Evidence in Support of the Border-Ownership Neurons for Representing Textured Figures.

We presented one eye with a monocular-boundary-contour (MBC) square, created by phase-shifting a central region of grating relative to a larger uniform grating surround, and the fellow eye with the larger uniform grating. In addition, the grating within the MBC region was rendered with lower contrast relative to the remaining stimulus. Despite this, we found the lower contrast MBC region dominated the perceived cyclopean contrast, with the corresponding region in the fellow eye being suppressed. Secondly, we found for dichoptic stimuli with half-images having square grating regions of different BC strengths, the interocular BC strength ratio determined the perceived contrast of the cyclopean square. Thirdly, we found perceived spatial phase of the cyclopean square was dominated by the spatial phase of the MBC half-image. Altogether, these psychophysical findings provided evidence for a border-to-interior representation strategy, that constructing surface begins at the boundary contour (BC), in binocular contrast and phase integration.

Introduction

The binocular visual mechanism represents three-dimensional (3-D) surfaces by utilizing the difference in visual scene perspectives imaged on the two retinas. But beyond this certainty, it is unclear how the visual system integrates the two retinal perspectives into a veridical surface representation of the 3-D scene. In theory, there are two computational strategies, either by integrating the local contrast information (region-based coding strategy) or by globally extracting the boundary contours (BC) of the surface image followed by the filling in of the local contrast information (border-to-interior representation strategy) (Grossberg and Mingolla, 1985; Julesz, 1960; Mumford et al., 1987; Paradiso and Nakayama, 1991). The region-based coding strategy could be implemented in the primary visual cortex where the early filtering processes exist (e.g., orientation and spatial frequency selective neurons). Consider the dichoptic grating stimulus in Figure 1A. According to the region-based coding strategy, the visual system begins by integrating the local contrast signals from the two eyes and then spatially groups these signals as a surface. Indeed, this hypothesis has gained much support from empirical and modeling studies, particularly from those using random-dot stereograms (RDS) that are devoid of global BC and object information and those using binocular stimuli with equal BC strength in the two eyes (Ding and Sperling, 2006; Georgeson et al., 2016; Halpern and Blake, 1988; Julesz, 1960; Legge and Gu, 1989; Marr and Poggio, 1976; Ohzawa et al., 1990; Poggio and Fischer, 1977).

Figure 1

Experiment 1: Influence of BC on Binocular Contrast Integration

(A–E) The five stimulus conditions (see Transparent Methods and Table S1 for detailed descriptions).

(F) Average results (n = 5) showing the average perceived binocular contrast in the five conditions. Notably, perceived contrast is significantly higher in the D-NS condition than D-MBC condition, indicating the role of BC. We defined perceived contrast in log unit. Therefore, the labeled y axis values of 1.9, 1.7, 1.5, 1.3, and 1.1 correspond, respectively, to 79.4%, 50.1%, 31.6%, 20.0%, and 12.6%. The error bars represent standard errors of the mean.

In contrast, there are fewer empirical studies investigating the border-to-interior representation strategy, although those that did have shed important insights on the roles of extrastriate cortical mechanisms in representing BC and stereoscopic edges (Bakin et al., 2000; Duncan et al., 2000; Hesse and Tsao, 2016; Qiu and von der Heydt, 2005; von der Heydt, 2015; von der Heydt and Zhang, 2018; von der Heydt et al., 2000; Williford and von der Heydt, 2016; Zhou et al., 2000). Specifically, Zhou et al. (2000) discovered a large percentage of macaque V2 neurons signaling border ownership of 2-D texture-free surface/figure in a manner mainly consistent with Gestalt rules. Furthermore, the border-ownership selective neurons' responses could be affected by the image's context that was located beyond the classical receptive fields of V2 neurons, suggesting that they were modulated by feedback signals representing larger surface areas. Critically, Qiu and von der Heydt (Qiu and von der Heydt, 2005) also found the V2 neurons' depth sign of border ownership was consistent with their depth sign revealed with RDS (Qiu and von der Heydt, 2005; von der Heydt et al., 2000). This indicates the global depth determination based on Gestalt rules and local stereoscopic depth signals converge on the border-ownership selective neurons in V2.

An important implication of von der Heydt and colleagues' studies is that the border-to-interior representation strategy is implemented at the extrastriate cortices. However, this speculation requires further empirical evidence. This is because most neurophysiological studies of border-ownership selectivity of the V2 neurons employed texture-free (uniform color) figures, where only the figures' BCs (edges) carried the figural information. There are only two recent studies that revealed border-ownership selective neurons could also respond to natural stimuli such as to single faces and occluding faces (Hesse and Tsao, 2016) and to complex natural scenes (Williford and von der Heydt, 2016) with both BCs and interior textures. However, it remains unknown how the border-ownership selective neurons interact with neurons selective for local textures to form a global surface representation. To address this, the current study provided behavioral evidence for the border-to-interior representation strategy by testing images with interior features (gratings). Our psychophysical experiments confirmed a prediction of the border-to-interior representation strategy that binocular surface perception of pattern and contrast is significantly affected by the relative BC strengths of the dichoptic stimulus.

Results

To represent the binocular surface in Figure 1A, the visual system can interocularly integrate the local contrast signals and then spatially group them as a surface (Ding and Sperling, 2006; Georgeson et al., 2016; Halpern and Blake, 1988; Legge and Gu, 1989). With this region-based coding strategy, the perceived cyclopean contrast of the binocular surface depends only on the contrast of the gratings in each half-image. However, if the visual system adopts the border-to-interior representation strategy, the BC signals from the two eyes will first be integrated according to their relative strength. From the integrated binocular BC, the grating texture is sequentially added by a “filling-in” process whereby local grating patches adjacent to the BC spread inward until the entire area within the BC is filled. Importantly, the relative strength of the BC signals is the factor determining how the grating textures from the two eyes are integrated. Specifically, a stronger BC in one eye will lead to its grating texture carrying a larger weight when integrating with the grating texture from the fellow eye.

Experiment 1 directly investigated the role of the BC by testing whether the contrast of a half-image with BC contributed more to contrast perception over the half-image without the BC (see Figure 1B). The stimulus in Figure 1B differs from that in Figure 1A in one critical aspect, in that the former has BC in one half-image, whereas the latter has BCs in both half-images. In Figure 1A, the gratings in the left and right half-images, respectively, have +45° and −45° phase relative to the horizontal green reference lines. (We define 0° phase as that when the darkest horizontal zone of the grating aligns with the green reference lines, and a positive or negative phase indicates the darkest zone either being above or below the reference lines, respectively). The two gratings also differ in contrast (1.3 versus 1.7 log unit). With free fusion, one perceives the contrast of the cyclopean grating resembling that of the right half-image's (1.7 log unit). For simplicity, we call this a Dichoptic-with-No-Surround (D-NS) stimulus. Previous modeling works posit the perceived cyclopean contrast of D-NS stimulus is accounted for by an interocular contrast gain control mechanism (Ding et al., 2013; Ding and Sperling, 2006; Hou et al., 2013; Huang et al., 2010). To investigate the influence of BC, our new stimulus in Figure 1B added a high-contrast surround grating with −45° phase shift relative to a pair of green reference lines (Table S1). Accordingly, a monocular boundary contour (MBC) square is created in the left half-image of Figure 1B, due to the 90° phase difference between the central and surrounding gratings. There is no BC in the right half-image. We call this a Dichoptic-grating-with-MBC (D-MBC) stimulus. In Figure 1C, we added a surround grating with 180° phase, which causes the phase difference between the central and surround gratings to be 135° in both eyes. Since both half-images have BC, we called it a Dichoptic-with-Binocular Boundary Contours (D-BBC) stimulus.

We hypothesized that if BC significantly contributed to interocular contrast integration, then perceived cyclopean contrast of the central square would not be the same for all three stimuli. Specifically, if a preference existed for representing BC, then the lower contrast of the MBC grating in the D-MBC stimulus would have a larger contribution leading to lower perceived cyclopean contrast, unlike those of D-NS and D-BBC with BC in both eyes (leading to higher perceived contrast). One can qualitatively confirm this by free fusing the stimuli in Figures 1A–1C (Note that in both Figures 1B and 1C, the BCs of the larger grating square in each half-image have the same strength in the two eyes. Therefore, these BCs are unlikely to affect the interocular contrast integration that occurs in the center). We verified these observations empirically using a suprathreshold contrast matching task wherein observers matched the contrast of the test stimulus (D-NS, D-MBC, or D-BBC) with a standard matching stimulus. The matching stimulus was similar to the testing stimulus except that the left and right gratings had the same phase and contrast. Figure 1F depicts our observers' average results. Confirming the prediction, perceived contrast was lower with the D-MBC (1.39 ± 0.02 log unit, M + SE) than D-NS and D-BBC stimuli (1.68 ± 0.01 and 1.66 ± 0.01 log unit, respectively) [D-MBC versus D-NS: t(4) = −18.52 p = 5.0 × 10−5; D-MBC versus D-BBC: t(4) = −22.36, p = 2.4 × 10−5].

We also tested two monocular stimuli as controls, which we call the monocular-NS (Figure 1D) and monocular-MBC (Figure 1E) conditions. We found perceived contrast in D-MBC stimulus was slightly higher but close to both control conditions [D-MBC versus M-NS: t(4) = 4.95 p = 0.008; D-MBC versus M-MBC: t(4) = 4.06 p = 0.015]. This indicates the uniform grating in D-MBC (right half-image in Figure 1B) contributed little to perceived contrast. One explanation for this is that the eye receiving the MBC grating square exerted stronger interocular suppression onto the uniform grating. This is reminiscent of our previous binocular rivalry (BR) studies using an MBC-BR stimulus (Figure 2A), where the MBC disc (vertical grating) half-image largely dominates over the corresponding uniform horizontal grating half-image (Ooi and He, 2006; Su et al., 2009, 2011a, 2011b; Xu et al., 2010). Nonetheless, we recognized that the D-MBC stimulus (Figure 1B) differs from the MBC-BR stimulus in that the grating orientation in the two eyes are similar in the former, and, thus, fusible. To support our hypothesis that the uniform grating in D-MBC was also interocularly suppressed, Experiment 2 measured increment thresholds for detecting a monocular Gabor probe in the D-MBC stimulus (Figure 2B). The left pair in Figure 2B shows a Gabor probe being presented against the MBC grating disc pedestal (1.0 log unit), whereas the right pair shows the probe being presented against the uniform grating pedestal (1.6 log unit). We predicted that should the uniform grating be suppressed, detection threshold would be higher in the scenario of Figure 2B right pair than left pair (Su et al., 2009, 2010). We measured our observers' thresholds using a combined 2AFC and QUEST method and found the average threshold (circles in Figure 2E) was indeed higher on the uniform grating pedestal [1.37 ± 0.05 vs. 0.91 ± 0.03 log unit; t(3) = 15.33; p = 6.02 × 10−4].

Figure 2

Experiment 2: Influence of BC on Increment Contrast Threshold Measured with a Gabor Probe

(A) An example of an MBC-BR stimulus with orthogonal orientation features in each eye.

(B–D) The three types of stimuli tested. For each dichoptic pair, increment contrast threshold was measured by the Gabor probe (as pointed to by green arrows) that is depicted in one half-image.

(E) Average results (n = 4) showing the average increment contrast threshold for seeing the Gabor probe as a function of pedestal contrast. Notably, when the pedestal was 1.6 log unit, threshold is highest in the D-MBC condition indicating suppression of the homogeneous grating (1.6 log unit) by the MBC half-image with low contrast (1.0 log unit) in the fellow eye. We defined perceived contrast in log unit. Therefore, the labeled y axis values of 1.4, 1.2, 1.0, and 0.8 correspond to 25.1%, 15.9%, 10%, and 6.3%, respectively. The error bars represent standard errors of the mean.

In addition, we tested two control stimuli, the D-NS (Figure 2C) and D-BBC/ring (Figure 2D) stimuli, to rule out the confounding factor that the lower contrast pedestal of the MBC disc caused the lower detection threshold. If the higher threshold on the uniform grating in D-MBC (Figure 2B, right pair) was merely due to a higher pedestal contrast, we should expect similar high thresholds with the D-NS (Figure 2C, right pair) and D-BBC/ring (Figure 2D, right pair) stimuli. But this prediction was not borne out. As shown in Figure 2E, with the higher contrast pedestal being 1.6 log unit, the average probe threshold was significantly higher with the D-MBC stimulus (circular symbol) than with either the D-NS [t(3) = 8.07, p = 0.004] or D-BBC/ring [t(3) = 5.37, p = 0.013] stimuli. Further, for both control stimuli, the average probe thresholds did not differ significantly between the two contrast pedestals [D-NS: square symbol, t(3) = 1.97; p = 0.144; D-BBC/ring: triangular symbol, t(3) = 0.37; p = 0.738]. Also, with the lower contrast pedestal, the average probe threshold was significantly lower with the D-MBC stimulus than with the D-NS stimuli [t(3) = -8.01, p = 0.004] and D-BBC/ring [t(3) = -7.55, p = 0.005]. This suggests that in comparison against the D-NS and D-BBC/ring stimuli, the lower increment contrast threshold measured on the half-image with MBC in the D-MBC stimulus was due to reduced suppression from the fellow eye.

Experiments 1 and 2 together revealed that with the D-MBC stimulus, the visual system represents the MBC grating while suppressing the uniform grating (without BC) in the fellow eye even when the two gratings are fusible. But what happens when the two eyes receive BC of different strengths? To answer this, Experiment 3 capitalized on the D-BBC stimulus to measure how having an interocular BC strength ratio affected binocular contrast integration. Recall we created the BCs of the dichoptic stimuli in Experiment 1 by adding surround gratings of different phases (Figure 1). Similarly, the current experiment used a series of set phases (−45, −75, −90, −105, −120, −135, −180 deg) to manipulate the interocular BC strength ratio (Table S2). For example, the stimulus in Figure 3A is modified from the D-MBC stimulus (Figure 1B) by replacing the −45° surround grating with a −75° surround. This causes a significant increase of BC strength in the right half-image and hence changes the BC strength ratio between the right (high-contrast central grating) and left (low-contrast central grating) half-images. We then conducted two measurements. One, we measured perceived contrast of the various D-BBC stimuli using the same contrast matching task as in Experiment 1. Two, we measured observers' perceived strength (saliency) of the BC in each half-image by rating it from a scale of 0–10. This allows us to calculate the perceived interocular BC strength ratios between the right and left half-images. Figure 3B plots the average perceived contrast (y axis) against the perceived interocular BC strength ratio (x axis). Clearly, perceived contrast varied significantly with the perceived interocular BC strength ratio [F(6, 21) = 10.70; p = 1.8 × 10−5; One way ANOVA with repeated measures]. The data were fitted by the curve, . The finding that perceived contrast increases with the interocular BC strength ratio indicates that BC strengths, besides grating contrast, contributes to binocular contrast integration of D-BBC stimuli.

Figure 3

Experiment 3: Interocular BC Saliency Affecting Binocular Contrast Integration

(A) A sample stimulus showing a pair of half-images with different relative phase shifts between the central and surround gratings creating unequal BC saliency in each half-image. The contrast of the central grating with the higher BC saliency (left half-image) is lower than that in the fellow (right) half-image. See Transparent Methods and Table S2 for detailed descriptions of stimuli.

(B) Average results (n = 4) showing the perceived contrast as a function of interocular BC strength ratio between the central square grating with high contrast and central square grating with low contrast (filled symbols). The curve fitting the data is derived from observers' ratings of the BC saliency. Perceived binocular contrast is higher, which is largely contributed by the high-contrast grating (right), when the high-contrast grating's BC strength is about the same as that of the low contrast grating (left) (ratio = 1). We defined perceived contrast in log unit. Therefore, the labeled y axis values of 1.7, 1.5, and 1.3 correspond to 50.1%, 31.6%, and 20.0%, respectively. Both the vertical and horizontal error bars represent standard errors of the mean.

Experiment 4 generalized our conclusion by extending to another aspect of perceived spatial feature of the interior surface, namely, its spatial phase. We modified the stimuli in Experiment 1 by assigning all gratings with a contrast of 1.5 log unit (Figure 4). With free fusion of the D-NS stimulus in Figure 4A, an observer with balanced eyes perceives the cyclopean grating has zero phase, ie, its darkest zone is aligned with the green reference lines. This confirms the cyclopean grating's phase is the mean phase of the two gratings from the two eyes [i.e., (−45° + 45°)/2 = 0°] (Badcock and Derrington, 1987; Ding et al., 2013; Ding and Sperling, 2006; Han et al., 2018; Huang et al., 2010). However, in Figure 4B, the perceived phase of the D-MBC stimulus is no longer zero but at a positive phase where the darkest zone of the cyclopean grating is above the reference lines. In fact, the perceived phase appears similar to that of the MBC grating in the left half-image. This observation resembles the perceived cyclopean contrast results of Experiment 1 where the contrast of the MBC grating dictated the perceived contrast. Also similar to Experiment 1, for the D-BBC stimulus, perceived cyclopean phase of the grating is almost zero, similar to the D-NS stimulus. We empirically verified these observations by measuring the perceived cyclopean phases of the central grating square of the D-NS, D-MBC, and D-BBC stimuli (Figures 4A–4C). Figure 4F depicts the average results showing perceived phase of the central grating in the D-MBC stimulus (above 45°) differed significantly from those in the D-NS and D-BBC stimuli [D-MBC versus D-NS: t(4) = 41.72, p = 2.0 × 10−6; D-MBC versus D-BBC: t(4) = 39.48, p = 2.5 × 10−6], indicating the effect of BC on binocular phase integration. As in Experiment 1, we tested two monocular control conditions, M-NS (Figure 4D) and M-MBC (Figure 4E). We found perceived phase in M-MBC was not significantly different from D-MBC (t(4) = -2.14; p = 0.10). This further confirms perceived cyclopean phase of the grating in D-MBC was dictated by the MBC grating. Curiously, we found perceived phase in the M-MBC stimulus was significantly larger than that in M-NS (t(4) = 5.64; p = 0.005). This suggests a phase-repulsion phenomenon between the central and surround gratings of different spatial phases.

Figure 4

Experiment 4: Influence of BC on Phase Integration

(A–E) The stimuli were the same as those in Experiment 1 except that all gratings had a contrast of 1.5 log unit.

(F) Results showing the average perceived binocular phase in the five conditions. Notably, perceived phase in the D-MBC condition is higher than in the D-NS condition, indicating the role of BC (n = 5). The error bars represent standard errors of the mean.

Discussion

Our results indicate binocular BC signals influence perceived cyclopean surface features (contrast and phase) of a grating surface. In the limiting case where one eye sees a BC-defined grating patch and the fellow eye sees the grating with the same orientation but without BC, the observer perceives predominantly the MBC grating patch alone. In other scenarios where there are BCs in both half-images, the interocular BC strength ratio affects the perceived cyclopean contrast of the grating surface. And in the case where the BCs in the two eyes have similar strengths (e.g., when interocular BC strength ratio is unity in Experiment 3), the high-contrast grating dominates the cyclopean contrast perception. In other words, both the interocular BC and local contrast signals determine the perceived interior features of a surface.

The current finding that BC influences the perception of a surface's interior supports the notion that surface representation begins at the BC. It is also consistent with our previous studies using binocular rivalry stimuli (Su et al., 2010, 2011a, 2011b; Xu et al., 2010, 2016). We showed with non-fusible binocular rivalry (BR) stimulus that the visual system first registers the BC, before the local texture region adjacent to the BC, for determination of the dominant surface (Su et al., 2011a, 2011b). In one study (Su et al., 2011b), we presented an MBC-BR stimulus similar to that in Figure 2A, except for the MBC being rectangular in shape and its texture horizontal-grating, to the observers for varying durations (30–500 ms). The fellow eye received a vertical-grating half-image. We then asked observers to report the pattern of the textures (vertical, horizontal, or checkerboard) they saw in the rectangular area interior to the MBC. Our data showed at short durations, observers saw the (dominant) horizontal grating adjacent to the MBC, whereas the central area of the MBC rectangle was filled with checkerboard texture. With longer presentation durations, the coverage of the dominant horizontal grating spread toward the center while the central checkerboard area shrank, until the entire MBC rectangle was covered with the dominant grating. We then calculated the speed of filling-in and found it to be constant when scaled according to cortical distance in the early visual cortical areas (V1 and V2). Of significance, the current study supports a similar border-to-interior strategy for representing texture surface with fusible features (same orientation). Here, we further showed perceived binocular surface's (e.g., interior contrast perception) dependence on both local contrast and BC signals (Figure 3). In the case of the D-MBC stimulus, the cyclopean percept of the interior surface (grating) is mainly contributed by the half-image with the BC.

Although there is yet no neurophysiological evidence directly showing how the border-to-interior representation strategy is implemented to represent textured-surfaces, there are empirical findings suggesting the involvement of both top-down cortical feedback networks and bottom-up visual information. Based on these studies, it is reasonable to posit the border-to-interior representation strategy requires an image's border ownership (BC) to be determined before the texture and color fill-in the interior of the surface. The border-ownership selective neurons in V2 are likely modulated by feedbacks from higher-level visual cortical areas where characteristics of global surface properties, 3-D surface layouts, and natural objects are represented (Chen et al., 2014; Craft et al., 2007; Hesse and Tsao, 2016; Ko and von der Heydt, 2018; Lamme and Roelfsema, 2000; Qiu et al., 2007; Qiu and von der Heydt, 2005; von der Heydt, 2015; von der Heydt and Zhang, 2018; von der Heydt et al., 2000; Williford and von der Heydt, 2016; Zhou et al., 2000). In addition, these V2 neurons receive bottom-up local feature information, such as T-junctions and binocular disparity (Qiu and von der Heydt, 2005; von der Heydt and Zhang, 2018). For example, Qiu and von der Heydt found the V2 neurons' depth sign (selectivity) of border ownership was consistent with their depth sign (selectivity) revealed with RDS stimuli (Qiu and von der Heydt, 2005; von der Heydt et al., 2000). Presumably, the border-ownership selective neurons can carry surface texture and color features to initiate the filling-in process from the borders. This being the case, it requires a battery of border-ownership selective neurons having tuning functions to represent various types of texture patterns and colors. Indeed, single-unit recording studies of V2 neurons have revealed neural activities of border-ownership selective neurons are affected by surface texture and color features (Hesse and Tsao, 2016; Zhou et al., 2000). This further suggests border ownership is determined by a population of neurons distributed along the visual pathways (Hesse and Tsao, 2016; Ko and von der Heydt, 2018). It would be interesting for future studies to investigate whether the border ownership selective neurons directly participate in the filling-in process. In addition, our previous psychophysical findings as well as those of the current Experiment 2 have led us to hypothesize that the V2 border ownership neurons play a broader role in mediating interocular suppression to represent a global binocular surface, by integrating the often times ambiguous information from the two eyes (Ooi and He, 2006; Su et al., 2009, 2011a, 2011b; Xu et al., 2010).

The border-to-interior representation strategy notwithstanding, there is evidence that the binocular visual system is capable of representing surfaces devoid of BC information by employing the region-based coding strategy. In fact, much research has shown computer-generated RDS images can lead to 3-D surface perception (Julesz, 1960). Nevertheless, there are advantages in adopting a border-to-interior representation strategy to represent natural surfaces. This is because most natural surfaces are smooth, with abrupt surface discontinuation or curvature changes only occurring at their borders. As a consequence, useful information regarding the overall 3-D surface attributes pertaining to surface shapes and depth discontinuities are carried by BCs (Grossberg and Mingolla, 1985; Nakayama and Shimojo, 1990). One could further speculate that in the natural scene, it is arguably more efficient for the visual system to first find the corresponding BCs from the two eyes and then within each corresponding binocular BC to match the local features in the two eyes and compute their binocular depth relationship. In fact, a number of stereoscopic depth studies have shown that BCs can be critical in determining 3-D surface representation (He and Ooi, 2000; McKee, 1983; Mitchison and McKee, 1990; Ramachandran and Cavanaugh, 1985; Richards, 1977; Su et al., 2009). In addition, the color and texture of the surface region adjacent to the BC usually provides a good estimate of the entire surface, as the interior of most natural surfaces have common optical properties of color and texture (Elder and Goldberg, 2002; Fine et al., 2003). Moreover, neurophysiological studies have showed that V2 neurons with border ownership selectivity respond to both binocular depth discontinuities and luminance BC (Hesse and Tsao, 2016; Qiu and von der Heydt, 2005; von der Heydt, 2015; von der Heydt et al., 2000; Zhou et al., 2000). Consequently, using the border-to-interior strategy could optimize the efficiency of representing the interior structures of natural surfaces. In other words, the choice of representing binocular surfaces using the border-to-interior strategy among other possible strategies reflects the visual system's preference to utilize the ecological constraints inherent in natural scenes (Gillam and Borsting, 1988; Nakayama et al., 1995; Ooi and He, 2005; Paffen et al., 2006; Shimojo and Nakayama, 1990; van Bogaert et al., 2008).

Resource Availability

Lead Contact

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Teng Leng Ooi (ooi.22@osu.edu).

Materials Availability

This study did not generate new materials.

Data and Code Availability

The experimental code and data that support the findings of this study are available from our repository (GitHub: https://github.com/V-sci/BC_contrast-phase).

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.