Integration and suppression interact in binocular vision

Abstract

Contingent on stereo compatibility, two images presented dichoptically can lead to either binocular integration, thus generating stable stereopsis, or interocular suppression that induces binocular rivalry with bistable perception that alternates between the two images. The relationship between binocular integration and interocular suppression concerns how our brain processes binocular inputs to form unified visual awareness but remains unclear. Here, a series of psychophysical experiments were conducted to address this question, revealing that these collaborative and competitive binocular interactions are interconnected and would mediate one another according to their strength. Specifically, Experiments 1a and 1b showed that the presence of binocular rivalry inhibited peripheral stereopsis, significantly elevating the stereo threshold, with higher elevation resulting from increasing rivalry contrast. Experiments 2a and 2b showed that existing stereopsis with increasing binocular disparity balanced the dynamics of peripheral binocular rivalry, rendering more equivalent eye dominance. Based on these interactions, we suggest that binocular integration and interocular suppression may mediate one another through an overlapping mechanism for regulating eye dominance, with strong stereo percepts tending to reduce eye dominance and strong rivalry tending to increase eye dominance.

Integration and suppression interact in binocular vision

Binocular inputs can engage in both collaborative and competitive binocular processing. Normally, images projected to the binocular retina differ slightly due to the lateral separation between our eyes, and the visual system integrates such images into stable stereopsis based on the disparity of these images. This ability to see a three-dimensional world through binocular integration is remarkable: By requiring only slight binocular disparity, we can integrate two meaningless monocular images consisting of numerous random dots and form meaningful stereopsis (Julesz, 1971). Interestingly, however, if two distinctly incompatible images are presented dichoptically, binocular integration is replaced by interocular suppression, leading to a phenomenon referred to as binocular rivalry, with our perceptual awareness alternating between the two images (Blake, 2022; Wheatstone, 1838). This process of rivalry, characterized by fluctuations in perception despite constant physical stimulation, has been an invaluable tool for exploring neural concomitants of consciousness (Alexander, Panagiotaropoulos, Naotsugu, & Keliris, 2012; Melloni, Mudrik, Pitts, & Koch, 2021; Tong, Meng, & Blake, 2006). Thus, the processing of binocular inputs engages in either binocular integration or interocular suppression, resulting in stable or bistable perceptual awareness. What is the relationship between the two processes? 

A conventional view to addressing this question presumes that integration should be interrupted when inputs from one eye are suppressed during binocular rivalry and thus considers binocular integration and interocular suppression as mutually exclusive. Some studies have indeed provided direct evidence showing the inhibitory effects of interocular suppression on binocular integration (Harrad, McKee, Blake, & Yang, 1994), and vice versa (Blake & Boothroyd, 1985; Blake, Yang, & Wilson, 1991). However, a growing number of studies that used dichoptic stimuli containing both binocular matched and unmatched components have found that binocular rivalry and stereopsis can be observed simultaneously at the same spatial location (Andrews & Holmes, 2011; Buckthought & Wilson, 2007; Su, He, & Ooi, 2009), suggesting that these two processes are not mutually exclusive but independent of each other (Wolfe, 1986; Wolfe & Jeremy, 1988). To reconcile these seemingly contradictory observations, Buckthought and Wilson (2007) proposed that binocular integration and interocular inhibition may interfere with each other within the same processing channel while being independent in different channels. They discovered that rivalry and stereo stimuli with similar spatial frequencies interacted with each other, whereas stimuli with large spatial differences were independent (Buckthought & Wilson, 2007). This compelling experimental evidence indicates that whether binocular integration and interocular suppression are mutually exclusive or independent is affected by their spatial frequency processing channels. 

Here, from another perspective, we wonder whether interactions between binocular integration and interocular suppression are also mediated by their respective strength. If so, mutually exclusive and independent results may represent the extreme cases of progressive interactions between these two processes. By arguing that binocular integration and interocular suppression can both coexist and interact, our hypothesis is compatible with all of the abovementioned previous studies. In a progressive instead of all-or-none manner, we hypothesize that, although binocular rivalry and stereopsis may coexist, they interact contingent on their respective intensities, suggesting a general mechanism that mediates the two processes. 

Furthermore, recent studies have challenged the notion that binocular integration and interocular suppression are opposing processes. By carefully manipulating experimental parameters, Riesen, Norcia, and Gardner (2019) showed that subjects’ perception of Gabor patches with a moderate orientation difference exhibited a tristable state in which fused, left-eye–dominant or right-eye–dominant percepts could all occur over time. Statistical modeling of these perceptual dynamics supported a single-process model with three substates capable of direct competition rather than a separate-process model with stereopsis and rivalry as competing components, indicating that stereopsis and binocular rivalry are not opposed but may be regulated by a shared mechanism. 

Meanwhile, studies in the field of binocular vision have implied that the shared mechanism of binocular rivalry and stereopsis may relate to the modulation of ocular dominance. In binocular processing, a phenomenon referred to as sensory eye dominance can be quantified by diverse tasks, including binocular rivalry (Ooi & He, 2001; Ooi & He, 2020). Studies show that observers have lower stereoacuity in locations with higher sensory eye dominance measured by binocular rivalry (Dieter, Sy, & Blake, 2017); moreover, sensory eye dominance measured through stereo thresholding positively correlates with that measured by examining binocular rivalry (Han, He, & Ooi, 2019). In addition, perceptual training and short-term monocular deprivation can change sensory eye dominance measured by rivalry (Lunghi, Burr, & Morrone, 2011; Lunghi et al., 2019; Xu, He, & Ooi, 2010) and improve stereoacuity (Lunghi et al., 2019; Sauvan et al., 2019; Xu et al., 2010). All of these results imply a shared mechanism underlying binocular rivalry, stereopsis, sensory eye dominance, and binocular plasticity. From the perspective of eye dominance, binocular integration and interocular suppression may interact with each other through an overlapping mechanism related to binocular balance: The presence of binocular rivalry may impair binocular balance and inhibit stereopsis, whereas the presence of stereopsis may facilitate binocular balance and level rivalry dynamics. 

To elucidate the connection between binocular integration and interocular suppression, this study investigated the interactions between adjacent stereopsis and binocular rivalry. In particular, Experiment 1 explored the influence of central binocular rivalry on peripheral stereopsis, and Experiment 2 examined how central stereopsis affects peripheral binocular rivalry. Through the utilization of rivalry stimuli with varying contrast levels in Experiment 1 and stereopsis involving increasing binocular disparity in Experiment 2, we manipulated the strength of interocular suppression(Brascamp, Klink, & Levelt, 2015; Qiu, Caldwell, You, & Mendola, 2020) and binocular integration, aiming to determine whether the interaction between these two processes is contingent upon their strength. By evaluating whether the presence of stereopsis diminishes or balances the intensity of interocular suppression, observable through changes in rivalry switch rate and dominance proportion of each eye, we aimed to clarify whether binocular integration and interocular suppression oppose each other or contribute to a shared mechanism associated with binocular balance. 

Experiment 1: How binocular rivalry affects stereopsis

Experiment 1a

Methods

Participants

Participants included 13 undergraduate or graduate students (seven females, six males; ages 18–25 years); one participant, whose stereo detection accuracy was lower than 0.5 in all conditions, was excluded from the analysis. All participants had normal or corrected-to-normal visual acuity, were naïve to the purpose of the experiment, and provided written informed consent before data collection. This study was approved by the Human Research Ethics Board at South China Normal University (SCNU) and conducted in accordance with the tenets of the Declaration of Helsinki. All participants in this study got paid according to the time spent in the experiment. 

Apparatus

A gamma-corrected 24-inch VIEWPixx CRT screen (100-Hz refresh rate; 1920 × 1080 pixel resolution, 95-cd/m² peak brightness; VPixx Technologies, Saint-Bruno, QC, Canada) was used to present the stimuli. The gamma correction followed a Psychtoolbox program (Brainard, 1997), and the luminance was measured with a Konica Minolta CS-150 luminance and color meter. Each participant viewed the screen through a mirror stereoscope in a darkened room at a distance of 142 cm with their head stabilized on a chin rest. This viewing distance ensured that 1 pixel on our screen corresponded to a 40′′ visual angle. An identical apparatus and viewing distance were used in all experiments in this study. All experimental programs were run using MATLAB (MathWorks, Natick, MA) installed on a Dell PC (Dell, Round Rock, TX). 

Stimuli

Stimuli in all experiments in this study were generated using MATLAB with Psychtoolbox 3.0.18 (Brainard, 1997). In this experiment, we set a central yellow point (0.083° diameter) to help participants maintain fixation and two gray rectangular frames, one small (0.8° × 0.8°) and one large (2° × 2°), centered to aid eye alignment. Rivalry stimuli consisted of two diagonal sinusoidal gratings (spatial frequency, 7.5 c/°; diameter, 0.8°) with a counterbalanced eye-stimulus match across trials. The average of diagonal gratings, which constituted a plaid pattern, was used as the nonrivalry stimulus (see Figure 1B). Both rivalry and nonrivalry stimuli had three levels of Michelson contrast: 0.16, 0.32, and 0.64, defined by  

\begin{eqnarray}{C_m} = \frac{{{L_{max}} - {L_{min}}}}{{{L_{max}} + {L_{min}}}} \quad \end{eqnarray}

(1)

where L_max and L_min indicate the luminance of the brightest and darkest parts of the stimulus, respectively. In each trial, a random-dot stereogram was placed between the two rectangular frames containing a stereo circle (0.4° diameter) at the top, bottom, left, or right of the stereogram. Inside this circle, dots in the right eye had a left shift, leading to crossed disparity. 

Figure 1.

 

Design of Experiment 1, which measured whether binocular rivalry affects peripheral stereopsis. (A) Participants were told to maintain fixation on the central point and to press a corresponding key as soon as they perceived a stereo circle at the top, bottom, left, or right position. (B) Experimental conditions included nonrivalry stimuli as a control condition and three levels of Michelson contrast for the central stimuli. In addition, binocular disparity of the stereogram varied at eight levels: 0′′, 40′′, 80′′, 120′′, 160′′, 200′′, 240′′, and 280′′.

Design and procedure

The main experiment adopted a three-factor within-subjects design, with the three variables being (1) experimental conditions (rivalry vs. nonrivalry), (2) the contrast of central stimuli (low, 0.16; middle, 0.32; high, 0.64), and (3) the binocular disparity of the stereo circle (0′′, 40′′, 80′′, 120′′, 160′′, 200′′, 240′′, and 280′′). Increasing binocular disparity would facilitate stereopsis and elevate participants’ stereo detection accuracy. The stereo threshold was measured based on the stereo detection accuracy. By comparing the stereo threshold with and without the presence of different contrast binocular rivalry, we examined whether binocular rivalry affected stereopsis and, if so, how rivalry contrast would mediate this effect. 

Figure 1A shows the experimental flow chart. In each trial, rivalry stimuli were first presented in the center for 1.5 seconds (plaid patterns were shown to both eyes in the nonrivalry condition). Then, a random-dot stereogram appeared around the central stimuli. In the stereogram, participants may perceive a stereo circle popping out in the top, bottom, left, or right position. They were instructed to maintain fixation on the central point and to press a corresponding key according to the position of the stereo circle as soon as the stereo circle was seen. After the keypress or 1 second, a blank screen was presented for 0.5 second. Except for the 0′′ disparity condition, not pressing a key within 1 second was treated as a wrong response. For our 2 × 3 × 8 = 48 conditions, each condition was repeated 24 times. Thus, each participant completed a total of 1152 trials with a break for every 128 trials completed. 

To obtain rivalry characteristics of different contrasts for correlating rivalry intensity with stereo performance, participants completed six trials of a regular binocular rivalry tracking task after the main experiment. Each trial of this supplementary task lasted for 60 seconds, during which participants were asked to maintain fixation on the central point and to report their perceptual status of binocular rivalry by pressing and holding a corresponding key while releasing all keys when perceiving a mix of the left and right eye stimuli. The rivalry stimuli were identical to the main experiment, with each contrast level repeated twice. 

The whole experiment lasted approximately 1 hour. In both the main and supplementary tasks, different conditions were presented in a random order. 

Data analysis

To obtain psychometric functions of stereo detection accuracy with disparity in different conditions, we fitted our data with logistic psychometric functions for both group and single-subject analyses. For group analysis (see Figure 2A), the data were well fitted in all conditions (R² > 0.99, p < 0.001). For single-subject analysis, 10 participants were successful with curve fitting in all conditions (R² > 0.9, p < 0.05); however, two failed, and their data were excluded from subsequent threshold and correlation analysis. For each participant, despite the 25% correct chance of a random response, the disparity corresponding to 50% accuracy was taken as the stereo threshold in this condition because participants were asked not to respond without seeing a stereo circle. For the data obtained in the supplementary task, we calculated an indicator of the interocular suppression strength, noted as the rivalry switch rate, by dividing the number of exclusive percepts by the total duration of each trial. In this study, all parametric tests were performed on the data that passed the Shapiro–Wilk test, a recommended normality test (Ghasemi & Zahediasl, 2012). 

Figure 2.

 

Effects of binocular rivalry on stereopsis. (A) Fitting curves of stereo detection accuracy with disparity in different conditions. Compared to the three inseparable red curves (nonrivalry condition), the blue curves (rivalry condition) shifted to the right with increasing shift distance resulting from higher rivalry contrast. (B) For each participant, the disparity corresponding to 50% accuracy was taken as the stereo threshold in this condition. The average stereo detection thresholds in different conditions are shown. Error bars represent ±1 standard error. The data for each individual are shown in Supplementary Figure S1.

Results

We found that the presence of binocular rivalry inhibited stereopsis with greater inhibition resulting from higher rivalry contrast. Figure 2A shows the fitting curves of stereo detection accuracy as a function of disparity in different conditions. Compared with the three inseparable curves corresponding to nonrivalry conditions, the curves corresponding to rivalry conditions shifted to the right when rivalry contrast increased; that is, the shift distance increased with higher rivalry contrast. This indicated that, although binocular rivalry and stereopsis could coexist, the presence of binocular rivalry, particularly with higher contrast, impaired stereo detection. 

Figure 2B shows the average stereo detection threshold in different conditions. A two-way ANOVA revealed a highly significant interaction between rivalry/nonrivalry conditions and contrast levels: F(2, 18) = 41.288, p < 0.001, η²_p = 0.821. Further simple effect analysis indicated that, in the nonrivalry condition, there were no significant differences among the stereo thresholds at different contrast levels; for low versus high, t(9) = −1.821, p = 0.102, and Cohen's d = 0.58. In the rivalry condition, the stereo thresholds were significantly elevated with increasing rivalry contrast: for low < middle, t(9) = 5.080, p < 0.001, Cohen's d = 1.61; for middle < high, t(9) = 6.995, p < 0.001, Cohen's d = 2.21. In addition, when the contrast level was low, no significant difference was found between the stereo detection threshold in the rivalry and nonrivalry conditions, t(9) = 1.470, p = 0.176, Cohen's d = 0.46. With increasing stimulus contrast, the difference between the rivalry and nonrivalry conditions became larger and significant: for middle contrast, t(9) = 8.356, p < 0.001, and Cohen's d = 2.64; for high contrast, t(9) = 10.067, p < 0.001, and Cohen's d = 3.18. These results indicated that the presence of binocular rivalry would inhibit stereopsis, elevating participants’ stereo threshold for peripheral stimuli, with a greater effect resulting from higher rivalry contrast. 

To further analyze how the intensity of interocular inhibition modulated its effect on stereopsis, we calculated the repeated measures correlation (Bakdash & Marusich, 2017) between stereo detection performance and the rivalry switch rate obtained in the supplementary task (see Figure 3). We found a significant positive correlation between the rivalry switch rate and stereo detection threshold: r_rm(19) = 0.51; 95% confidence interval (CI), 0.07–0.78; p = 0.018. No significant correlation was found between the rivalry switch rate and stereo reaction time: r_rm(19) = 0.10; 95% CI, −0.38 to 0.53; p = 0.682. The correlation results indicated that, with increasing rivalry contrast, individuals who had greater increases in rivalry intensity (reflected in the rivalry switch rate) in the supplemental experiment also had greater stereo threshold elevation in the main experiment. Such a covariant relationship supported a partial, progressive effect of interocular suppression on stereopsis formation but not on the time required to generate stereopsis, with the latter deduction limited by the lack of fine-grained manipulation of time in the current study. 

Figure 3.

 

Repeated measures correlation between the rivalry switch rate and stereo detection performance. To assess the strength of interocular suppression, participants completed a supplementary task involving 60 seconds of rivalry tracking. Based on the intermediate variable of rivalry contrast, we conducted within-individual correlations between the rivalry switch rate and stereo performance in the main task. The results showed that rivalry switch rates positively correlated with stereo detection thresholds (A) but did not significantly correlate with stereo reaction time (B). These results support a partial, progressive effect of interocular suppression on stereopsis formation but not on the time required to generate stereopsis. Different participants are shown in different colors.

Experiment 1b

Experiment 1b aimed to rule out the possibility that lower stereo detection accuracy in the presence of binocular rivalry found in Experiment 1a may have resulted from possible perceptual switches and involuntary attention shifts caused by rivalry stimuli. In Experiment 1b, participants performed a similar detection task, with the difference being that the circle popping out was defined by Gaussian-distributed luminance enhancement within a circular region instead of binocular disparity. If the presence of binocular rivalry solely affects the detection of stereo circles but not the detection of luminance enhancement circles, it strongly supports that the reduced detection accuracy in the presence of binocular rivalry in Experiment 1a was indeed a result of the negative impact of interocular suppression on binocular integration. 

Methods

Participants

Twelve undergraduate or graduate students (nine females, three males; ages 19–26 years), including the first author, participated in this experiment. All participants had normal or corrected-to-normal visual acuity and, except for the author, were naïve to the purpose of the experiment. Participants provided written informed consent before data collection. 

Stimuli and procedure

In Experiment 1b, the circle popping out was defined by Gaussian-distributed (sigma = 0.1°) luminance enhancement within a circular region (0.4° diameter) in both eyes (see Figure 4). The peak of luminance enhancement varied at eight levels: 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, and 0.7, corresponding to eight levels of binocular disparity in Experiment 1a. The experimental procedure of Experiment 1b followed that for the main task in Experiment 1a. 

Figure 4.

 

Stimuli of Experiment 1b. In Experiment 1b, the participants’ task was similar to that of Experiment 1a: Respond to whether a circle popped out in the top, bottom, left, or right position. The difference is that the circle in Experiment 1a was defined by binocular disparity with its perception relying on binocular integration, whereas the circle in Experiment 1b was defined by Gaussian-distributed luminance enhancement and did not rely on binocular integration.

Data analysis

Similar to Experiment 1a, we fitted the data for both group and single-subject using logistic psychometric functions (for group: R² > 0.99, p < 0.001; for single-subject: R² > 0.90, p < 0.05). For each participant, the luminance enhancement corresponding to 50% accuracy was taken as the detection threshold in this condition. 

Results

We found that the presence of binocular rivalry did not affect the detection accuracy of a circle defined by luminance enhancement (see Figure 5). A two-way ANOVA showed no significant interactions between the rivalry/nonrivalry conditions and contrast levels, F(2, 22) = 0.752, p = 0.483, η²_p = 0.064, nor a significant main effect between the rivalry and nonrivalry conditions, F(2, 22) = 1.308, p = 0.277, η²_p = 0.106. These results indicate that, compared to nonrivalry stimuli, the presence of rivalry stimuli could affect the detection accuracy of a peripheral circle only when its perception relies on binocular integration. 

Figure 5.

 

No impact of binocular rivalry on non-stereo detection. The presence of binocular rivalry, regardless of contrast, does not affect the detection accuracy of a circle defined by luminance enhancement. Therefore, the reduced stereo detection accuracy in the presence of binocular rivalry in Experiment 1a was unlikely due to more attention attracted by rivalry stimuli but indeed a result of the negative impact of interocular suppression on binocular integration. The data for each individual are shown in Supplementary Figure S2.

Experiment 2: How stereopsis affects binocular rivalry

Experiment 2a

Methods

Participants

Twenty-four undergraduate or graduate students (14 females, 10 males; ages 18–24 years), including the first author, participated in this experiment. All participants had normal or corrected-to-normal visual acuity and, except for the author, were naïve to the purpose of the experiment. Participants provided written informed consent before data collection. One participant was excluded from further analysis due to constant mixed perceptions in some trials of binocular rivalry tracking. 

Stimuli

A central yellow point (0.111° diameter) was presented to help participants maintain fixation. In addition, two centered gray rectangular frames, one small (0.8° × 0.8°) and one large (2° × 2°), were shown to aid binocular alignment. A random-dot stereogram was placed inside the small rectangular frame. A stereo circle (0.7° diameter) was contained in the middle of the stereogram. Inside this circle, dots shown to the left eye were shifted to the right, and dots shown to the right eye were shifted to the left, leading to crossed disparity. Rivalry stimuli consisted of a red radial sinusoidal grating (Michelson contrast, 0.32; spatial frequency, 7.5 c/°; diameter, 0.6°) and a green circular sinusoidal grating (Michelson contrast, 0.32; spatial frequency, 11 c/°; diameter, 0.6°) with counterbalanced eye-stimulus matching across trials. 

Design and procedure

The main experiment adopted a one-factor, within-subjects design, including five conditions for the central stereo stimuli: (1) fixation only condition (i.e., no random-dot stereogram contained); (2) a random-dot stereogram with 0′′ disparity; (3) a random-dot stereogram with 80′′ disparity; (4) a random-dot stereogram with 160′′ disparity; and (5) a random-dot stereogram with 240′′ disparity. Because increasing binocular disparity would facilitate stereopsis in the center, we focused on examining whether the presence of stereo stimuli with increasing disparity would affect peripheral binocular rivalry. Specifically, we examined the effects on the rivalry switch rate and the dominance ratio of each eye, two metrics that represent the strength and balance of interocular inhibition, respectively. 

Figure 6 shows the experimental procedure. In each trial, stereo stimuli surrounded by four red radial gratings were first presented for 2 seconds. Then, stimuli for binocular rivalry were displayed in the top, bottom, left, or right position lasting for 60 seconds. In most cases, participants first perceived the green grating during rivalry due to the previously shown red radial gratings (an adaptation effect). Participants were told to maintain fixation on the central point and to report their percepts by pressing and holding corresponding keys while releasing all keys when perceiving a mixture of both gratings. After completion of the binocular rivalry display, a blank screen was presented for 5 seconds. Participants were allowed to rest after each trial. Each participant completed a total of 40 trials in a random order with eight repetitions of each experimental condition. 

Figure 6.

 

Flow chart of Experiment 2a, which examined whether stereopsis affects peripheral binocular rivalry. Participants were told to maintain fixation on the central point and to report their percepts of the gratings by pressing and holding the corresponding keys. Red gratings presented before rivalry ensured that the perception of rivalry started with a green grating for most trials (an adaptation effect). There were five conditions for the central stereo stimuli: (1) fixation only, (2) binocular disparity of 0′′, (3) binocular disparity of 80′′, (4) binocular disparity of 160′′, and (5) binocular disparity of 240′′.

To confirm whether increased disparity in our experimental range indeed facilitated stereopsis, participants completed 120 trials of a stereo detection task before the main task. In this task, a random-dot stereogram with four conditions (disparity of 0′′, 80′′, 160′′, or 240′′) was presented in the center of the screen after a 1-second presentation of a fixation point. Participants were asked to press the down arrow key as soon as they perceived a stereo circle. After the keypress, or 1 second, a 0.5-second blank screen followed. Except for the 0′′ disparity condition, not pressing the key within 1 second was considered an incorrect response. 

Data analysis

During binocular rivalry, the key presses/releases lasting less than 300 ms (for this experiment and for Experiment 2b) more likely reflected errant key presses and were excluded from further analysis (Dieter et al., 2017). We calculated a dominant index to represent the balance degree of rivalry dynamics, specifically to what extent a dominant eye predominated in rivalry dynamics, as follows:  

\begin{eqnarray} && Dominant\ index = \ \frac{{Dominan{t_{prop}} - Nondominan{t_{prop}}}}{{Dominan{t_{prop}} + Nondominan{t_{prop}}}} \quad \end{eqnarray}

(2)

where Dominant_prop and Nondominant_prop denote the dominance time proportions of the dominant eye and nondominant eye, respectively. For trials in the same control condition (control conditions: four locations × two eye-stimulus matches), the dominant eye was considered the eye that had a higher average dominance time proportion across these trials. The rivalry switch rate was calculated in the same way as in Experiment 1. We normalized both the dominance index and rivalry switch rate within the same control condition by using Z-score transformation. This approach enabled us to focus on the variation of these values under different experimental conditions while addressing the problem of large differences in these values across control conditions and between individuals. 

Results

The results from the supplementary task showed that stereo detection accuracy significantly improved with increasing disparity: χ²(2) = 39.263, p < 0.001 (Friedman test was used instead of ANOVA because the data distribution failed the normality test; see Figure 7A). More importantly, we found that the presence of stereopsis with increasing binocular disparity lowered the dominance index and caused a more balanced rivalry between eyes. Figure 7B shows the Z-score of the rivalry switch rates and dominant indexes in different conditions. The dominant index Z-score decreased with increasing disparity. We successfully fitted this trend using a quadratic function: y = (−4.1x² − 12.9x) × 10⁻⁶ + 0.065 (R² = 0.998, p = 0.045). The results of two paired-samples t-tests showed the following: (1) In the presence of stable stereopsis, the dominant index Z-score significantly decreased: 240′′ disparity < 0′′ disparity, t(22) = −2.133, p = 0.044, Cohen's d = 0.44. (2) No significant difference was found between 0′′ disparity and fixation-only conditions, t(22) = −0.428, p = 0.673, Cohen's d = 0.09, indicating that merely adding binocular matching stimuli without stereopsis did not affect rivalry dynamics. For the rivalry switch rate, though, its average Z-score increased with increasing disparity. We found no significant difference of the rivalry switch rate Z-score between 240′′ disparity and 0′′ disparity conditions, t(22) = 0.574, p = 0.571, Cohen's d = 0.12, nor between 0′′ disparity and fixation-only conditions, t(22) = 1.573, p = 0.130, Cohen's d = 0.33. These results indicate that the presence of stereopsis did not significantly affect the switch rate of binocular rivalry. 

Figure 7.

 

Effects of stereopsis on peripheral rivalry. (A) Stereo detection accuracy significantly improved with increasing disparity. (B) The dominance index and the rivalry switch rate reflect the balance and intensity of binocular rivalry, respectively. A smaller balance index indicates more balanced binocular rivalry. We found that the presence of strong stereopsis significantly decreased the dominant index Z-core: for 240′′ disparity < 0′′ disparity, t(22) = −2.133, p = 0.044, Cohen's d = 0.44. But, it did not have a significant effect on the rivalry switch rate Z-score: for 240′′ disparity versus 0′′ disparity, t(22) = 0.574, p = 0.571, and Cohen's d = 0.12. These results indicate that existing stereopsis leads to a more balanced binocular rivalry. Error bars represent ±1 standard error. The data for each individual are shown in Supplementary Figure S3.

Experiment 2b

Experiment 2a showed that the presence of stereopsis, induced by random-dot stereograms with crossed disparity, could balance peripheral rivalry dynamics. Experiment 2b further examined whether this effect could be replicated and generalized when stereopsis was induced by common black rings with uncrossed disparity. Additionally, in Experiment 2b, we manipulated the balance degree of binocular rivalry by using rivalry stimuli with interocular contrast differences. We hypothesized that the balancing effect of stereopsis on rivalry dynamics may be modulated by the original balance degree of binocular rivalry. After all, only an uneven rivalry can become more balanced. 

Method

Participants

Eight undergraduate or graduate students (seven females, one male; ages 20–24 years), including the first author, participated in this experiment. All participants had normal or corrected-to-normal visual acuity and, except for the author, were naive to the purpose of the experiment. Participants provided written informed consent before data collection. One participant was excluded from further analysis due to constant mixed perceptions in some trials of binocular rivalry tracking. 

Stimuli

A central cross consisting of two crossed rectangles (0.022° × 0.111°) was presented to help participants maintain fixation. In addition, a checkerboard frame (outer and inner dimensions: 3.2° × 3.2° and 2.8° × 2.8°) was shown against a gray background to aid binocular alignment. A black ring (0.3° diameter) was presented at the center of both eyes. In the stereo condition, the ring presented to the left eye was shifted left, and the ring shown to the right eye was shifted right, leading to uncrossed disparity. Rivalry stimuli consisted of two diagonal sinusoidal gratings (spatial frequency, 3 c/°; diameter, 1.5°). The Michelson contrast (log unit) of rivalry stimuli presented in the left and right eyes varied at five levels: (1) 1.6:1.8; (2) 1.65:1.75; (3) 1.7:1.7; (4) 1.75:1.65; and (5) 1.8:1.6. Note that a 1.7 log unit contrast corresponded to a 10^1.7 = 50.12% contrast. 

Design and procedure

The main experiment adopted a two-factor, within-subjects design, with the two variables being (1) stereo conditions (240′′ disparity vs. 0′′ disparity), and (2) interocular contrast difference of the rivalry stimuli between the left eye and right eye (−0.2, −0.1, 0, 0.1, and 0.2 log units). By using rivalry stimuli with different interocular contrast differences, we manipulated the balance degree of binocular rivalry. We examined whether the effect of stereopsis on binocular rivalry was modulated by balance degree of rivalry. 

Figure 8 shows the experimental procedure. Participants started each trial by pressing the space key. After the keypress, a black ring around a fixation cross was presented at the center of both eyes. Participants were asked to press a corresponding key to determine whether the central black ring was deeper than the fixation cross. Then, the color of the fixation cross turned red, and 1.5 seconds later diagonal gratings around the black ring were presented to different eyes to induce binocular rivalry. Participants were instructed to maintain fixation on the central cross and to report their perceptual status of binocular rivalry by pressing and holding a corresponding key while releasing all keys when perceiving a mix of the left and right eye stimuli. The rivalry lasted for at least 60 seconds until the end of the final exclusive percept. After that, a blank screen was presented for 1 second, followed by binocular matching plaid patterns for 5 seconds to prevent rivalry dynamics from changing in continuous presentations (Klink, Brascamp, Blake, & van Wezel, 2010). Participants were allowed to rest after each trial. For our 2 × 5 = 10 conditions, each condition was repeated four times in a random order. Each participant completed a total of 40 trials in two sessions separated by at least half a day. 

Figure 8.

 

Flow chart of Experiment 2b. In Experiment 2b, participants were first asked to press a corresponding key to determine whether the central black ring was deeper than the fixation cross, then they were required to maintain fixation on the central cross and report their percepts of the rivalry gratings by pressing and holding the corresponding keys. In the stereo condition, the central black rings had 240′′ uncrossed disparity between eyes to trigger stereopsis. The interocular contrast difference of rivalry stimuli between eyes varied at five levels: −0.2, −0.1, 0, 0.1, and 0.2 (log units).

To confirm whether the black ring with 240′′ disparity indeed caused stable stereopsis, participants completed 100 trials of a stereo detection task before the main task. In this task, dichoptic black rings with 0′′ or 240′′ disparity were presented after the 1-second presentation of a fixation cross. Participants were asked to press a corresponding key to determine whether the ring was deeper than the fixation cross. After the keypress, or 2 seconds, a 0.5-second blank screen followed. In the 240′′ disparity condition, no keypress within 2 s was considered an incorrect response. 

Data analysis

We calculated the balance index (BI) to represent the balance degree of the two eyes in rivalry dynamics as follows:  

\begin{eqnarray}Balance\ index = \ \frac{{Righ{t_{prop}} - {Left_{prop}}}}{{Righ{t_{prop}} + {Left_{prop}}}} \quad \end{eqnarray}

(3)

where Left_prop and Right_prop denote the dominance time proportions of the left eye and right eye, respectively. The values of the balance indexes in each trial ranged from −1 to 1, with a value closer to −1 indicating a predominance of the left eye and closer to 1 indicating a predominance of the right eye. Each participant had four repetitions in each experimental condition. We calculated the median rather than the mean of these four repetitions to avoid the disturbance of extreme values. Then, we defined the stereo effect as:  

\begin{eqnarray}Stereo\ {effect} = B{I_{stereo}} - B{I_{non - stereo}} \quad \end{eqnarray}

(4)

where BI_stereo and BI_non-stereo are the balance index median in the stereo and non-stereo conditions, respectively. In a Pearson correlation between Stereo effect and BI_non-stereo, three outlier data points with BI_non-stereo outside ±0.8 were excluded. 

Results

All participants had an accuracy rate greater than 0.9 for detecting stereo stimuli in the supplementary task, indicating that black rings with 240′′ disparity could lead to stable stereopsis. Importantly, we found that the presence of stereopsis, induced by common black rings with uncrossed disparity, can also balance rivalry dynamics, and this effect is contingent upon the initial balance degree of binocular rivalry in the absence of stereopsis. At the individual level, Figure 9A shows the median BI in different conditions for each participant. For each individual, as the contrast of the rivalry stimuli varied from strong left-eye versus weak right-eye to strong right-eye versus weak left-eye, the dominant eye of binocular rivalry also shifted from the left eye to the right eye, with the BI value changing from negative to positive. Notably, the results for certain participants (particularly S1, S3, and S4) demonstrated that, as the BI changed from negative to positive, the direction of the stereo effect between two eyes reversed, making the BI closer to 0 in the stereo condition. In other words, regardless of which eye predominated in binocular rivalry, the presence of stereopsis led to more balanced rivalry dynamics. If stereopsis had no effect on binocular rivalry, the difference of BI between the stereo and non-stereo conditions should be random, without any discernible pattern or reversals associated with ocular dominance in the absence of stereopsis. 

Figure 9.

 

Effects of stereopsis on binocular rivalry with an interocular contrast difference. (A) The horizontal axis labels NS and S represent the non-stereo and stereo conditions, respectively. The vertical axis label balance index (BI) reflects whether the left or right eye predominates in rivalry dynamics. The value of the BI ranges from −1 to 1, with the value closer to −1 indicating a predominance of the left eye and closer to 1 indicating a predominance of the right eye. Interestingly, the BI shifting from NS to S in the same contrast condition tends to approach zero, regardless of whether the left or right eye predominates (especially S1, S3, and S4). This trend indicates a balancing effect of stereopsis on rivalry dynamics at the individual level. (B) Stereo effect = BI_stereo – BI_non-stereo. A significant negative correlation was observed between the stereo effect and the balance index in the non-stereo condition, r(30) = −0.41, p = 0.019. This finding suggests that the impact of stereopsis on eye dominance was opposite to its initial state, resulting in more balanced rivalry dynamics.

At the group level, we tested whether the stereo effect was modulated by the balance degree of rivalry through a correlation analysis between the stereo effect and balance index in the non-stereo condition. The result showed a significant negative correlation between the stereo effect and non-stereo BI, r(30) = −0.41, p = 0.019, indicating that the stereo effect on eye dominance was opposite to its original state and led to more balanced rivalry dynamics. 

General discussion

Our study investigated the relationship between binocular integration and interocular suppression. Previous research on this issue has yielded conflicting findings, with some studies suggesting that these processes are mutually exclusive (Blake & Boothroyd, 1985; Blake et al., 1991), whereas others propose that they are independent of each other (Andrews & Holmes, 2011; Buckthought & Wilson, 2007; Su et al., 2009). In our study, we found evidence of progressive interactions between stereopsis and binocular rivalry: (1) the presence of binocular rivalry inhibited stereopsis with increasing rivalry contrast, and (2) existing stereopsis leveled the dynamics of binocular rivalry as binocular disparity increased. By demonstrating these progressive interactions, we reconcile the contradictory results of previous studies and propose that mutual exclusivity and independence may represent the extreme cases of the relationship between binocular integration and interocular suppression. Furthermore, our findings suggest that, whether binocular rivalry and stereopsis can coexist is not constant but contingent on their respective strength. Specifically, stereopsis generated from moderate binocular disparity may survive shallow interocular suppression but would be disrupted by intense suppression. 

Our findings support that binocular integration and interocular suppression are not opposing processes but may contribute to a shared mechanism related to binocular balance. Regarding the impact of rivalry on stereopsis, we observed that the presence of binocular rivalry inhibited stereopsis, leading to lower stereo detection accuracy. This indicates that interocular suppression does indeed impair binocular integration, a result also found in other studies (Buckthought & Wilson, 2007; Harrad et al., 1994). However, instead of simply impairing interocular suppression, which should have altered the rivalry switch rate, we surprisingly found that existing stereopsis levels the rivalry dynamics, rendering more equivalent eye dominance. Considering that the direction and strength of interocular suppression change over time during binocular rivalry (Alais, Cass, O'Shea, & Blake, 2010), this balancing effect of stereopsis on rivalry dynamics suggests that binocular integration promotes a more symmetrical interocular suppression between the two eyes rather than diminishing the overall strength of interocular suppression. The balancing effect of binocular integration on interocular suppression also supports that these two may interact with each other through an overlapping mechanism related to binocular balance: On the one hand, interocular suppression impairs binocular balance and inhibits stereopsis; on the other hand, binocular integration maintains binocular balance and levels rivalry dynamics. Notably, such mechanisms highly align with a recent modeling result that suggests that stereopsis and rivalry are not competing components but may belong to a single process (Riesen et al., 2019). 

Our findings revealed interactions between binocular integration and interocular suppression in nearby spatial locations, providing insights into the mechanisms involved in both processes and the development of general models of binocular rivalry and stereopsis. Interocular suppression in binocular rivalry is known to extend beyond the immediate location of conflict (Blake, 2001). This characteristic of interocular suppression forms the foundation for the observed influence of binocular rivalry on peripheral stereopsis. Interestingly, we also found that existing stereopsis can impact peripheral binocular rivalry, suggesting that the influence of binocular integration on binocular balance (Hou, Huang, Liang, Zhou, & Lu, 2013) can extend to adjacent locations. This sheds light on how the visual system can achieve a stable and unified binocular vision despite persistent biases in eye dominance within the visual field (Dieter et al., 2017). Moreover, our results challenge existing general models of binocular rivalry and stereopsis (Hayashi, Maeda, Shimojo, & Tachi, 2004; Riesen et al., 2019; Wilson, 2017), which have predominantly focused on the relationship between stereopsis and binocular rivalry within the same spatial location. Our study underscores the necessity of considering not only their interactions within the same space (Buckthought, Kim, & Wilson, 2008; Buckthought & Wilson, 2007) but also the mechanisms of their interactions in different spatial locations to develop a more comprehensive model that can account for our findings. 

The relationship between stable binocular stereopsis and bistable binocular rivalry underscores other fundamental visual mechanisms, including sensory eye dominance and binocular plasticity. Recent studies have implied tight connections among stereopsis, binocular rivalry, sensory eye dominance, and binocular plasticity (Dieter et al., 2017; Han et al., 2019; Lunghi et al., 2011; Sauvan et al., 2019; Xu et al., 2010). Given that top–down attention and past experience are well known to bias representation strength, it becomes conceivable through the strength regulation mechanism that past experiences, such as perceptual training and short-term monocular deprivation, may change sensory eye dominance that is measured by binocular rivalry (Lunghi et al., 2011; Xu et al., 2010) and thus potentially improve stereoacuity (Sauvan et al., 2019; Xu et al., 2010). Indeed, our findings that stereopsis affects the dominance ratio between the two eyes in rivalry dynamics indicate that sensory eye dominance may be modulated in the presence of stereopsis. Although many studies addressing binocular plasticity of sensory eye dominance have primarily focused on interocular suppression processes, including continuous binocular rivalry (Klink et al., 2010) or continuous flash suppression (Kim, Kim, & Blake, 2017) and short-term monocular deprivation (Lunghi et al., 2011; Lunghi et al., 2019), our findings suggest a novel possibility that such binocular plasticity may also be triggered through binocular integration. Future studies can further systematically examine whether continuously viewing stereopsis generated from stimuli with interocular contrast differences would also alter sensory eye dominance and affect stereoacuity. If so, then binocular dysfunctions, such as amblyopia, may be ameliorated based on a more natural training strategy that involves daily binocular integration. 

The relationship between stereopsis and binocular rivalry is important for understanding the mechanisms underlying stable and bistable visual awareness in binocular vision. Binocular integration and binocular rivalry lead to stable and bistable visual perceptions, respectively. Our study uncovered a mutual modulation between existing binocular rivalry and stereopsis, indicating an interconnection between the formation of stable and bistable perceptions during bottom–up processing in binocular vision. Given that both bottom–up processing and top–down regulation influence our perception (Gilbert & Sigman, 2007; Intaitė, Noreika, Šoliūnas, & Falter, 2013; Kornmeier, Hein, & Bach, 2009; Theeuwes, 2019; Tong et al., 2006), studies have shown that top–down attention can also modulate both stereopsis (Zou, Liu, & Wolfe, 2022) and binocular rivalry (Meng & Tong, 2004; Paffen, Alais, & Verstraten, 2006; Zhang, Jiang, & He, 2012). However, it remains unanswered whether top–down processes, such as attention, can simultaneously influence both binocular rivalry and stereopsis through the regulation of binocular balance, which we propose as a potential mechanism underlying the connection between binocular integration and interocular suppression. Further investigations are needed to explore this potential interaction and its implications for stable and bistable visual awareness in binocular vision. 

In conclusion, our study reveals how adjacent binocular rivalry and stereopsis interact: (1) Central binocular rivalry with higher contrast elevates peripheral stereo threshold, and (2) central stereopsis with increasing binocular disparity levels peripheral rivalry dynamics. These results support the notion that integration and suppression in binocular vision are not opposing processes, suggesting that dynamic regulation of binocular balance may underlie the mechanism shared between the two processes. 

Acknowledgments

Supported by grants from the STI2030-Major Projects (2021ZD0204200), the Sino-German Center for Research Promotion (M-0705), and the National Natural Science Foundation of China (31871136). 

Commercial relationships: none. 

Corresponding author: Ming Meng. 

Email: [e-mail]. 

Address: School of Psychology, South China Normal University, 55 W Zhongshan Ave, Guangzhou, Guangdong Province, China.