Short-term monocular deprivation strengthens the patched eye's contribution to binocular combination

Abstract

Abstract

Abstract:

Abstract  Binocularity is a fundamental property of primate vision. Ocular dominance describes the perceptual weight given to the inputs from the two eyes in their binocular combination. There is a distribution of sensory dominance within the normal binocular population with most subjects having balanced inputs while some are dominated by the left eye and some by the right eye. Using short-term monocular deprivation, the sensory dominance can be modulated as, under these conditions, the patched eye's contribution is strengthened. We address two questions: Is this strengthening a general effect such that it is seen for different types of sensory processing? And is the strengthening specific to pattern deprivation, or does it also occur for light deprivation? Our results show that the strengthening effect is a general finding involving a number of sensory functions, and it occurs as a result of both pattern and light deprivation.

Introduction

A fundamental feature of primate vision is the extent to which information is combined between the two eyes' inputs. The first stage of binocularity is seen in the anatomical arrangement of left and right eye inputs at the earliest cortical level (Hubel & Wiesel, 1977). The ocular dominance columns, which are the first stage of the anatomical foundation of binocularity, are present at birth (Horton & Hocking, 1996) and can only be disrupted by a prolonged period of monocular deprivation (Hubel & Wiesel, 1970). The contribution of the two eyes is often not totally balanced, with one predominating over the other, resulting in a left or right eye sensory dominance (Li et al., 2010; Zhang, Bobier, Thompson, & Hess, 2011). 

Recently, it has been demonstrated that, in normal adults, short-term patching can result in a stronger contribution of the previously patched eye to binocular combination if the patching excludes pattern information but not mean luminance (i.e., a diffuser as opposed to an occluder) (Lunghi, Burr, & Morrone, 2011). The rate of binocular alternation (i.e., binocular rivalry) was taken as the measure of sensory dominance. This result was unexpected, being at odds with a number of reports suggesting depriving one eye leads to either a reduction in sensitivity of the patched eye (Lou et al., 2011) or an increased plasticity of the unpatched eye (Shibata, Kawato, Watanabe, & Sasaki, 2012). 

To resolve this issue, we used a number of different and possibly more accepted indices of sensory dominance to assess the generality of this finding. Our measures were designed to reflect different visual functions at different sites along the cortical pathway. We were also particularly interested to see whether these beneficial patched-eye effects were specific to the use of a diffuser as opposed to an occluder. To test this, we compared light-tight patching, which excludes pattern and mean luminance information, with a diffuser, which only excludes pattern information. 

Methods

Participants

Five normal adults (mean age: 30.4 ± 4.9 years old; three males) with normal or corrected-to-normal vision participated in the patching experiment. Except the first author, all subjects were naïve to the purpose of this study. An information consent form was signed by each of them. The study was approved by the Institutional Review Board of McGill University. 

Apparatus

All the interocular sensory balance measurements were conducted on a Mac computer using Matlab and PsychToolBox 3.0.9 extensions. The stimuli were dichoptically presented by head-mount goggles (eMagin Z800 pro, OLED) with a refresh rate of 60 Hz and a resolution of 800 × 600. The mean luminance of the OLED goggles was 190 cd/m². 

The monocular contrast threshold was measured on a Mac computer using PsyKinematix (Beaudot, 2009), and the stimuli were presented on a Mitsubishi Diamond Pro 2070SB monitor. 

Experimental design

The experiment consisted of three consecutive stages: a prepatching measurement of binocular balance, a patching stage (2.5 hours), and a postpatching measurement of binocular balance (four sessions immediately after the removal of the patch and one session 30 minutes after the removal of the patch). The effects of two patching methods, transparent patching (also called a diffuser, which transmits light but not pattern) and opaque patching (light-tight patching, which excludes pattern and mean luminance information), were quantified using three interocular suppression measure techniques: a binocular phase combination task (Ding & Sperling, 2006; Huang, Zhou, Lu, Feng, & Zhou, 2009), a dichoptic global motion coherence task (Mansouri, Thompson, & Hess, 2008), and a binocular contrast matching task (Huang, Zhou, Lu, & Zhou, 2011; Huang, Zhou, Zhou, & Lu, 2010). In these three interocular suppression measures, the contrast of the stimulus in the unpatched eye was fixed at 64%. The contrast of the stimulus in the patched eye was chosen based on an individual's performance in the preliminary experiment: Performance was investigated at several contrast levels around 64%, and the contrast that achieved matched performance (when the two eyes were nearly equally effective in binocular combination) in the prepatching measurement was chosen. In particular, for the binocular phase combination task (Figure 1A), the contrast at which the perceived phase of the binocular combined grating was near 0°, and for the dichoptic global motion coherence task (Figure 1B), the contrast at which the coherence thresholds were nearly equal no matter which eye saw the signal was chosen. For the binocular contrast matching task (Figure 1C), the same contrast as that used in the binocular phase combination task was used. Averaged across subjects, the interocular contrast ratio we used in the binocular phase combination task (δ in Figure 1A) and in the dichoptic global motion coherence task (θ in Figure 1B) was 0.946 ± 0.037 (mean ± SE) and 0.934 ± 0.044, respectively. For the two subjects who participated in the binocular contrast matching task (i.e., N1 and N2), the interocular contrast ratios (δ in Figure 1C) were 0.87 and 0.80, respectively. For each observer, the patched eye was randomly selected and fixed in different probed task and patching method pairs (on different days). 

Figure 1

 

Three interocular suppression measurement techniques. (A) Binocular phase combination task. C = 64%; δ = 0.946 ± 0.037 (mean ± SE). (B) Dichoptic global motion coherence task (28 dots in the two eyes are illustrated here while the total number of dots in two eyes was fixed as 100 in the measure). C = 64%; δ = 0.934 ± 0.044. (C) Binocular contrast matching task. C = 64% for the two subjects who participated in this test (i.e., N1 and N2), δ was 0.87 and 0.80, respectively.

To further illustrate the patching effect, the effective contrast ratio when the two eyes were balanced (i.e., balance point) in the dichoptic global motion coherence task was measured in two subjects (i.e., N1 and N2) before and after the transparent patching. A quick balance point assessment method that had been previously validated (Black, Thompson, Maehara, & Hess, 2011) was used to quantify the effective contrast ratios at the previously used time sessions. 

Monocular contrast thresholds at 0.3 c/d (corresponded to that used in the phase combination and contrast matching tasks) of both the patched and unpatched eyes were also measured for the transparent patching in subjects N1 and N2 (on different days). 

Experimental procedures

Binocular phase combination task

Two horizontal sine-wave gratings (0.3 c/d, 6.6° × 6.6°) with phase shifts in opposite directions of the same magnitude were dichoptically presented to the two eyes. The perceived phase of the grating after binocular combination depends on the internal representations of the two inputs. Therefore, any variations of the binocular sensory balance could be quantified by the changes of the perceived phase. To cancel potential positional bias, two configurations were used to access the perceived phase: In configuration 1, the phase shift was +22.5° in the unpatched eye and −22.5° in the patched eye; in configuration 2, the phase shift was −22.5° in the unpatched eye and +22.5° in the patched eye. Each configuration was repeated eight times using constant stimuli in one measure session. In all, there were 16 trials randomly interleaved in one measure (two configurations × eight repetitions). Normally, subjects could finish one measure in three minutes after a short period of practice. The perceived phase at configuration 1 minus the perceived phase at configuration 2 and then divided by 2. Thus, if the patched eye becomes stronger after patching, the perceived phase should be decreased and vice versa if the unpatched eye becomes stronger (see illustration in Figure 1A). 

In measuring perceived phase, observers were asked to adjust the height of a one-pixel sided reference line to indicate the perceived phase of the grating after binocular combination, defined as the location of the center of the dark stripe of the grating. The initial height of the reference line was randomly (−9 to 10 pixels) positioned relative to the center in each trial. The reference line was moved with a fixed step size of one pixel, corresponding to a 4° phase angle of the sine-wave grating. More details of the use of this paradigm can be found elsewhere (Huang et al., 2009). 

Dichoptic global motion coherence task

One hundred moving dots (each with a diameter 0.40° of the visual angle) with randomized position were presented in a circular display window (diameter 8.4° of the visual angle). Some of them were assigned as signal dots and had a coherent motion direction while others were assigned as noise dots and had a random motion direction. The signal and noise dots, with a total number of 100, were always dichoptically presented in the two eyes. The coherence threshold, i.e., the number of coherently moving dots that is needed in one eye to discriminate the coherent motion direction from those randomly moving dots in the other eye, depends on the degree of interocular suppression. Therefore, any variations of the binocular sensory balance could be quantified by the changes in the coherence threshold ratio, which was obtained by dividing the coherence threshold that was measured when the patched eye viewed the signal and the unpatched eye viewed the noise (configuration 1) by the coherence threshold that was measured when the patched eye viewed the noise and the unpatched eye viewed the signal (configuration 2). If the patched eye becomes stronger after patching, this coherence threshold ratio should be decreased; otherwise, the coherence threshold ratio should be increased (see illustration in Figure 1B). 

In this measure, the number of coherently moving dots (in one eye) was varied using a three-up, one-down staircase method with a proportional step size of 50% before the first reversal and 25% in the following trials; the number of randomly moving dots (in the other eye) was varied correspondingly to make the total number of dots in two eyes fixed at 100. The two staircases at different configurations were randomly interleaved and terminated at the sixth reversal point. The coherence thresholds (79% correct performance) were then calculated based on the last five reversals of each staircase. Normally, subjects could finish one measure in three minutes after a short period of practice. For each subject, the patching effect was quantified by the average of two repetitions in two days. More details of the use of this paradigm can be found elsewhere (Mansouri et al., 2008). 

Binocular contrast matching task

Two horizontal in-phase sine-wave gratings (0.3 c/d, 2.2° × 6.6°) with the same relative contrasts as the stimuli in the binocular phase combination task were dichoptically presented on the left side of fixation in the two eyes; a monocular probe grating with randomized contrast but of the same size and phase as these two dichoptic pairs was presented on the right side of fixation in the patched eye (configuration 1) or the unpatched eye (configuration 2). Observers were instructed to adjust the contrast of the probe grating on the right side of the fixation to match the standard binocular-combined cyclopean grating on the left side of the fixation in these two configurations, respectively. Each configuration was repeated eight times using the method of constant stimuli in one measure session. Normally, subjects could finish one measure in three minutes after a short period of practice. The patching effect could be reflected by the ratio of the contrast matching results (i.e., monocular perceived contrast) from configuration 1 to configuration 2 because the monocular probe gratings in configuration 1 and 2 are perceptually equal (with the standard binocular-combined cyclopean grating). If this ratio declines, this means the patched eye needs less contrast to be perceptually equal to the unpatched eye (i.e., the patched eye getting stronger); otherwise, the ratio should be increased. This is illustrated in Figure 1C. More details of the use of this paradigm can be found elsewhere (Huang et al., 2011; Huang et al., 2010). 

Quick balance point assessment method

The quick balance point assessment method we used was identical to method 2 in Black et al. (2011), which involved two steps: First, the number of signal dots necessary for a binocular threshold was measured with the same stimulus parameters (64% contrast) as previously described. Second, using this number of signal dots in the unpatched eye (fixed at 64% contrast), the contrast threshold of the noise stimulus shown to the previously patched eye was varied until a dichoptic coherence threshold was reached. A three-up, one-down staircase method, beginning at 0% contrast in the patched eye with the contrast increasing in steps of 10% before the first reversal and 5% in the following trials and decreasing in steps of 10% in all trials, was used. The staircase terminated at the sixth reversal point. The contrast threshold (79% correct performance) was then calculated based on the last five reversals. Normally, subjects could finish one measure in three minutes after a short period of practice. The ratio of the contrast threshold (in the patched eye) to the base contrast (i.e., 64%, in the unpatched eye) represents the effective contrast ratio when the two eyes are balanced, i.e., the noise dots in the patched eye are having the same effect on global motion perception under the dichoptic viewing condition as that under the binocular viewing condition. If this ratio declines, this means the patched eye needs less contrast to be balanced with the unpatched eye (i.e., the patched eye is getting stronger); otherwise, the ratio should be increased. 

Monocular contrast threshold

Monocular contrast thresholds for a 0.3 c/d sine-wave grating seen by the patched eye and by the unpatched eye were measured using an orientation discrimination task and a constant stimuli method before and after the patching stage. During the measurement, the unmeasured eye was patched by an opaque patch. Subjects were instructed to answer whether the orientation of the grating was horizontal or vertical by key pressing. Probabilities of correct identification were measured at six contrast levels; each level contains 75 trials. Each measure lasted for 15 minutes. We measured one psychometric function before the patching stage and one psychometric function immediately after the patching stage. We fitted these two psychometric functions using Quick functions with parametric maximum likelihood estimation (Watson, 1979). The two parameters of the Quick function, alpha and beta, represent the threshold corresponding to 81.6% accuracy and the slope of the psychometric function at that point, respectively. The mean and variances of these two parameters were determined by 500 times' bootstrap simulation (Efron & Tibshirani, 1994). A two-independent-sample Z-test was conducted based on the bootstrap results to statistically estimate the changes after patching. 

Results

The effect of 2.5 hours of monocular occlusion with either a light-tight patch or a diffuser is shown in Figure 2 for the binocular phase combination task and the dichoptic global motion coherence task. The patching effect is plotted for five normal subjects for a range of different time points as illustrated above in the figure inset. All thresholds were normalized to that measured at time point T0, just before monocular deprivation. 

Figure 2

 

The effect of 2.5 hours of monocular occlusion with either a light-tight patch or a diffuser in the binocular phase combination task and the dichoptic global motion coherence task. (A) Experimental processing. (B) Patching effects in the binocular phase combination task (left panel) and dichoptic global motion coherence task (right panel). Error bars represent standard errors.

The measures using the binocular phase combination task showed that monocular deprivation with the light-tight patch or diffuser produced similar large changes in binocular balance in the direction of the patched eye's contribution being strengthened (Figure 2B, left panel). This effect peaked at around 0–6 minutes post occlusion (i.e., time points T1–T2) and diminished with time. The effects were still significant at 30 minutes post occlusion (i.e., time point T5) in both the light-tight patch conditioning, t(4) = 5.87, p = 0.004, two-tailed, and the diffuser patching condition, t(4) = 2.57, p = 0.062, two-tailed. The patching effects in the transparent and opaque patching were similar in their direction and time course but were not the same in magnitude, especially at time point T1, t(4) = −3.38, p = 0.028, two-tailed. A repeated measured-within-subjects ANOVA based on results from T1 to T5 showed that the main effect of patching was not significant: F(1, 4) = 2.69, p = 0.18. The main effect of time was significant: F(4, 16) = 5.46, p = 0.006. The interaction of these two factors was also significant: F(4, 16) = 3.19, p = 0.042. 

The dichoptic global motion coherence task also showed a similar effect in terms of interocular motion coherence ratios for five normal subjects (Figure 2B, right panel) in both patching methods [the main effect of patching: F(1, 4) = 0.36, p = 0.58]. Normalized motion coherence threshold ratios are plotted for different time points after monocular deprivation. Monocular occlusion with either the light-tight patch or diffuser strengthened the contribution of the patched eye in terms of its contribution to the binocular contrast percept. The strengthening effect was maximum at around 0–6 minutes post occlusion (i.e., time point T1–T2), which lasted for at least 12 minutes (i.e., time point T4). This effect recovered at around 30 minutes post occlusion (i.e., time point T5) in the light-tight condition. t(4) = 1.05, p = 0.35, two-tailed, but was still significant in the diffuser condition, t(4) = 3.65, p = 0.022, two-tailed. 

A binocular contrast matching task was also used to gauge how monocular perceived contrast changed as a result of short-term monocular deprivation. These results are displayed in Figure 3 for two subjects for whom the normalized monocular perceived contrast ratio is shown for different time points after monocular deprivation with either a light-tight patch or diffuser. Similar results were obtained as previously described for the phase and motion coherence tasks; a strengthening of the patched eye's contribution that lasted for at least 12 minutes post occlusion (i.e., time point T4) and returned to the baseline at around 30 minutes post occlusion. This result for binocular contrast complements the between-eye (i.e., the dichoptic) contrast matching results provided by Lunghi et al. (2011). 

Figure 3

 

The effect of 2.5 hours of monocular occlusion with either a light-tight patch or a diffuser in the binocular contrast matching task. (A) Experimental processing. (B) Normalized monocular perceived contrast ratio at different time spots of two subjects. Error bars represent standard errors.

These previously described measurements that focused on the change of visual performance when the interocular contrast ratio was fixed clearly showed that the patched eye was strengthened after 2.5 hours of patching. On the other hand, one would expect that the interocular contrast ratio that resulted in a fixed balanced performance (i.e., balance point) should also change as a result of patching in a direction consistent with the above effects for stimuli of the fixed interocular contrast ratio. To check this, we measured the interocular contrast needed to achieve a fixed sensory balance for the global coherence paradigm using the quick procedure previously validated (Black et al., 2011). The interocular contrast ratios (patched eye/unpatched eye) for a fixed balanced task performance (i.e., effective contrast ratio at balance point) in different time sessions are shown in Figure 4 for two subjects. Indeed, 2.5 hours of diffuser patching strengthened the patched eye in which the effective contrast ratio at balance point decreased after patching and returned to the baseline at around 30 minutes post occlusion. 

Figure 4

 

The effect of 2.5 hours of monocular occlusion with a diffuser on the effective contrast ratio at the balance point in the dichoptic global motion coherence task. (A) Experimental processing. (B) The effective contrast ratio at balance point (patched eye/unpatched eye) at different time spots of two subjects. Error bars represent standard errors.

In Figure 5, we show the effect of patching with a diffuser for the monocular contrast threshold. The spatial frequency used corresponded to that used in the phase combination and contrast matching tasks (i.e., 0.3 c/d). Two and half hours of diffuser patching shifted the patched eye's psychometric function at this low spatial frequency to the leftward in each of the two subjects, indicating a decrease in contrast threshold (Figure 5B, above panels). The contrast threshold (corresponding to 81.6% accuracy) decreased from 1.66 ± 0.087 (mean ± SD, variance was estimated from 500 times' bootstrap simulation) to 1.47 ± 0.085 (mean ± SD) in observer N1 and from 1.39 ± 0.081 (mean ± SD) to 0.97 ± 0.082 (mean ± SD) in observer N2. A right-tailed Z test showed that the decrease in contrast threshold was significant in both N1 (Z score = 1.52, p = 0.06) and N2 (Z score = 3.67, p = 0.0001). No significant change of the slope of the psychometric function was found in these two subjects (p > 0.2). While for the unpatched eye, 2.5 hours of diffuser patching shifted the psychometric function to the rightward in both N1 and N2 (Figure 5B, below panels), indicating an increase in contrast threshold. The contrast threshold (corresponding to 81.6% accuracy) increased from 1.63 ± 0.082 (mean ± SD) to 2.04 ± 0.088 (mean ± SD) in observer N1 and from 1.06 ± 0.069 (mean ± SD) to 1.77 ± 0.12 (mean ± SD) in observer N2. A left-tailed Z test showed that the increase in contrast threshold was significant in both N1 (Z score = −3.42, p = 0.0003) and N2 (Z score = −5.25, p < 0.00001). Again, no significant change of the slope of the psychometric function was found in both subjects (p > 0.4). A summary for the changing of contrast threshold in subjects N1 and N2 is provided in Figure 5C. 

Figure 5

 

The effect of 2.5 hours of monocular occlusion with a diffuser on the two eyes' contrast threshold at 0.3 c/d. (A) Experimental processing. (B) Probability of correct identification at different contrast levels of two subjects (N1 and N2) before and after patching for the patched eye (above panels) and unpatched eye (below panels). (C) Contrast threshold (corresponding to 81.6% accuracy) in the unpatched eye (○) and patched eye (□) of two subjects (N1 and N2) before and after patching. Error bars represent variance estimated from 500 times' bootstrap simulation. #, 0.05 < p < 0.1; *, p < 0.001.

Discussion

It was previously demonstrated that a short period of monocular deprivation with a diffuser that excludes all pattern information without affecting the mean luminance resulted in strengthening the contribution of the patched eye to binocular rivalry and dichoptic contrast matches (Lunghi et al., 2011). We were keen to know if the effects generalize to what have become more accepted means of quantifying sensory dominance, namely using measures of binocular phase combination, dichoptic global motion coherence, and binocular contrast matching. Because these measures reflect different aspects and possibly different sites of cortical processing (Zhou, Huang, & Hess, 2013), we hoped to establish how general the effects were. Our results confirm that the novel findings Lunghi et al. (2011) reported for binocular rivalry are also reflected in a range of different visual functions, including local phase, local contrast, and global motion. In all cases (i.e., phase, motion, and contrast, all measured under binocular conditions), the eye that was deprived had its subsequent contribution to the binocular percept strengthened. We show this for stimuli whose interocular contrast is fixed in the way they change the sensory balance, and we also show it for stimuli whose interocular contrast is varied to maintain a fixed binocular sensory balance. The effect had a similar timescale for all three functions, and we conclude that it is a general phenomenon reflecting processing at a number of distributed cortical sites; for example, the phase and contrast measurements are likely to reflect different aspects of striate function (Huang et al., 2010; Hubel & Wiesel, 1968) while the motion coherence is likely to reflect dorsal extrastriate function (Newsome & Pare, 1988; Simmers, Ledgeway, Hess, & McGraw, 2003). However, until we know more about the underlying physiological basis of each of these tasks, it is impossible to know if the small differences we find between the tasks have any physiological meaning. One possible reason why this effect has not been seen before in a number of previous studies of monocular deprivation (Hubel & Wiesel, 1970; Shibata et al., 2012; Wiesel & Hubel, 1963) is because the duration of this effect is very brief, lasting only 30 minutes at most—well before testing began in previous studies. 

The second question we addressed is whether the results reported by Lunghi et al. (2011) critically depended on the use of a diffuser that excludes pattern information but not mean luminance. Our results for all three visual functions show similar effects (magnitude and time course) for a light-tight occluder and a diffuser, suggesting that the effects are not critically dependent on the mean luminance remaining unchanged during pattern deprivation. Finally, the decrease in contrast threshold of the patched eye and the increase in contrast threshold of the unpatched eye are subtle but significant, indicating a strengthening of the patched eye and a weakening of the unpatched eye occurs. The fact that the monocular contrast threshold effects are reciprocal further strengthens the conclusion that these effects are the result of a binocular-based interaction and not simply due to a monocular-based effect of the patched eye. This finding is not unexpected because in the current models of binocular combination there is no purely monocular processing due to the reciprocal contralateral inhibition that occurs prior to excitatory binocular combination (Ding & Sperling, 2006; Meese, Georgeson, & Baker, 2006). We feel it is likely that both the threshold effects and the suprathreshold effects reflect a common underlying cause. 

One possible common cause could be that short-term patching alters the contrast gain, thereby varying the two eyes' contribution in binocular combination. Binocular combination is now thought of as having both excitatory and inhibitory components. The inhibitory influences reflect an inhibitory input to the contrast gain control of the contralateral eye prior to excitatory combination of the two eyes' inputs, the so-called contrast gain control theory of binocular combination (Ding, Klein, & Levi, 2013a, 2013b; Ding & Sperling, 2006; Huang et al., 2009; Huang et al., 2010; Meese et al., 2006; Meese & Hess, 2004). Thus, the sensory balance is depended on by the internal representation of the monocular input, and any change of the contrast gain control prior to binocular combination will change the sensory balance. It is quite possible that the patching increased the contrast gain of the patched eye by shifting the contrast response function to the left and/or decreased the contrast gain of the unpatched eye by shifting the contrast response function to the right (the so-called intrinsic plasticity, see Turrigiano & Nelson, 2000 for review), thereby strengthening the patched eye's contribution to the binocular percept. This speculation gains some support from our findings in the binocular contrast matching task and the monocular contrast threshold measure, which showed that (a) after removal of the patch, the patched eye needed less contrast to perceptually equate the unpatched eye's contribution to the binocular percept; (b) the time course of the change of apparent contrast matched that of the sensory balance change in the binocular phase combination paradigm and the dichoptic motion coherence paradigm; and (c) by the reciprocal change in monocular contrast thresholds (not psychometric slope) for the patched and unpatched eyes. 

Acknowledgments

This work was supported by a CIHR grant (#53346) to RFH. The authors would like to thank Dr. Anthony Norcia and two anonymous reviewers for the helpful comments and thoughtful suggestions. 

Commercial relationships: none. 

Corresponding author: Jiawei Zhou. 

Email: [e-mail]. 

Address: McGill Vision Research, Department of Ophthalmology, McGill University, Montreal, Canada.