto, 2006; Kliegl, Rolfs, Laubrock, & Engbert, 2009). Although it was shown that saccadic latency is increased when targets are presented during microsaccades (Hafed & Krauzlis, 2010), the oculomotor data related to the characteristics of microsaccades in amblyopia are rather inconclusive. Ghasia (2015) found a decrease in the frequency of microsaccades with the increasing severity of amblyopia. Shaikh et al. (2016) studied fixational eye movements in amblyopic as well as healthy age-matched controls and reported an increase in the amplitude with decreased frequency of fixational saccades in children with amblyopia. On the other hand, Chung et al. (2015) examined 44 (twenty-eight had amblyopia) adult subjects and found the increased frequency of microsaccades for the amblyopic eye of strabismic amblyopes, compared with the control or the amblyopic eye of anisometropic amblyopes. Thus, additional research (using high-resolution measuring system) on both pediatric and adult amblyopes is needed to further investigate the effect of microsaccades characteristics in amblyopia on the saccadic latency. 

Although in our recent work we have reported that an amblyopic eye in strabismic group showed an increased saccadic latency when compared to an anisometropic group, to our surprise, in the present study we did not find any statistical difference between the amblyopic eyes of anisometropic and strabismic subjects. One plausible explanation for this discrepancy may be a sampling problem, since not all of our strabismic subjects manifested increased saccadic latency compared with anisometropic subjects for amblyopic eye viewing. Furthermore, in the present, as well as in our previous research (Perdziak et al., 2014; Perdziak et al., 2016), we have used the same eye movements recording system, Saccadometer Plus device (Ober Consulting, Poland), providing very bright, maximum contrast stimulation; targets are provided by the laser spot projectors. Thus, it seems that the stimulus attributes (e.g., luminance, contrast) should not account for this discrepancy. Interestingly, Gambacorta et al. (2018) showed that for some amblyopes (especially strabismic), saccadic latencies were increased, even when the signal strength in the two eyes was equated. Thus, future research on a larger group of amblyopic subjects is needed to further explore the source of difference in saccadic latency between the strabismic and anisometropic amblyopes. 

Gap effect and express saccades

The visual fixation disengagement during the gap trials results in faster saccadic responses towards the target, known as the gap effect (Dorris & Munoz, 1995; Liversedge, Gilchrist, & Everling, 2011). Removing the fixation target reduces the inhibition upon the saccadic system, allowing it to respond more quickly to the onset of peripheral target. But how it works in amblyopic visual system is still under debate. The decrease in activity of SC fixation cells during the gap period reduces the inhibition upon the saccade system, allowing it to respond more quickly to the target (Dorris & Munoz, 1995). It was shown that a maximum reduction in saccadic latency occurs with a gap period of ∼200–300 ms, and multiple components may contribute to the gap effect. By itself the disappearance of a fixation target may both disengage the visual fixation system and act as a warning signal of impending target appearance (Saslow, 1967; Dorris & Munoz, 1995; Stevenson, Elsley, & Corneil, 2009). However, in most research, this reduction in saccadic latency is usually considered in terms of visual fixation disengagement, prior to target reappearance which allows an increase in neural activity of saccade-related neurons in the brainstem reticular formation, superior colliculus (SC) and frontal eye fields (FEF; Munoz et al., 2000). Since the disengagement of visual attention from the current fixation target is essential for programming and initiating the next saccade, the gap paradigms are often used to investigate the efficiency of attentional disengagement (Van der Stigchel, Hessels, van Elst, & Kemner, 2017). 

Whereas our knowledge regarding saccadic latency increase for an amblyopic eye is rather well established, the data regarding gap effect and the ability to produce express saccades in amblyopia is yet to be explored. Recently, Gambacorta et al. (2018) compared latency of saccadic response with and without gap condition and found that removing the fixation target speeds up the reaction time in amblyopic eyes, and the size of the gap effect is comparable between both amblyopic and fellow eyes. Although Gambacorta et al. (2018) eliminated from the analysis saccadic latencies below 120 ms, included a small group of amblyopic subjects (two anisometropic and three strabismic amblyopes), and used different stimulation (Gabor target chosen randomly from one of six possibilities: 0 ms, 100 ms, 200 ms, 300 ms, and 400 ms gap conditions), they have demonstrated similar patterns regarding the gap effect, as the authors found in the present study. In the present study, the size of the gap effect was comparable between the amblyopic and control subjects, both for 50 ms gap and 200 ms gap periods (see Figure 6 and Table 11). Similar values of the gap effect were found also in previous studies (Braun & Breitmeyer, 1988; Dorris & Munoz, 1995; Munoz et al., 2000). Very short gap periods (e.g., 50 ms or 75 ms) were also used in previous research on normal subjects (Braun & Breitmeyer, 1988; Takagi et al., 1995). However, to our best knowledge, neither of previous research used a very short (∼50 ms) gap period to test the ability of visual attention disengagement in amblyopic subjects. Braun and Breitmeyer (1988) noted that at least a 50 ms gap period is needed to evoke the gap effect, an observation which is in a good agreement with our findings. Similarly, Takagi et al. (1995) showed reduced saccadic latency under the short (75 ms) gap condition compared with the “no gap” (simultaneous) condition. Thus, it seems that even very short (50 ms) gap period is sufficient to evoke the gap effect in both normal and amblyopic eyes, a conclusion that may indicate that amblyopia does not delay the neural circuitry associated with visual fixation disengagement. Based on our results, one may additionally speculate that amblyopia affects mainly the supply of visual information to the central nervous system, delaying the saccadic response. At the same time, the processes related to saccades programming remain intact during the gap conditions, probably because they are rather well established already during the early childhood. This interpretation may by supported by the fact that the gap effect as well as an increased occurrence of express saccades is observed not only in adults but also in young children during the gap conditions (Munoz et al., 1998; Klein & Foerster, 2001; Bucci et al., 2005; Bucci & Seassau, 2012). However, future studies on a larger group of amblyopic subjects are needed to further explore this hypothesis. 

The reaction time of express saccades approach minimal time required for visual information to travel from the retina to the extraocular muscle motoneurons via the shortest possible route (Munoz et al., 1998) and the express saccades provide a way of testing the status of collicular function in humans (Leigh & Zee, 2015). In the present study, we did not observe any significant difference in terms of express saccades production between the normal and amblyopic subjects. As we expected, the highest number of express saccades occurred during the 200 ms gap condition. Similarly, Braun and Breitmeyer (1988) observed that express saccades were most frequent at a gap duration of 200 ms compared with a 50 ms gap duration. Further, Mayfrank et al. (1986) observed an increased number of express saccades for 200 ms gap durations and a slow decrease for longer or shorter gap periods. Based on the above studies, one may additionally speculate that there is an optimal (∼200–300 ms) gap duration for express saccades generation, since both significantly longer (500–800 ms) and shorter (50–100 ms) gap durations were associated with a reduced percent of express saccades compared with the 200–300 ms gap period (Mayfrank et al.,1986; Braun & Breitmeyer, 1988). Further, an animal study (Sparks, Rohrer, & Zhang, 2000) showed that a brief (∼100 ms) period of reduced collicular activity began about 60–70 ms after the fixation target disappearance and that was followed by an increase in activity that continued until saccade initiation, a study that seems to support our observation that express saccades tend to occur more often under the long (200 ms) gap compared with short (50 ms) gap. Thus, our findings may indicate that the ability to produce express saccades (which is associated with adequate functioning of saccadic burst cells in the superior colliculus; Clementz, 1996) present similar patterns in both amblyopic and control subjects (Figure 5). Furthermore, it seems that amblypia does not affect the ability to turn off the visual fixation mechanism, probably because subcortical areas involved in oculomotor control such as SC are already matured during early childhood (Fukushima et al., 2000). However, in our experiment we have recorded only 50 saccadic trials for each gap period and viewing condition. Consequently, the number of express saccades was relatively small in all groups, and future research on larger population of saccades is needed to further explore the impact of amblyopia on the express saccades generation. Note that after afferent and efferent delays are subtracted from the latency time, very little time remains to compute the location and to program the movement (Sparks et al., 2000). Thus, it might be interesting for future research to examine also other parameters of express saccades such as amplitude and direction in amblyopic subjects. 

To our surprise, we have observed that for all three groups, the BV condition was associated with an increased number of express saccades as compared to the NDE condition. Although our findings will not provide a conclusive explanation, one may speculate that binocular inputs provide an increased (accumulated) probability of gap detection between the offset of a fixation target and onset of a peripheral target, allowing short-latency saccadic responses to occur more often during the gap conditions. Nonetheless, future research is needed to explore the effect of binocular viewing on express saccades generation. Several previous studies suggested that short latency saccades tend to occur more often when there is a less cortical inhibition from higher regions like supplementary and frontal eye fields towards more primitive neural regions controlling saccades, such as the superior colliculus (Noorani, 2014). However, based on our results, it seems that amblyopia does not affect neural mechanisms of inhibitory control from higher cortical regions toward subcortical areas involved in express saccades generation. 

In conclusion, our findings may suggest that abnormalities in visual stimulation (strabismus and/or anisometropia) during early childhood do not alter physiological mechanisms related to the efficiency of visual attention/fixation disengagement as supported by the authors' observation that the gap effect and express saccades occurrence was comparable between the groups and viewing conditions. 

Acknowledgments

Commercial relationships: none. 

Corresponding author: Maciej Perdziak. 

Email: . 

Address: Nalecz Institute of Biocybernetics and Biomedical Engineering, Polish Academy of Sciences, Warsaw, Poland.