Using the visual cliff and pole descent assays to detect binocular disruption in mice

Héctor De Jesús-Cortés; Teresa L. M. Cramer; Daniel A. Bowen; Francis Reilly-Andújar; Sophie Lu; Eric D. Gaier; Mark F. Bear; Héctor De Jesús-Cortés; Teresa L. M. Cramer; Daniel A. Bowen; Francis Reilly-Andújar; Sophie Lu; Eric D. Gaier; Mark F. Bear

doi:10.48130/vns-0026-0007

Amblyopia, a neurodevelopmental binocular visual disorder, is commonly modeled in animals using monocular deprivation (MD) during the critical period of visual development. Despite extensive research at synaptic, cellular and circuit levels, reliable behavioral assays to study stereoscopic deficits in mice are limited. This study aimed to characterize the visual cliff assay (VCA) and the pole descent cliff task (PDCT), and to evaluate their utility in detecting binocular dysfunction in mice. By using manipulations of binocular vision, including monocular occlusion, pupillary dilation, and amblyopia (by prior MD), we show that optimal performance in both the VCA and PDCT requires matched binocular input. However, deficits after MD in the VCA exhibited relatively small effect sizes (7%–14%), requiring large sample sizes. By comparison, the PDCT demonstrated larger effect sizes (43%–61%), allowing for reliable detection of binocular dysfunction with fewer animals. Following validation through multiple monocular manipulations that are relevant to clinical paradigms, the PDCT emerged as the preferred assay for detecting deficits in stereoscopic depth perception in mice. These findings provide a robust framework for using the VCA and PDCT in mechanistic and therapeutic studies in mice that will offer insights into the neural mechanisms of binocular vision and potential interventions for amblyopia.

HTML

Introduction

Amblyopia ("lazy eye") is a prevalent neurodevelopmental visual disorder stemming from abnormal visual experience early in life. Of the many visual deficits associated with amblyopia, impaired stereoacuity is among the most common, and it is functionally impactful and resistant to restoration with treatment^[1−3]. Monocular deprivation (MD) via unilateral eyelid suture during the critical period is the predominant paradigm used to model and study amblyopia in animals^[4,5]. MD elicits an ocular dominance (OD) shift in the primary visual cortex (V1), reflecting the loss of strength of the synaptic inputs from the deprived eye. Studies in cats and primates have shown that the primary locus of the underlying functional plasticity is the synaptic neuropil of V1 itself. To gain an understanding of the molecular mechanisms for such plasticity, work increasingly has focused on the mouse. The limitations of the mouse as a model to understand primate vision are numerous and include limited binocular vision, poor spatial acuity, and the absence of the columnar organization in V1 that is characteristic of carnivores and primates. Nevertheless, studies in mice have been instrumental in pinpointing changes at the molecular level that shift OD in V1 after MD. This work is directly relevant to the understanding of the most treatment-resistant form of amblyopia in humans, deprivation amblyopia, which arises from optical obstruction of the visual axis (e.g., by a cataract)^[6].

Studies of MD in rodents have suggested novel treatment approaches to promote recovery from amblyopia after the critical period. These include environmental enrichment^[7], dark exposure^[8], pharmacologic retinal silencing^[9,10], and ketamine^[11,12]. Assays used to assess treatments' effectiveness typically include electrophysiological or optical measures of the visual responses in V1, and behavioral estimates of monocular spatial acuity. Reliable behavioral assays to study stereoscopic deficits in mice, however, are needed.

To test the stereoscopic depth perception of human infants, Gibson and Walk developed the visual cliff assay (VCA)^[13]. The VCA comprises a glass plate atop a patterned floor with a sharp drop-off. Subjects with functional stereoscopic vision are inherently averse to the "cliff" side and tend to spend more time on the "safe" side. The VCA has been adapted for use in several animal species to assess the perception of stereoscopic depth^[14−20]. The assay has been used in rats to study the development of depth perception and the role of early-life visual experience^{[14,15,19,20]}, including the effect of MD^[20,21]. Although it was initially questioned whether mice are capable of visually perceiving depth^[22], emerging evidence has demonstrated that mice are able to discriminate stereoscopic surfaces despite having a binocular visual field of only 40°^[23−27]. The performance of mice in the VCA can be impaired by restricting vision to one eye, albeit modestly^[28].

Recently, a modified version of the VCA has been described, called the pole descent cliff task (PDCT)^[29]. In this task, mice descend a vertical pole to a conical platform which engages the upper visual field, where there is more binocular overlap in visual representations. The vertical pole is suspended over a glass plate divided into four quadrants; three of the four quadrants appear as a visual cliff using stereoscopic depth cues. Neurotypical mice, viewing with both eyes, will exit the pole onto the "safe" quadrant. Acutely closing one eye results in errors that increase as the apparent depth of the cliff quadrants increases. Although promising, its performance has not yet been shown to be sensitive to the disruption of stereopsis caused by deprivation amblyopia following long term MD (LTMD) in mice.

Robust and reliable behavioral assays of binocular vision are clearly needed to validate the functional outcomes observed in experimental treatment paradigms in mice. We therefore undertook the current study to test the utility of the VCA and PDCT to probe deficits in binocular vision in mice caused by LTMD.

Results

The contribution of vision to the VCA

To characterize the VCA, we first quantified the inherent tendency of mice to stay on the safe side under normal lighting conditions. Female and male mice (P45–58, n = 7) were placed on the middle of the apparatus (Fig. 1a) facing the safe side and allowed to freely explore the arena for 5 min as previously described^[28,30]. Sessions were video-recorded from above, and the behaviors were scored as the percentage of time spent on the safe side per minute (Fig. 1b), the time to the first zone transition (Fig. 1c), and the number of transitions between zones per minute (Fig. 1d). Performance did not differ between female and male mice, so data from both sexes were combined for all analyses. On average, animals spent 96.0% ± 4.8% (95% confidence interval [CI]) of the first minute on the safe side. A nonparametric repeated Friedman test showed the main effect of test time on the proportion of time spent on the safe side over the 5-min test (p = 0.0060). Pairwise comparisons revealed that time spent on the safe side decreased significantly to 60.0% ± 22.2% (∆ = 37.5%; p = 0.0162) and 57.0% ± 22.2% (∆ = 40.6%; p = 0.0021) during the third and fourth min, respectively (Fig. 1b). On average, animals made the first zone transition to the cliff side at 66.0 ± 15.9 s (Fig. 1c). The repeated Friedman test showed a main effect of test time on the number of zone transitions (p = 0.0088). We observed a significant increase in the number of zone transitions by the third (first min: 0.6 ± 0.7, third min: 4.1 ± 2.5; p = 0.0211), fourth (4.1 ± 1.7; p = 0.0162), and fifth min (3.9 ± 1.8; p = 0.0211) (Fig. 1d). The exploration index (EI), a standard single measure of VCA performance^[31,32], was 0.44 ± 0.23 (Fig. 1e).

To evaluate the contribution of vision, we tested VCA performance in complete darkness in a separate cohort of animals (P45–58, n = 7) (Fig. 1f). On average, animals spent 58.6% ± 8.2% of the first minute on the safe side (Fig. 1g), significantly less than in the light (p = 0.0006; Mann–Whitney test). Repeated measures ANOVA showed no effect of test time on the proportion of time spent on the safe side (p = 0.3765) (Fig. 1g), which ranged between 49.3% and 59.6% over the 5-min test. On average, animals made the first zone transition to the cliff side at 10.1 ± 7.9 s (Fig. 1h), significantly sooner than in the light (p < 0.0001, unpaired t-test). Repeated measures ANOVA showed no effect of test time on the number of zone transitions (p = 0.8995) (Fig. 1i). The EI was significantly decreased in the dark (0.11 ± 0.09) compared with the light (p = 0.0006, unpaired t-test) (Fig. 1e vs j). Taken together, these data unambiguously indicate that vision is required for normal performance on the VCA in mice, and that visual cues deter zone transitions in the first 2 min of the test.

Test–retest stability of VCA performance
Recognizing that experiments involving a treatment or intervention may benefit from multiple runs of the VCA by the same animals, we next examined the test–retest reliability. We modified the assay to only test animals for 2 min to avoid habituation to the cliff side in the first trial (see Fig. 1). A new cohort of female and male mice (P48–68, n = 20) was run 4 times at 1-week intervals. The repeated measures Friedman test showed a borderline, but ultimately statistically nonsignificant, effect of trial number on the percentage of time spent on the safe side (p = 0.0589) (Fig. 2a). Given the marginal effect of trial number and our objective to assess repeatability over weeks, we examined pairwise comparisons. Between Week 0 and Week 1, there was no significant change in time spent on the safe side (p = 0.0825). There was a main effect of trial number on the time to the first zone transition (p = 0.0248) that was carried by a decrease between the first and second trial (Fig. 2b, p = 0.0212). There was a main effect of trial number on the number of zone transitions (p = 0.0218), with no pairwise differences with the first trial (Fig. 2c, p-values > 0.15). Thus, the VCA performance remained relatively stable over the weeks.

Acute disruptions of binocularity impair VCA performance
Knowing that VCA performance requires vision, we next tested the contribution of binocularity in the 2-min test. VCA performance has previously been shown to drop under monocular conditions (with one eye sutured closed) in both rats^[20] and mice^[28], so we began by attempting to replicate this result. Male and female mice (P42–60, n = 11 per group) had one or both eyes sutured closed under isoflurane anesthesia 1–3 h prior to VCA testing. Control (sham) animals underwent the same manipulations, but their eyes were immediately re-opened. Eye closure decreased performance in a dose-dependent manner across all metrics (Fig. 3a–d). ANOVA showed a main effect of eye closure on the percentage of time spent on the safe side (p < 0.0001). Time spent on the safe side decreased by 14.0% with one eye closed (p = 0.0082) and 29.7% with both eyes closed (p < 0.0001) (Fig. 3a). There was a main effect of eye closure on time to the first zone transition (p = 0.0013) with a decrease of 38.5 ± 8.8 s with one eye closed (p = 0.0068) and 11.2 ± 6.7 s with both eyes closed (p = 0.0025) (Fig. 3b). There was a main effect of eye closure on the number of zone transitions (p = 0.0006) (Fig. 3c), with an increase of 5.3 ± 2.5 with one eye closed (p = 0.0085) and 6.4 ± 2.1 with both eyes closed (p = 0.0019). The dose-dependence of eye closure on VCA performance corroborates the dependence of the VCA on vision (Fig. 1) and also suggests a contribution of binocular function. Considering that differences in movement may confound VCA performance comparisons between groups, we used open-source pose estimation software (ezTrack^[35] to quantify the total distance traveled during the 2-min test. There was a main effect of eye closure on the total distance traveled (p = 0.0006) (Fig. 3d). Group comparisons revealed a significant increase of 501 cm (45%) in total distance traveled when both eyes were closed (p = 0.0009) but no significant change when only one eye was closed (248 cm; p = 0.0990). Therefore, we can at least partially attribute VCA performance deficits under monocular conditions to a lack of binocularity.

Monocular occlusion significantly narrows the visual field, particularly in mice, and could influence VCA performance, which is globally dependent on vision (Fig. 1). To better delineate whether VCA performance is dependent on balanced binocular inputs rather than a narrowing of the visual field, we next tested the influence of optical manipulations. Spherical aberration degrades spatial acuity and stereopsis, particularly with interocular disparity^[36]. The optics of the murine eye impart significant susceptibility to optical aberration as a function of pupil size^[37], likely owing to the relative sphericity of the lens. We applied atropine to dilate the pupil of one or both eyes and tested the animals for 2 min via the VCA ≥ 60 min later (P38–42, n = 12–13 per group, Fig. 3e–h). ANOVA showed a main effect of atropine on time spent on the safe side (p = 0.0160). Group comparisons revealed that monocular atropine decreased the percentage of time spent on the safe side by 11.5% (p = 0.0083) whereas binocular atropine imparted no significant change (p = 0.1464) (Fig. 3e). Atropine imparted no main effect on the time to the first zone transition (p = 0.3551) (Fig. 3f) or the number of zone transitions (p = 0.1826) (Fig. 3g). In contrast to eye closure, atropine imparted no change in total distance traveled during the test (p = 0.5630) (Fig. 3h). Therefore, the effect of atropine on time spent on the safe side can be attributed to binocular imbalance. These data demonstrate that disruption of optically matched binocular input is sufficient to impair VCA performance.

Amblyopic mice demonstrate binocular dysfunction in the VCA
We next tested the effects of amblyogenic rearing. Female and male mice underwent 3 weeks of LTMD or sham eyelid closure/opening from P21 to P42 and were tested on the VCA 1–3 h after eye opening (Fig. 4a). To ensure the study was adequately powered to detect meaningful group differences, an interim power analysis was performed after testing the first cohort of the sham control (n = 14) and LTMD (n = 14) animals. According to this analysis, with a Type I error rate set at 5% (two-sided) and a desired power of 80%, a sample size of 44 animals per group was determined. Amblyopic mice (sham control, n = 44; LTMD, n = 52) showed impaired performance across all VCA metrics (Fig. 4b–d). LTMD decreased time spent on the safe side by 7.5% compared with the control animals (p = 0.0070) (Fig. 4b). LTMD decreased the time to the first zone transition by 16.5 s (20.6%) (p = 0.0243) (Fig. 4c) and increased the number of zone transitions (p = 0.0393) (Fig. 4d), but there were no differences between sham and LTMD mice in the routes taken when transitioning from the safe to the cliff side (Supplementary Fig. S1). LTMD had no effect on total distance traveled (p = 0.6218) (Fig. 4e), suggesting that amblyopia does not increase ambulatory movement the way occlusion does. Together, these data demonstrate that LTMD imparts an amblyopic deficit in binocular function in mice, to which multiple VCA metrics are sensitive.

The PDCT reveals visual deficits following acute disruption of binocularity
Previous behavioral work has demonstrated that mice use binocular information in the PDCT to infer depth and exit the pole onto the safe quadrant^[29]. We attempted to replicate these results and began by quantifying the overall tendency of mice to exit the pole. Female and male mice (P75–P110, n = 9) were placed on the pole facing downwards to motivate descent (Fig. 5a). Sessions were video-recorded from above, and behaviors were quantified as the percentage of exits. Performance did not differ between female and male mice, so data from both sexes were combined for all analyses. Repeated measures ANOVA revealed a significant effect of trial on the percentage of mice exiting the pole onto any of the four quadrants (p = 0.0003), with a marked reduction observed from Trial 4 onwards across the 10 interleaved 3-min trials (Fig. 5b). Although 100% of mice exited the pole on the first trial, only 33.3% ± 16.7% exited on the fifth trial and no mice exited on the tenth trial (Dunnett's multiple comparisons: Trial 1 vs. Trial 5, p = 0.0224; Trial 1 vs. Trial 10, p < 0.0001). These findings differ from published results obtained using the PDCT. Boone et al.^[29] reported 10–15 exits per mouse by repeatedly repositioning the animals back onto the top of the pole whenever they failed to initiate a descent or engaged in nondescending behaviors. In the absence of such interventions, we found that the mice reliably self-initiated one or two descents. Because repeated trials could not be obtained under these minimally handled conditions, we adopted a single-trial PDCT paradigm. To evaluate the contribution of vision, as with the VCA, we next tested PDCT performance under normal lighting conditions and in complete darkness using two separate cohorts of animals (P65–P95, n = 9–10 per group) (Fig. 5c). On average, 88.9% ± 11.1% of animals exited the pole onto the safe quadrant under normal lighting conditions, whereas only 36.3% ± 16.1% of animals exited onto the safe quadrant in darkness (p = 0.015, unpaired t-test). These findings indicate that vision is required for performance in the PDCT. While observing the animals' behavior over the course of the assay, we noticed that animals tested in darkness took longer to descend the pole. The lack of visual cues in darkness prevents mice from using vision to gauge distances, which likely results in slower descent times. We quantified the descent time and found that animals tested in darkness took significantly longer to exit the pole (11.8 ± 3.6 s) than animals tested in the light (7.6 ± 3.7 s, p = 0.021, unpaired t-test). Taken together, these findings demonstrate that vision is required for descending and exiting the pole onto the safe quadrant in the PDCT.

We next tested whether binocularity is required for PDCT performance. Using the VCA, we had found that monocular atropine and eye suturing produce comparable deficits in performance, as measured by time spent on the safe side (Fig. 3). Because atropine is simple to apply, does not narrow the visual field, and may better approximate the effect of amblyopia, we focused on this manipulation in the PDCT. Atropine was applied to one or both eyes, and the animals were tested via the PDCT ~60 min later when pupil dilation was maximal (P55–65, n = 13–14 per group, Fig. 5d,d'). Though atropine imparted no main effect on mice exiting the pole onto the safe quadrant (control: 92.3% ± 0.7%, monocular atropine: 100%; binocular atropine: 92.8% ± 0.7%, p = 0.60, Fig. 5d), hypothesis-driven group comparisons revealed that monocular atropine significantly increased the descent time compared with controls (p = 0.02, unpaired t-test) and binocular atropine did not (p = 0.38, unpaired t-test, Fig. 5d'). These findings indicate that the PDCT metric "safe/unsafe exit" may not be sensitive enough to detect subtle imbalances in binocular vision. In contrast, disruption of optically matched binocular input is sufficient to impair pole descent time in the PDCT.

Amblyopic mice demonstrate binocular dysfunction in the PDCT
We next tested the effects of amblyogenic rearing via the PDCT. Female and male mice (n = 8–10 per group) underwent 3 weeks of LTMD or sham eyelid closure/opening from P21 to P42 and were tested on the PDCT 1–3 hours after eye opening, as shown in Fig. 4a. Similar to our previous experiment using atropine, we found that induction of amblyopia by LTMD had no effect on the success of mice exiting the pole onto the safe quadrant (sham: 100%, LTMD: 100%, data not shown). However, LTMD significantly increased the descent time compared with the sham controls (p = 0.008, unpaired t-test, Fig. 5e). These findings indicate that LTMD is sufficient to impair pole descent time in the PDCT.

Together, these experiments revealed that increased descent time rather than a decreased percentage of safe exits reports an imbalance in vision between the two eyes. To confirm this result in a separate set of experiments, mice that underwent 3 weeks of LTMD were tested using a second version of the PDCT apparatus that more closely resembled the configuration described in a previous study^[29]. Although the physical dimensions of the apparatus differed from the first experiment (see Methods), the behavioral effects of LTMD were comparable. Again, LTMD had no effect on pole–platform exit choice but significantly increased the descent time compared with the sham controls (p = 0.017, unpaired t-test, Fig. 5e'), independently replicating the effect of amblyopia on the task and validating the modified configuration for future use.

Discussion

We characterized the VCA and PDCT as methods to study binocular perceptual deficits in mice. Mice are the preferred species to study the experience-dependent neuroplasticity of visual processing despite their important differences from humans with respect to neuroanatomy and the circuitry of the visual system^[38]. Mechanistic studies in mice are best validated through behavioral studies that demonstrate functional relevance. Here, we provide evidence that the VCA and PDCT rely on balanced binocular inputs for optimal performance in mice. Both assays have advantageous features, including relatively fast execution, no need for operant training, and readout through several simple metrics. However, the assays also have important limitations, as evidenced through our results, that should be weighed carefully when considering their use.

Impairments in binocular function are detected by the VCA, but effect sizes are modest
Monocular occlusion, pupillary dilation (and consequential optical aberration), and amblyopia all degraded performance in the VCA, supporting the conclusion that binocular sensory integration is required for the task. Monocular pupillary dilation significantly reduced the percentage of time spent on the safe side, whereas bilateral atropine did not (Fig. 3e), highlighting the need for balanced visual input in binocular integration. Monocular blur more significantly degrades stereopsis compared with binocular blur^[39,40]. However, the manipulations that disrupted binocular function did not drive VCA performance to the levels we observed for animals in the dark (Fig. 1g–j) or with both eyes closed (Fig. 3a–c). One explanation could be that the manipulations of binocular function in our studies, such as unilateral pupillary dilation and amblyopia after LTMD, may only partially disrupt binocular function. However, this is unlikely to be the sole explanation for the difference, as the unambiguous loss of binocular function caused by closing one eye similarly did not reduce performance to the levels observed in darkness or with both eyes closed.

Another explanation is that VCA performance can be partially maintained using monocular cues. One mode by which animals can detect depth is by discerning changes in monocular cues with movement, called motion parallax^[41,42], and this may account for the partial effects of the complete disruption of binocularity on VCA performance seen with monocular occlusion. Although we did not test acutely monocularly occluded animals (unilateral eyelid suture just prior to VCA testing in adulthood; Fig. 3a–c) directly alongside those who were rendered amblyopic (LTMD through the critical period; Fig. 4b–d), the apparently better performance by the latter cohorts supports a potential effect of adaptation to monocular conditions and learned reliance on monocular cues. The impact of monocular cues is an unavoidable potential confounder in the VCA.

The main limitation in using the VCA to detect deficits in binocular function in mice is the relatively small effect sizes (7%–14%), which we found consistently across experiments and VCA metrics. Our observations are consistent with those reported previously in mice^[28], although that study reported a larger effect of MD (~20%) when tested in younger animals during the critical period. Modest effect sizes limit the utility of this assay to detect partial improvement after an experimental treatment. According to our observations in amblyopia caused by 3-week LTMD from P21 (Fig. 4), to achieve 80% power, hypothetical treatments restoring 100%, 75%, and 50% binocular function would require 42, 73, and 165 animals per group to statistically demonstrate an effect. We made multiple iterations of and adjustments to the VCA to address this limitation without success.

The PDCT is a recently developed approach to study deficits in binocular vision in amblyopic mice
While the VCA has a rich history of use in species from rodents to humans^[13,20,43], only one prior study has described and applied the PDCT to evaluate binocular vision in depth perception in mice^[29]. To our knowledge, we are the first to attempt to replicate their findings and use the assay to study the effect of amblyogenic rearing. Under normal lighting conditions, we found that animals preferentially exited the pole onto the safe quadrant, whereas animals tested in darkness did not (Fig. 5d). In agreement with Boone et al.^[29], we thus found that vision is required for targeting the safe platform in the PDCT. We additionally found that animals tested in darkness require longer times to descend to the pole (Fig. 5d). The absence of light likely makes the descent more challenging and slower by preventing mice from using vision to gauge distances, detect obstacles, and maintain balance. Additionally, without visual input, mice may rely more on other senses, such as tactile input from the paws and nose (with the whiskers trimmed), which could be less effective for quick navigation compared with visual guidance in a well-lit environment. Our observed increase in descent time is thus not surprising.

In our experiments, we utilized pupillary dilation and amblyogenic rearing to test whether balanced binocular vision is required for the PDCT. Both paradigms degraded descent time performance in the PDCT but did not affect the tendency of mice to exit onto the safe platform (Fig. 5d, e). Thus, while we found the tendency of mice to exit onto the safe quadrant to be dependent on vision, only descent time was dependent on balanced binocular vision. These findings differ from published results obtained using the PDCT^[29], where a lack of binocularity caused by acute monocular occlusion impaired the identification of the safe platform without affecting the descent time. These differences in behaviors may be attributed to how each assay engages visual processing and depth perception. Monocular occlusion eliminates binocular vision and substantially restricts the visual field. This absence of binocular function hinders the animal's ability to accurately judge distances between platforms and determine the safest quadrant. Despite this, mice with monocular vision can unambiguously detect contrast, light, and motion with normal acuity. As a result, monocular cues alone may provide sufficient information for spatial navigation, which may explain why descent time was unaffected in Boone et al.'s study^[29]. Unlike occlusion, monocular pupillary dilation introduces optical aberration in one eye, bringing the inputs from the two eyes into rivalry or conflict (which is resolved when both eyes receive atropine). Optical mismatch through pupillary dilation is unlikely to completely eliminate the binocular fusion required for depth perception but diminishes it somewhat. Similarly, and even though amblyopia following LTMD drastically affects the development of normal binocular vision, visual input from the deprived (amblyopic) eye is not completely eliminated the way it is following acute monocular occlusion. Closer to the platform, both monocular and binocular cues are larger and likely sufficient to enable mice to accurately identify and target the safe platform. These cues are more subtle (and smaller in the case of interocular angular disparity) when the animal initiates their descent higher on the pole. Less apparent monocular cues further from the platform increases reliance on binocular cues. This may explain why we found the metric of descent time to be more sensitive to manipulations of binocular vision such as monocular pupillary dilation and LTMD, and likely accounts for the discrepancies between our findings and those of Boone et al.^[29].

A key methodological difference between our implementation of the PDCT and that of Boone et al.^[29] concerns the degree of handling during the assay. In our paradigm, the mice were allowed to self-initiate descents without repositioning. By contrast, Boone et al.^[29] reported 10–15 trials per mouse, which were obtained through repeated manual repositioning. These two approaches focus on distinct behavioral readouts: Uninterrupted, self-initiated trials allow the assessment of naturalistic descent behaviors, whereas repeated repositioning supports quantification of fractional platform choice. Both studies provide strong evidence for an essential role of binocular vision for task performance. Future work examining subtle animal behavior along the pole and cone may reveal decision points indicative of distinct depth perception strategies.

Though we could not detect subtle imbalances in binocular vision caused by the application of atropine or LTMD using the metric "safe/unsafe exit", we found pole descent time to be sensitive enough to detect these subtle imbalances. As LTMD in mice has been the standard model used to study the pathophysiology and novel treatment approaches of amblyopia, our findings strongly support the use of descent time in the PDCT to study stereoscopic deficits in mice. The major advantage of using descent time in the PDCT is its relatively large effect size (37%–61%), which we found when utilizing atropine or LTMD to disrupt binocular vision (Fig. 5d, e). This necessitated the use of <15 animals to make reliable comparisons, even when binocular vision is only partially disrupted.

While the effect of LTMD was maintained, the values of descent times showed variability across experiments that differed only by a slight change in the apparatus (Fig. 5e, e'). In addition to the effect of altering the apparatus, different absolute measures likely reflect subtle cohort effects as well as differences in handling and the nature of manual scoring. Though our scoring methodology offers behavioral specificity, it is inherently subject to variation between experimenters and across sessions. Future studies that incorporate additional camera views and automated analysis will not only reduce variability but also allow the identification of decision points earlier in the descent, which could reveal how binocular cues are integrated over space and time and provide important insights into the depth perception strategies that guide this behavior. Nevertheless, we found that although exact descent times varied when the videos were scored by two independent, blinded observers, relative group differences were consistently observed. This underscores the importance of running interleaved controls—i.e., testing LTMD and control animals side by side under matched conditions—and having a consistent, group-blinded rater (ideally, more than one) for video scoring when evaluating descent time as a behavioral readout. This approach is critical for avoiding confounding batch-specific or experimenter-related effects and accurately measuring the effect size of LTMD or treatment interventions. Taken together, the strengths of our study include the systematic approach used to characterize VCA performance, replicate the PDCT, and determine their dependencies on binocular vision in mice. Our primary experiments are well-powered and thus provide a guide for future experiments to use these assays. The pattern of behavior was consistent across experiments in our study, in which manipulations that reduced binocular function decreased the time spent on the safe side in the VCA and increased descent time in the PDCT.

[1]	McKee SP, Levi DM, Movshon JA. 2003. The pattern of visual deficits in amblyopia. Journal of Vision 3(5):380−405 doi: 10.1167/3.5.5 CrossRef Google Scholar
[2]	Levi DM, Knill DC, Bavelier D. 2015. Stereopsis and amblyopia: A mini-review. Vision Research 114:17−30 doi: 10.1016/j.visres.2015.01.002 CrossRef Google Scholar
[3]	Birch EE. 2013. Amblyopia and binocular vision. Progress in Retinal and Eye Research 33:67−84 doi: 10.1016/j.preteyeres.2012.11.001 CrossRef Google Scholar
[4]	Li YT, Chou XL, Tao HW. 2014. Monocular deprivation in mice. Bio-Protocol 4:e1024 doi: 10.21769/bioprotoc.1024 CrossRef Google Scholar
[5]	Wiesel TN, Hubel DH. 1963. Single-Cell Responses in Striate Cortex of Kittens Deprived of Vision in One Eye. Journal of neurophysiology 26:1003−17 doi: 10.1152/jn.1963.26.6.1003 CrossRef Google Scholar
[6]	Duffy KR, Bear MF, Patel NB, Das VE, Tychsen L. 2023. Human deprivation amblyopia: treatment insights from animal models. Frontiers in Neuroscience 17:1249466 doi: 10.3389/fnins.2023.1249466 CrossRef Google Scholar
[7]	Sale A, Maya Vetencourt JF, Medini P, Cenni MC, Baroncelli L, et al. 2007. Environmental enrichment in adulthood promotes amblyopia recovery through a reduction of intracortical inhibition. Nature Neuroscience 10(6):679−681 doi: 10.1038/nn1899 CrossRef Google Scholar
[8]	Montey KL, Quinlan EM. 2011. Recovery from chronic monocular deprivation following reactivation of thalamocortical plasticity by dark exposure. Nature Communications 2:317 doi: 10.1038/ncomms1312 CrossRef Google Scholar
[9]	Fong MF, Duffy KR, Leet MP, Candler CT, Bear MF. 2021. Correction of amblyopia in cats and mice after the critical period. eLife 10:e70023
[10]	Fong MF, Mitchell DE, Duffy KR, Bear MF. 2016. Rapid recovery from the effects of early monocular deprivation is enabled by temporary inactivation of the retinas. Proceedings of the National Academy of Sciences of the United States of America 113(49):14139−14144 doi: 10.1073/pnas.1613279113 CrossRef Google Scholar
[11]	Grieco SF, Qiao X, Zheng X, Liu Y, Chen L, et al. 2020. Subanesthetic ketamine reactivates adult cortical plasticity to restore vision from amblyopia. Current Biology 30(18):3591−3603.e8 doi: 10.1016/j.cub.2020.07.008 CrossRef Google Scholar
[12]	Venturino A, Schulz R, De Jesús-Cortés H, Maes ME, Nagy B, et al. 2021. Microglia enable mature perineuronal nets disassembly upon anesthetic ketamine exposure or 60-Hz light entrainment in the healthy brain. Cell Reports 36(1):109313 doi: 10.1016/j.celrep.2021.109313 CrossRef Google Scholar
[13]	Gibson EJ, Walk RD. 1976. The "visual cliff". Scientific American 202(4):64−71 doi: 10.1038/scientificamerican0460-64 CrossRef Google Scholar
[14]	Kaufman LW. 1976. Duration of early visual experience and visual cliff behavior of chicks. Developmental Psychobiology 9(1):1−4 doi: 10.1002/dev.420090102 CrossRef Google Scholar
[15]	Rodkey EN. 2015. The visual cliff's forgotten menagerie: rats, goats, babies, and myth-making in the history of psychology. Journal of the History of the Behavioral Sciences 51(2):113−140 doi: 10.1002/jhbs.21712 CrossRef Google Scholar
[16]	Witherington DC, Campos JJ, Anderson DI, Lejeune L, Seah E. 2005. Avoidance of heights on the visual cliff in newly walking infants. Infancy 7(3):285−298 doi: 10.1207/s15327078in0703_4 CrossRef Google Scholar
[17]	Green PR, Davies IB, Davies MN. 1993. Interaction of visual and tactile information in the control of chicks' locomotion in the visual cliff. Perception 22(11):1319−1231 doi: 10.1068/p221319 CrossRef Google Scholar
[18]	Cornwell P, Overman W, Levitsky C, Shipley J, Lezynski B. 1976. Performance on the visual cliff by cats with marginal gyrus lesions. Journal of Comparative and Physiological Psychology 90(10):996−1010 doi: 10.1037/h0077281 CrossRef Google Scholar
[19]	Morrison PR. 1982. Distance cues and depth avoidance on the visual cliff. erceptual and Motor Skills 54:1195−1198 doi: 10.2466/pms.1982.54.3c.1195 CrossRef Google Scholar
[20]	Baroncelli L, Braschi C, Maffei L. 2013. Visual depth perception in normal and deprived rats: effects of environmental enrichment. Neuroscience 236:313−319 doi: 10.1016/j.neuroscience.2013.01.036 CrossRef Google Scholar
[21]	Sansevero G, Baroncelli L, Scali M, Sale A. 2019. Intranasal BDNF administration promotes visual function recovery in adult amblyopic rats. Neuropharmacology 145:114−122 doi: 10.1016/j.neuropharm.2018.02.006 CrossRef Google Scholar
[22]	Waugh KT. 1910. The rôle of vision in the mental life of the mouse. Journal of Comparative Neurology and Psychology 20(6):549−599 doi: 10.1002/cne.920200602 CrossRef Google Scholar
[23]	Samonds JM, Choi V, Priebe NJ. 2019. Mice discriminate stereoscopic surfaces without fixating in depth. The Journal of Neuroscience 39(41):8024−8037 doi: 10.1523/JNEUROSCI.0895-19.2019 CrossRef Google Scholar
[24]	Dräger UC. 1978. Observations on monocular deprivation in mice. Journal of neurophysiology 41(1):28−42 doi: 10.1152/jn.1978.41.1.28 CrossRef Google Scholar
[25]	Heesy CP. 2004. On the relationship between orbit orientation and binocular visual field overlap in mammals. The Anatomical Record 281A(1):1104−1110 doi: 10.1002/ar.a.20116 CrossRef Google Scholar
[26]	Scholl B, Burge J, Priebe NJ. 2013. Binocular integration and disparity selectivity in mouse primary visual cortex. Journal of neurophysiology 109(12):3013−3024 doi: 10.1152/jn.01021.2012 CrossRef Google Scholar
[27]	Williams B, Del Rosario J, Muzzu T, Peelman K, Coletta S, et al. 2021. Spatial modulation of dark versus bright stimulus responses in the mouse visual system. Current Biology 31(18):4172−4179.e6 doi: 10.1016/j.cub.2021.06.094 CrossRef Google Scholar
[28]	Mazziotti R, Baroncelli L, Ceglia N, Chelini G, Della Sala G, et al. 2017. Mir-132/212 is required for maturation of binocular matching of orientation preference and depth perception. Nature Communications 8:15488 doi: 10.1038/ncomms15488 CrossRef Google Scholar
[29]	Boone HC, Samonds JM, Crouse EC, Barr C, Priebe NJ, et al. 2021. Natural binocular depth discrimination behavior in mice explained by visual cortical activity. Current Biology 31(10):2191−2198.e3 doi: 10.1016/j.cub.2021.02.031 CrossRef Google Scholar
[30]	Han KS, Cooke SF, Xu W. 2017. Experience-Dependent Equilibration of AMPAR-Mediated Synaptic Transmission during the Critical Period. Cell Reports 18(4):892−904 doi: 10.1016/j.celrep.2016.12.084 CrossRef Google Scholar
[31]	Sansevero G, Torelli C, Mazziotti R, Consorti A, Pizzorusso T, et al. 2020. Running towards amblyopia recovery. Scientific Reports 10(1):12661 doi: 10.1038/s41598-020-69630-7 CrossRef Google Scholar
[32]	Baroncelli L, Scali M, Sansevero G, Olimpico F, Manno I, Costa M, et al. 2016. Experience affects critical period plasticity in the visual cortex through an epigenetic regulation of histone post-translational modifications. The Journal of Neuroscience 36(12):3430−3440 doi: 10.1523/jneurosci.1787-15.2016 CrossRef Google Scholar
[33]	Pennington ZT, Dong Z, Feng Y, Vetere LM, Page-Harley L, Shuman T, et al. 2019. ezTrack: An open-source video analysis pipeline for the investigation of animal behavior. Scientific Reports 9(1):19979 doi: 10.1038/s41598-019-56408-9 CrossRef Google Scholar
[34]	Faul F, Erdfelder E, Lang AG, Buchner A. 2007. GPower 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods* 39(2):175−191 doi: 10.3758/BF03193146 CrossRef Google Scholar
[35]	Pennington ZT, Diego KS, Francisco TR, LaBanca AR, Lamsifer SI, et al. 2021. ezTrack-A Step-by-Step Guide to Behavior Tracking. Current Protocols 1(10):e255 doi: 10.1002/cpz1.255 CrossRef Google Scholar
[36]	Jiménez JR, Castro JJ, Jiménez R, Hita E. 2008. Interocular differences in higher-order aberrations on binocular visual performance. Optometry and Vision Science 85(3):174−179 doi: 10.1097/OPX.0b013e31816445a7 CrossRef Google Scholar
[37]	Gomes J, Sapkota K, Franco S. 2023. Central and peripheral ocular high-order aberrations and their relationship with accommodation and refractive error. A review. Vision 7(1):19 doi: 10.3390/vision7010019 CrossRef Google Scholar
[38]	Hooks BM, Chen C. 2020. Circuitry Underlying Experience-Dependent Plasticity in the Mouse Visual System. Neuron 106(1):21−36 doi: 10.1016/j.neuron.2020.01.031 CrossRef Google Scholar
[39]	Legge GE, Gu YC. 1989. Stereopsis and contrast. Vision Research 29(8):989−1004 doi: 10.1016/0042-6989(89)90114-4 CrossRef Google Scholar
[40]	Westheimer G, McKee SP. 1980. Stereogram design for testing local stereopsis. Investigative Ophthalmology & Visual Science 19(7):802−809 Google Scholar
[41]	Ellard CG, Goodale MA, Scorfield DM, Lawrence C. 1986. Visual cortical lesions abolish the use of motion parallax in the Mongolian gerbil. Experimental Brain Research 64(3):599−602 doi: 10.1007/BF00340498 CrossRef Google Scholar
[42]	Ellard CG, Goodale MA, Timney B. 1984. Distance estimation in the Mongolian gerbil: the role of dynamic depth cues. Behavioural Brain Research 14(1):29−39 doi: 10.1016/0166-4328(84)90017-2 CrossRef Google Scholar
[43]	Fox MW. 1965. The visual cliff test for the study of visual depth perception in the mouse. Animal Behaviour 13(2):232−233 doi: 10.1016/0003-3472(65)90040-0 CrossRef Google Scholar

{{lists.name}}

Using the visual cliff and pole descent assays to detect binocular disruption in mice

Abstract

Supplementary information

Rights and permissions

References

About this article

Cite this article

Article Metrics

Access History

Other Articles By Authors