Mask contrast and size do not alter suppression depth in the tracking continuous flash suppression paradigm

Abstract

Continuous flash suppression (CFS) is a variant of interocular conflict that occurs when one eye views a dynamic high-contrast mask that increases the duration of target suppression. A variant of CFS known as tracking continuous flash suppression (tCFS) was developed, allowing the depth of interocular suppression to be measured. Although previous research has measured how the duration of suppression may be modulated by the contrast and size of the masking stimulus, no study has assessed how mask features impact suppression depth. In our first study, we manipulated mask contrast to measure the consequent impact on suppression depth as measured by the tCFS procedure. We observed that high mask contrast increased the threshold required for a target to break into awareness. Critically, the decrease in contrast required to re-suppress each target was proportionately the same across all conditions so that suppression depth—the ratio of the two thresholds—remained constant. In the second experiment, we manipulated the size of the masking stimulus and found no change in breakthrough/suppression thresholds or suppression depth (i.e., the difference between the thresholds when using log-contrast). These findings clarify that, although changes in mask contrast may alter the threshold to enter awareness, there is no overall change in suppression depth as the changes in breakthrough threshold are reflected by proportionately equivalent changes in suppression threshold. This result matches findings obtained with binocular rivalry showing that suppression depth is constant despite changes in stimulus contrast. Differing levels of mask contrast and size, therefore, can be used by researchers in CFS without altering the strength of suppression, consistent with the perspective that interocular suppression operates in small local spatial zones determined by receptive field size in the primary visual cortex.

Introduction

The human visual system is constantly presented with complex stimuli that it must process and interpret. The lateral separation of our eyes provides us with two similar yet offset perspectives of the world. Typically, the resulting visual images are similar and consequentially are fused together, resulting in three-dimensional (3D) stereovision (Anderson, 1998; Blake & Sekuler, 2006). However, when each eye is presented with a unique image, the two stimuli cannot be fused and instead binocular rivalry ensues, and we consciously perceive one stimulus at a time while the other is suppressed (Blake & Alais, 2005; Blake & Logothetis, 2002; Leopold & Logothetis, 1996). One popular example of this phenomenon of interocular suppression, known as continuous flash suppression (CFS), has a static, low-contrast target image viewed in one eye and a dynamic, high-contrast mask (i.e., a Mondrian) presented in the contralateral eye (Pournaghdali & Schwartz, 2020; Tsuchiya & Koch, 2005; Tsuchiya, Koch, Gilroy, & Blake, 2006). The resulting conflict can create a powerful and deep suppression that prevents the target from being seen for up to three minutes (Tsuchiya & Koch, 2005).

Psychophysicists have used CFS as a tool to probe the unconscious visual system by assessing how long it takes for a suppressed stimulus to enter conscious awareness. In a popular version of the paradigm known as breaking CFS (bCFS) (Jiang, Costello, & He, 2007; Stein, Hebart, & Sterzer, 2011), the target stimulus in one eye, suppressed by the Mondrian mask in the other, steadily increases in contrast until it breaks through suppression. An example of the bCFS paradigm is shown in Figure 1A. This bCFS paradigm has proved very popular, and breakthrough times have been claimed, controversially, to be an index of how much an unconscious image is processed without awareness (Moors, 2019; Stein, 2019). Proponents of the unconscious processing perspective argue that certain types of stimuli, such as emotional faces (Yang, Zald, & Blake, 2007), incongruent scenes (Mudrik, Breska, Lamy, & Deouell, 2011), and words (Jiang et al., 2007) break suppression faster than other stimuli. This faster breakthrough is thought to indicate an unconscious preference for the brain to see these stimuli compared with other stimuli.

Figure 1.

Examples of the tCFS paradigm and stimuli used in the experiments. (A) An example of the bCFS paradigm. In this example, the left eye was presented with a dynamic Mondrian masking stimulus, and the right eye was presented with a target. The target gradually increased in RMS contrast from 0% to 100% in log-linear steps until the participant indicated it had broken through into visual awareness. (B) An example of the reCFS paradigm. Here, the contrast of the target was initially 100% RMS and decreased toward 0% until the participant indicated that it had disappeared from awareness. (C, D) An example of data from a tCFS trial, with bCFS thresholds recorded in blue and reCFS thresholds recorded in red (C). The target began as visible at maximum contrast and gradually declined. The participant pressed a key when the target became suppressed from awareness (red dots indicate suppression thresholds), which reversed the direction of contrast change. Eventually, the suppressed target rose to a contrast level where it emerged from suppression and became visible (blue dots represent breakthrough thresholds). Each time the participant indicated a change in perceptual state, the direction of target contrast reversed. The process continued until enough turning points had been obtained to average into reliable means for breakthrough and suppression threshold (here, 20 turning points). This method is a very efficient and fast (∼60 s or so) way to estimate suppression depth compared with (D), the traditional method of comparing dominance and suppression thresholds measured with separate staircases involving hundreds of trials. Suppression depth is shown to the right of the plot. (E) An example of the masking stimuli used in Experiment 1, at low (−32 dB/2.5% RMS), medium (−26 dB/5% RMS), and high (−20 dB/10% RMS) levels respectively. (F) An example of the masking stimuli used in Experiment 2, at small (2°), medium (4°), and large (8°) sizes, respectively.

Notably, this preference was attributed to the emotional or other high-level content of the images rather than the low-level stimulus features. Sceptics argue that these differences in bCFS breakthrough times can instead be attributed to differences in low-level stimulus features that render a target more or less salient to early visual neurons (Gayet, Van der Stigchel, & Paffen, 2014; Moors, 2019; Moors, Boelens, van Overwalle, & Wagemans, 2016; Moors & Hesselmann, 2017; Moors, Hesselmann, Wagemans, & van Ee, 2017).

Recently, Alais, Coorey, Blake, and Davidson (2024) developed a new tracking CFS (tCFS) technique, which demonstrated that differences in bCFS thresholds alone are insufficient to infer the strength of suppression. This technique measured bCFS thresholds for targets of increasing contrast, as in previous experiments, but notably also measured the amount of contrast change required to re-suppress the target (see Figures 1A and 1B). This is achieved by having participants indicate via button press when a target breaks into conscious awareness (as in previous bCFS studies). The target contrast when the participant has indicated a breakthrough is recorded as a bCFS threshold. When the participant has indicated a bCFS event, the contrast of the target image then immediately begins decreasing in contrast until the participant now indicates that it has left conscious awareness. The point of contrast at which the target image disappears is recorded as a re-suppression (reCFS) threshold. A big advantage of the new tCFS method is that it allows easy quantification of the depth of suppression (how strongly the visual system suppresses the target during CFS) by calculating the difference between the breakthrough and reCFS thresholds. It is important to note that we manipulate stimulus contrast on a log scale (in decibels) to perceptually linearize the contrast changes. Thus, calculating the threshold difference is equivalent to the log-ratio of the thresholds when linear contrast is used. The vast majority of CFS studies do not measure suppression depth (for the only exceptions, see Alais et al., 2024; Tsuchiya et al., 2006; Yang & Blake, 2012) even though it was commonly done in binocular rivalry studies (Alais & Melcher, 2007; Blake & Camisa, 1979; Nguyen, Freeman, & Alais, 2003; Watanabe, Paik, & Blake, 2004). Alais et al. (2024) determined that different types of stimuli (such as faces, houses, gratings, and noise patches) exhibited different bCFS thresholds, as others had reported, yet all had the same proportion of suppression depth. This challenges claims based on bCFS thresholds alone that stimuli such as emotional faces (Jiang et al., 2007; Yang et al., 2007), incongruent scenes (Mudrik et al., 2011), and words (Jiang et al., 2007) are prioritized in visual processing.

There is a small but important difference between the CFS suppression depth studies by Tsuchiya et al. (2006) and Yang and Blake (2012) and the tCFS studies reported here and in Alais et al. (2024). In the former studies, the dominance threshold that forms the baseline measurement for suppression depth was not actually measured. Instead, the threshold for a monocular stimulus with no stimulus presented to the other eye (i.e., without a flickering Mondrian in the other eye) was measured and assumed to be the same as a dominance threshold measured with the Mondrian in the other eye. This was based on binocular rivalry studies showing that monocular thresholds and dominance thresholds are the same (Blake & Camisa, 1979; Westendorf, 1989). However, in CFS, this might not be true because in rivalry the dominance threshold is measured with a constant pattern in the untested eye whose response would habituate over time whereas in CFS there is a continuously refreshed high-contrast Mondrian in the untested eye. Any interaction at all between the eyes would likely add noise to the tested eye and raise the dominance threshold above the simple monocular threshold level. If so, using the lower monocular threshold instead of a dominance threshold would artificially increase suppression depth and might explain why Tsuchiya et al. (2006) reported such a high value of suppression depth of 1.4 log units (i.e., 28 dB). In our tCFS method, by contrast, dominance and suppression thresholds are both measured with the dynamic masker in the other eye and likely provide a better suppression depth estimate than using the monocular threshold as a proxy for the dominance threshold. We intend to test whether monocular and dominance thresholds are identical in CFS.

How might altering the properties of the masking stimulus affect suppression depth in the tCFS paradigm? Varying mask characteristics such as orientation or spatial and temporal content does alter bCFS thresholds (Alais & Melcher, 2007; Alais & Parker, 2012; Drewes, Zhu, & Melcher, 2020; Fahle, 1982; Han, Lunghi, & Alais, 2016; Hesselmann, Darcy, Ludwig, & Sterzer, 2016; Hong, 2015; Stuit, Cass, Paffen, & Alais, 2009; Stuit, Cass, Paffen, & Alais, 2023; Webb & Hibbard, 2020; Yang & Blake, 2012; Zhu, Drewes, & Melcher, 2016). Masks with lower temporal frequency of about 1 Hz create a longer suppression duration compared with masks that modulate at a higher temporal frequency (Adams, Gray, Garner, & Graf, 2010; Han et al., 2016; Han, Alais, & Blake, 2018a; Han, Blake, & Alais, 2018b). Similarly, masks that have a lower spatial frequency typically produce stronger suppression (Yang & Blake, 2012). Notably, the size and contrast of the masking stimulus also seem to be important. Han, Lukaszewski, and Alais (2019) manipulated both the size and contrast of the masking stimulus and found that, as the mask contrast increased, so too did the suppression duration, as measured by bCFS thresholds. Han et al. (2019) also compared suppression durations when small targets (radius 1°) were paired with masks of differing sizes (radii ∼ 1°–3°). The effects of mask size were inconsistent and did not have an effect at low mask contrasts. Larger mask sizes were not consistently shown to increase suppression durations, yet larger masks with collinear gratings were shown to reduce suppression depth compared with matched mask and target sizes. This latter effect was consistent with earlier work in binocular rivalry (Paffen, Alais, & Verstraten, 2005; Paffen, te Pas, Kanai, van der Smagt, & Verstraten, 2004) and could be due to surround suppression in the primary visual cortex (Fries, Albus, & Creutzfeldt, 1977; Knierim & van Essen, 1992). Han et al. (2019) interpreted the lack of a clear mask size effect as arising from CFS operating in local spatial zones, consistent with a suppression mechanism based in early visual cortex, as has been proposed for binocular rivalry (Blake, O'Shea, & Mueller, 1992; O'Shea, Sims, & Govan, 1997). Notably, however, their study drew these conclusions based on bCFS breakthrough times—the time taken for a low contrast image to break into awareness. A more rigorous method would be to quantify the effectiveness of a Mondrian mask as a suppressor by measuring suppression strength in terms of contrast. Very few papers, however, have quantified CFS suppression in terms of contrast. One study quantified the effect of mask contrast on suppression thresholds (Ludwig, Kathmann, Sterzer, & Hesselmann, 2015) and found it had a modest effect in that there was an increase in suppression thresholds comparing medium-contrast (70%) and high-contrast (95%) masks, but there was no effect between low and medium contrast and, overall, no monotonic trend. The study did not measure dominance thresholds, so the effect of contrast on suppression depth (the log-ratio of the thresholds) cannot be computed. Based on binocular rivalry studies, however, no effect would be expected because variations in the contrast of the suppressing stimulus does not alter suppression depth, as shown in several studies (Blake & Camisa, 1979; Hollins & Bailey, 1981; Holopigian, 1989).

The other mask variable of interest in this paper, mask size, has never been quantified in terms of contrast for its effect on suppression depth. It is unclear, therefore, whether variations in mask size will vary suppression depth when using tCFS to measure both breakthrough and suppression thresholds in terms of contrast. One possibility, based on the results of Alais et al (2024), is that, although bCFS thresholds may vary with changes in these mask features, reCFS thresholds would change in parallel and thus maintain a constant level of suppression depth. We designed two experiments to test manipulations of mask contrast and mask size on suppression depth, as measured via the tCFS procedure. Measuring suppression depth will reveal a complete picture of how mask size and contrast affects interocular suppression in CFS. Moreover, the thresholds and suppression depth are measured in terms of contrast which, unlike the commonly used bCFS breakthrough time, maps naturally onto low-level visual cortical responses, which are particularly sensitive to contrast (Avidan et al., 2002).

Experiment 1

Experiment 1 manipulated mask contrast over three levels. Because higher contrast drives early visual neurons more strongly, we predict that suppression depth will increase as the masking stimulus increases in contrast due to the mask-driven neurons exerting more inhibition over the target-driven neurons. This will increase the breakthrough threshold of the target and deepen its suppression (greater difference between breakthrough and suppression thresholds).

Methods

Participants

Participants were 22 first- and second-year psychology students (19 female), who participated in this experiment in exchange for course credit, and one of the authors (JC). A pilot study was used conducted to determine the sample size. Using the effect size (${\eta }_p^2$ = 0.311) of the interaction effect between bCFS and reCFS thresholds with mask contrast, a sample size of six was estimated using G*Power. However, more participants were recruited in order to remain consistent with previous tCFS studies (Alais et al., 2024). The experiment was conducted under a pre-existing ethics approval granted by the University of Sydney (protocol number: 2021/048). The experimental procedures and associated costs and risks were explained to participants before informed consent was obtained. Participants were debriefed on the aims and background of the study after completion of the experiment.

Design

Experiment 1 manipulated mask contrast over three levels in a within-subjects repeated-measures design. The mask contrasts were separated by 6 dB: −32, −26, −20 dB, corresponding to 2,5%, 5%, and 10% RMS contrast, respectively. Note that, on a decibel scale of contrast, each 6-dB increment is equivalent to a doubling of contrast on a linear scale, meaning there is a 4× range from lowest to highest contrast. Participants tracked the visibility of the target as its contrast continually changed and reported when perceptual reversals occurred, with each report reversing the sign of the contrast change of the target (see Figure 1). The dependent variable was the contrast at which a perceptual reversal occurred, either when an invisible target became visible as it increased in contrast (breakthrough threshold) or when a visible target became invisible as it declined in contrast (re-suppression threshold).

Variables

The dependent variable in tCFS is target contrast rather than duration. An additional variable of suppression depth is calculated from the threshold difference between breakthrough and suppression.

Apparatus

Visual stimuli were viewed through a mirror stereoscope approximately 57 cm from the display on an Apple LED Cinema Display (24-inch, 1920 × 1200-pixel resolution, 60-Hz refresh rate; Apple, Cupertino, CA). Experiments were programmed using MATLAB R2017b (MathWorks, Natick, MA) and Psychtoolbox 3.0.13 (Brainard, 1997), running on OS X El Capitan (10.11.6) on an Apple 2013 Mac Pro (3.7-GHz Quad-Core Intel Xeon E5) computer. Responses were collected with the right hand as a left-button mouse click.

Stimuli and materials

Participants viewed the stimuli through a mirror stereoscope that subtended targets to approximately 2° × 2° (of visual angle) in their right eye. A Mondrian pattern of size 8° × 8° was presented to their left eye and acted as the suppressing stimulus. Examples of mask stimuli are presented in Figure 1. Stimuli were four naturalistic object images obtained from the THINGS database (Hebart et al., 2019). The four stimuli were a basket, a bowl of almonds, a dandelion, and Easter eggs. Stimuli were modified to be grayscale, consistent with the mask, and were resized to 2° × 2° of visual field. The trial began with the target stimulus visible and declining from maximum contrast. Target contrast changed at a rate of 0.07 dB per video frame, either increasing or decreasing depending on whether the participant's most recent response indicated breakthrough or suppression (see Figure 1). Target stimuli were four images that were matched for root mean squared (RMS) contrast and mean luminance. Images were all naturalistic objects. Stimuli were placed within two square fusion contours, as shown in Figure 1, which allowed participants to properly align and maintain stable fusion of the images.

Procedure

Participants sat in front of a mirror stereoscope and rested on a chin rest. They were then taken through a calibration procedure, to ensure that the fixation squares for stimuli were aligned for consistent interocular conflict. Participants were then presented a Mondrian mask to the left eye and the target to the right eye. Initially, the target was presented at maximum contrast. It would then decrease in contrast according to a logarithmic scale (in order to appear perceptually linear) in steps of 0.07 dB per video frame. Participants were instructed to click the left mouse button as soon as the target disappeared from awareness. When the participant clicked to indicate the disappearance of the target, the direction of contrast change reversed and thus increased until the participant clicked again to indicate that the image had entered into visual awareness. In this experiment, participants were instructed to indicate that a target had disappeared the moment it was completely gone from their visual awareness. For a breakthrough, participants were instructed to click when 75% of the image had reappeared. This criterion was chosen to avoid false-positive indications of a breakthrough, as fractions of a target have been known to appear in visual awareness in bCFS paradigms, even though the target itself is still suppressed (Moors et al., 2017; Zadbood, Lee, & Blake, 2011). Target contrast was recorded at the point at which the participant clicked to indicate that the target had emerged from suppression (bCFS) or had become suppressed again (reCFS). Suppression depth was calculated as the target contrast in decibels for bCFS thresholds minus the contrast for reCFS thresholds. When using log contrast values (i.e., decibels) to define thresholds, calculating suppression depth as the threshold difference is mathematically equivalent to taking the log of the dominance-to-suppression ratio when thresholds are measured in linear contrast, as in some previous suppression depth studies (Alais & Melcher, 2007; Tsuchiya et al. 2006; Yang & Blake, 2012). This procedure continued for three blocks of eight trials, with each trial containing a single image. We recorded 10 bCFS and reCFS thresholds, respectively, per trial per image.

Data cleaning

Two participants were excluded from Experiment 1 due to a failure to comply with instructions and an abnormally fast completion speed. For the remaining participants, the data were cleaned to remove floor and ceiling values. This occurs when target contrast has reached its maximum or minimum value (1.0 or 0.01 RMS, respectively), and the participant has not indicated a response for more than 250 ms. These floor and ceiling responses were omitted from further analyses. It is important to remove these values, as suppression depth is ultimately a measure of contrast threshold. If a target is maintained at the maximum or minimum value for more than 250 ms, we cannot accurately determine the threshold. This problem is largely eliminated in piloting by choosing mask contrasts and target contrast increments that allow the thresholds to be reached within the floor and ceiling limits.

Results

A 2 × 3 repeated-measures ANOVA was conducted on the contrast thresholds: 2 categories of contrast threshold (bCFS and reCFS) × 3 levels of mask contrast. The results are plotted in Figure 2A (CFS thresholds) and Figure 2B (CFS suppression depth). There was a statistically significant main effect of tCFS threshold, such that the threshold for bCFS (M = −13.58 dB, SD = 5.79) was significantly higher than the threshold observed for reCFS (M = −25.54dB, SD = 6.30), F(1, 21) = 361.97, p < 0.001, ${\eta }_p^2$ = 0.945. The main effect of mask contrast was also statistically significant, F(2, 42) = 66.01, p < 0.001, ${\eta }_p^2$ = 0.759. On average, thresholds in the low mask contrast condition (M = −23.62 dB, SD = 4.56) were lower in contrast compared with the thresholds in the medium mask contrast condition (M = −19.61 dB, SD = 5.29), MD = −4.01, t(21) = −6.36, p < 0.001, d = −0.784. Similarly, thresholds in the medium mask contrast condition were lower than in the high mask contrast condition (M = −15.45, SD = 4.67), MD = −4.16, t(22) = −6.05, p < 0.001, d = −0.814. There was no statistically significant interaction effect between tCFS threshold and mask contrast (p = 0.51). Figure 2B plots suppression depth (the difference between the thresholds in Figure 2A). There was no effect of suppression depth varying with mask contrast, and overall average suppression depth was 11.97 dB.

Figure 2.

tCFS data from Experiment 1 acquired by manipulating mask contrast. (A) Plot of contrast thresholds for breakthrough (bCFS, blue) and suppression (reCFS, red) across the three levels of mask contrast. Thresholds rose significantly with increasing mask contrast. Gray lines represent mean thresholds for each participant. Error bars represent 1 standard error of the mean (SEM). (B) Suppression depth (difference between bCFS and reCFS thresholds) for each level of mask contrast showed no effect of mask contrast on suppression depth. Gray dots represent mean suppression depth values for each participant. Error bars represent 1 SEM.

Although we were able to replicate the effect of higher mask contrast being associated with higher bCFS thresholds from Han et al. (2018a), we did not identify any effect of mask contrast on suppression depth. As mask contrast increased, the difference between bCFS and reCFs thresholds remained constant, as shown in Figure 1. This demonstrates that an increase in the strength of the mask does not necessarily correspond to an increase in suppression depth, as suggested by Holopigion (1989). As we have demonstrated the lack of direct relationship between bCFS thresholds and suppression depth, further investigation into manipulations of suppression depth is required. In Experiment 2, we attempted to manipulate the size of the Mondrian mask, as done by Han et al. (2018b), who identified inconsistent effects of this manipulation on bCFS thresholds.

Experiment 2

Experiment 2 manipulated mask size over three levels: 2°, 4°, and 8°. We were motivated to investigate the effect of mask size over a larger range than had previously been tested (e.g., Han et al., 2019), and to use the tCFS paradigm so that the effect of mask size on suppression depth could be measured for the first time, to our knowledge.

Method

Participants

All participants in Experiment 1 also participated in Experiment 2 plus one additional participant, an author of the paper (DA).

Design

Experiment 2 adopted a 2 (target contrast change, increasing vs. decreasing) × 3 (mask size, 2° vs. 4° vs. 8°) within-subjects repeated-measures design. Measurements, dependent variables, materials, and procedures were the same as in Experiment 1. Participants were recruited via The University of Sydney participant pool.

Stimuli

Target stimuli were as in Experiment 1. Mask stimuli were one of three Mondrian masks that varied in size. Mondrian masks were square and subtended 2°, 4°, or 8° of visual angle on a side. Examples are shown in Figure 1. Mask contrast was set to –18 dB (12% RMS).

Data Cleaning

Data were cleaned as in Experiment 1. No participants were excluded from Experiment 2.

Results

A 2 × 3 repeated-measures ANOVA was conducted on the tCFS thresholds: 2 categories of contrast threshold (bCFS and reCFS) × 3 levels of mask size. The CFS thresholds are plotted in Figure 3A, and CFS suppression depth is plotted in Figure 3B. There was a statistically significant main effect of threshold, suggesting that breakthrough (bCFS) thresholds (M = −11.566 dB, SD = 7.10) were significantly higher than re-suppression (reCFS) thresholds (M = −23.19, SD = 7.40), F(1, 22) = 431.84, p < 0.001, ${\eta }_p^2$ = 0.952. There was no statistically significant main effect of mask size (p = 0.479). A statistically significant interaction effect was observed (p = 0.03), F(2, 44) = 3.78, ${\eta }_p^2$ = 0.15. The suppression depth in the small mask size condition (M = 12.23) was significantly greater than in the medium mask condition (M = 10.98), t(22) = 3.45, p = 0.002, d = 0.39. No significant difference in suppression depth was found between the medium mask and the large mask (M = 11.79; p > 0.05). Given the marginal significance of the interaction and the observed difference only between the small and medium mask sizes, a Bayesian repeated-measures ANOVA was conducted for further investigation. The best-fitting model included only the main effect of bCFS versus reCFS (BF₁₀ = 1.496e13). The evidence for the inclusion of the interaction effect was within the range of uncertainty and favored the null (BF_inclusion = 0.81). Considering this evidence, we are reluctant to treat this interaction as a truly significant result. Figure 3B plots suppression depth (the difference between the thresholds in Figure 3A). There was no effect of suppression depth varying with mask size, and the overall average suppression depth was 11.63 dB.

Figure 3.

tCFS data from Experiment 2 acquired by manipulating mask size. (A) Plot of thresholds for breakthrough (bCFS, blue) and suppression (reCFS, red) across different mask size conditions. Gray lines represent mean thresholds for each participant. Error bars represent 1 SEM. (B) Bars represent suppression depth, the difference between the two thresholds. Gray dots represent mean suppression depth values for each participant. Error bars represent 1 SEM.

Discussion

The aim of these experiments was to use the new tCFS method to re-examine the effects of mask size and mask contrast on CFS. These effects have previously been examined using the bCFS method which measures breakthrough times only. With tCFS, we measure the contrast at which the suppressed target (when rising in contrast) breaks into awareness (the bCFS threshold) and the threshold at which the visible target (when falling in contrast) disappears from awareness (reCFS threshold). Quantifying target thresholds in terms of contrast instead of a reaction time has two advantages. First, it is easier to relate the results to cortical activation and inhibited response levels. Second, the difference between the bCFS and reCFS thresholds provides the important index of suppression depth which is seldom measured in CFS studies.

It was noted above that the vast majority of CFS studies do not measure suppression depth and that the only exceptions are Tsuchiya et al. (2006), Yang and Blake (2012), and Alais et al. (2024). In essence, suppression depth expresses the ratio of threshold elevation under interocular suppression. An increment threshold is first measured when the test stimulus is visible to establish the baseline visibility of the stimulus (the dominance threshold in binocular rivalry, or what we term the reCFS threshold in the tCFS method). Then an increment threshold is measured when the stimulus is suppressed from awareness. There will inevitably be a threshold elevation in the suppression condition that will typically be several times higher than the dominance baseline. If thresholds are defined on a linear contrast scale, as in Tsuchiya et al. (2006) and Yang and Blake (2012), suppression depth is defined as the logarithm of the threshold elevation ratio. If the contrast thresholds are measured on a logarithmic contrast scale, as in our tCFS method, then the equivalent suppression depth value is given by the difference between the thresholds rather than the ratio.

In Experiment 1, we observed a significant main effect of mask contrast such that, as mask contrast increased, so too did mean bCFS and reCFS thresholds. This finding replicates the results of Han et al. (2019), who found increased bCFS thresholds as mask contrast increased (as measured by normalized bCFS durations). Our results go further by also measuring suppression thresholds (reCFS) so that suppression depth can be quantified. We calculated suppression depth as the difference between the thresholds because we used a logarithmic contrast scale, equivalent to calculating the log of the contrast ratios when linear contrast is used, as in Tsuchiya et al. (2006) and Yang and Blake (2012). The suppression thresholds exhibited the same increase as a function of mask contrast that bCFS showed so that their difference (suppression depth) was constant.

The early visual system, particularly area V1, is sensitive to contrast and shows a monotonically increasing response to contrast (Albrecht & Hamilton, 1982; Niemeyer & Paradiso, 2017). If the mask is at a high level of contrast, then the signal from the dominant eye would be stronger and, given the standard account of interocular suppression based on reciprocal inhibition (Blake, 1989; Wilson, 2007), this should induce greater inhibition of the contralateral (suppressed) eye (Alais, 2012). As a result, one prediction that follows is that there would have to be a greater amount of target contrast change to overcome the suppressive power of the high-contrast mask. Our observations instead showed that changes to mask contrast had the same impact on bCFS and reCFS thresholds so that the difference between them did not change and suppression depth was constant (equivalently, in terms of linear contrast, dominance and suppression thresholds exhibit the same ratio of response as contrast changes). Thus, it is reasonable to conclude that, regardless of the contrast driving the interocular suppression process, it exerts a constant proportion of inhibition. In our first experiment examining mask contrast, the level of inhibition was ∼13 dB. Our finding aligns with earlier work on binocular rivalry by Blake and Camisa (1979), Hollins and Bailey (1981), and Holopigian (1989), who found that suppression depth did not vary with the contrast of the suppressing stimulus.

To our knowledge, this is the first study to measure the effects of mask contrast on suppression depth in the CFS paradigm, and it has deepened our understanding of the modulators of suppression depth. Not only does CFS produce a constant proportionate suppression regardless of mask contrast, but there also appears to be a ratio between mask contrast and target contrast such that, as mask contrast changed by 6 dB across conditions, average thresholds changed by ∼4 dB. A tentative conclusion is that contrast thresholds will rise at a rate proportional to 2/3 of the contrast increment of the mask. However, further studies are necessary to confirm that this ratio is constant.

Concerning the lack of change in suppression depth, it is possible that increased similarity between target and mask could increase suppression depth. In binocular rivalry, suppression depth increases if the rivaling stimuli share similar characteristics (Alais & Melcher, 2007). In another study examining binocular rivalry between oriented gratings, there was a general level of suppression for any incongruent stimuli which then deepened when the orientation of the probe grating was similar to the rivalry stimulus in the probed eye (Stuit et al., 2009). A similar finding has been reported for CFS, where suppression, as measured by bCFS thresholds, was greater when the target and mask were matched (Han & Alais, 2018).

In Experiment 2, we manipulated mask size at a fixed level of contrast. We observed that mask size did not influence suppression depth or bCFS/reCFS thresholds, in alignment with prior research investigating the effect of mask size on bCFS thresholds (Han et al., 2019). This result is also in agreement with prior accounts of interocular suppression, which emphasize the importance of visual conflict in competing receptive fields at an early stage of visual processing, where receptive fields are small (Alais, 2012; Blake, 1989; Blake et al., 1992). If binocular rivalry takes place within small receptive fields, then increasing mask size beyond a degree or so (a typical V1 receptive field size) (Bentley & Salinas, 2009) will not further optimize the response of a cell as its receptive field is already fully exposed to the stimulus. In contrast, if rivalry suppression took place at a higher level of processing where receptive fields are larger, then expanding the stimulus to more fully occupy the receptive field would be expected to boost the response level and thus influence suppression.

Previous research in binocular rivalry has shown that a surrounding annulus that extends the size and is collinear with competing stimuli can impact suppression durations (Paffen et al., 2004; Paffen et al., 2005), perhaps via iso-surround inhibition (Yu, Klein, & Levi, 2003) or through feedback from subsequent cortical stages with larger receptive fields (van Boxtel, Alais, & van Ee, 2008). In CFS, the traditional Mondrian masking stimuli are comprised of many small circles or squares and thus lacking in collinearity. Han et al. (2019) investigated the effect of collinearity when extending mask sizes and found reduced target suppression durations when masks were larger and collinear, with the regions overlapping with the conflicting target. Importantly, extending the mask with orthogonal gratings increased target suppression durations, consistent with a mechanism of iso-orientation surround suppression that reduces the response of neurons in a receptive field when surrounded by similar stimuli with the same orientation. Thus, larger mask sizes may not impact upon suppression depth when using traditional Mondrian masks, as we have demonstrated; yet, if collinear or coherent information can be integrated over a larger receptive field, an effect may be possible. This is a fruitful avenue for further research to explore if attempting to deepen or reduce consequent suppression depth.

Taken together, the findings of the present study demonstrate the importance of measuring reCFS thresholds to quantify suppression depth because, as Experiment 1 demonstrates, changes in bCFS thresholds can occur that have no bearing on suppression depth. This cautions against drawing conclusions about suppression depth based on bCFS thresholds alone. We observed that alterations in mask contrast and mask size did not alter suppression depth, despite replicating prior work showing an effect of mask contrast on bCFS thresholds.

This study has demonstrated that a change in bCFS thresholds does not necessarily correspond to a change in suppression depth. It is not necessarily the case that all studies using bCFS thresholds or reaction time as a dependent variable must replicate findings using tCFS. However, research that claims to measure unconscious processing, preferential processing, or suppression strength would benefit from replication using the tCFS method, as bCFS thresholds alone are not a reliable indicator of the strength of interocular suppression (Alais et al., 2024).

Across two experiments, we have assessed the impact of manipulating mask contrast and size on CFS thresholds and suppression depth when using the tCFS procedure. Our goal was to use the tCFS paradigm (Alais et al., 2024) to provide a deeper level of understanding about how changing the masking stimulus can affect the depth of suppression. Our first hypothesis, that altering the contrast of the mask will increase suppression depth, was not supported, with target suppression depth remaining constant across a fourfold change in mask contrast (Figure 2), although we successfully replicated the bCFS findings of Han et al. (2019), as we observed increasing bCFS thresholds as mask contrast increased. However, by using the tCFS paradigm, we gained additional insight by showing that this mask contrast level did not modulate target suppression depth. Our second experiment examined whether the size of the mask impacts CFS thresholds and suppression depth. Thresholds for breakthrough and suppression did not vary with mask size, and neither did suppression depth (Figure 3). Suppression depth did not increase as mask size increased. Taken together, these findings demonstrate that, although manipulations of mask contrast may alter bCFS thresholds, the manipulation of neither mask size nor contrast corresponds with a change in suppression depth.

There is another small but important difference between the CFS studies by Tsuchiya et al. (2006) and Yang and Blake (2012) and the tCFS study by Alais et al. (2024). In the former studies, the dominance threshold that formed the baseline measurement for suppression depth was not actually measured. Instead, the threshold for a monocular stimulus with no stimulus presented to the other eye (i.e., without a flickering Mondrian in the other eye) was measured and assumed to be the same as a dominance threshold measured with the Mondrian in the other eye. This was based on binocular rivalry studies showing that monocular thresholds and dominance thresholds are the same (Blake & Camisa, 1979; Westendorf, 1989). However, in CFS, this might not be true because in rivalry the dominance threshold is measured with a constant pattern in the untested eye whose response would habituate over time, whereas in CFS there is a continuously refreshed high-contrast Mondrian in the untested eye. Any interaction at all between the eyes would likely add noise to the tested eye and raise the dominance threshold above the simple monocular threshold level. If so, using the lower monocular threshold instead of a dominance threshold would artificially increase suppression depth and might explain why Tsuchiya et al. (2006) reported such a high value of suppression depth of 1.4 log units (i.e., 28 dB). In our tCFS method, by contrast, dominance and suppression thresholds are both measured with the dynamic masker in the other eye and likely provide a better suppression depth estimate than using the monocular threshold as a proxy for the dominance threshold. We have a future study planned to test whether monocular and dominance thresholds are identical in CFS.