Interocular suppression weakens binocular disparity integration in the dorsal visual stream
Curation statements for this article:-
Curated by eLife
eLife Assessment
This important study uses multi-session fMRI and multivariate analyses to investigate how interocular suppression affects binocular disparity representations across the human visual cortex. The authors provide convincing evidence that interocular suppression weakens disparity-related information in the dorsal visual stream, particularly in V3A and IPS. The findings further suggest that suppression alters anticorrelated binocular disparity representations and reduces coordinated disparity-related information across dorsal visual areas. Together, these results advance our understanding of how binocular integration and interocular suppression interact and highlight the involvement of higher-level dorsal visual regions in these processes.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
A unified percept depends on the coordinated interplay of sensory integration and suppression, yet how this interaction is implemented across cortical representations remains unclear. Using binocular vision as a tractable model system, we combined fMRI with multivariate analyses to test how interocular suppression reshapes binocular disparity integration across human visual pathways. Local binocular rivalry weakened near–far disparity decoding, with disruption increasing along the dorsal pathway from V1 to V3A and intraparietal sulcus (IPS), but no reliable loss in ventral regions. In V3A, anticorrelated random-dot stereograms showed an inverted representational relationship relative to correlated stimuli, and this structure was abolished under suppression, indicating disrupted non-canonical disparity integration. Informational connectivity further revealed that suppression selectively weakened V3A-centered coordination while largely sparing coupling among early visual areas. Together, these findings identify higher-level dorsal regions as a key substrate for coordinating integrative and suppressive computations during unified percept formation.
Article activity feed
-
eLife Assessment
This important study uses multi-session fMRI and multivariate analyses to investigate how interocular suppression affects binocular disparity representations across the human visual cortex. The authors provide convincing evidence that interocular suppression weakens disparity-related information in the dorsal visual stream, particularly in V3A and IPS. The findings further suggest that suppression alters anticorrelated binocular disparity representations and reduces coordinated disparity-related information across dorsal visual areas. Together, these results advance our understanding of how binocular integration and interocular suppression interact and highlight the involvement of higher-level dorsal visual regions in these processes.
-
Reviewer #1 (Public review):
Summary:
This study aims to study stereoscopic vision under conditions of binocular rivalry by using "functional magnetic resonance imaging (fMRI) to examine how interocular suppression reshapes disparity processing".
Strengths:
The strategic goal of being able to study stereoscopic vision when one or the other eye is dominant in rivalry is of potential interest.
Weaknesses:
There is evidence from earlier publications that pure eye rivalry (complete dominance of perception by one eye) is difficult to achieve in a controlled way. The authors present two types of random dot stereograms to subjects in the MR scanner under two different conditions. The stereograms are annular regions that surround a central disk, which contains grating patterns that, for condition 1, have the same orientation presented to the …
Reviewer #1 (Public review):
Summary:
This study aims to study stereoscopic vision under conditions of binocular rivalry by using "functional magnetic resonance imaging (fMRI) to examine how interocular suppression reshapes disparity processing".
Strengths:
The strategic goal of being able to study stereoscopic vision when one or the other eye is dominant in rivalry is of potential interest.
Weaknesses:
There is evidence from earlier publications that pure eye rivalry (complete dominance of perception by one eye) is difficult to achieve in a controlled way. The authors present two types of random dot stereograms to subjects in the MR scanner under two different conditions. The stereograms are annular regions that surround a central disk, which contains grating patterns that, for condition 1, have the same orientation presented to the left and right eyes (concordant) but, for condition 2, have different orientations, 90 degrees apart, so that rivalry is perceived. In the centre of the visual field, a binocular fixation target is presented, which subjects must monitor for brief changes in orientation.
The idea is that when the grating pair is rivalrous, there may be some measurable disruption of stereo responses, on the reasonable assumption that non-rivalrous binocular input would favour accurate stereo vision. Thus, it is argued that eye rivalry would be disruptive for stereo vision. There are two phenomena that intrude upon the goal of achieving pure eye rivalry. The first is methodological: using the paradigm of obliging the subject to view a rivalrous stimulus, whilst testing the response to stereo stimuli, typically results in a mixture of eye rivalry (which is relevant for this study) and pattern rivalry (which is an unwanted intrusion). Simply viewing a pair of patterns with different inputs to the left and right eyes does not guarantee that the rivalrous state is driven by eye rivalry (Kovács et al., 1996; Logothetis et al., 1996; Blake, 2001). It may be driven by pattern rivalry, which is a separate phenomenon that is not expected to have any simple predictable effect on a spatially adjacent stereo stimulus.
The second issue with the study design is that there are earlier experimental investigations that show that rivalrous binocular stimuli do not simply switch between one fully formed percept and its alternative. There are intermediate states in which different portions of the visual field are dominated by one percept whilst other parts of the visual field are dominated by its alternative (Blake, 2001). In some studies, the case has been made that there are travelling waves of alternation that pass over the visual field (Wilson et al., 2001). Thus, the rivalrous stimulus is often reported as fragmented in space and time. On the basis of these previously published studies, it is entirely possible for there to be rivalry in the region occupied by the pair of gratings and adequate binocular fusion in the regions with the random-dot stereograms. As the main result is that the presence of rivalrous stimuli in one part of the visual field has at best a weak effect on processing of binocularly correlated random dot patterns in a different part of the visual field, it seems important to take this issue into account in the interpretation of results from this study.
Another area of concern is that there are previous reports that the pattern of microsaccades differs between rivalrous and non-rivalrous states (Sabrin and Kertesz, 1980). There are no measures of eye movements in this study, but microsaccades are very effective in stimulating V1 and later dorsal stream areas. Extra activations of cortex during the rivalrous state could decrease the ability of a near/far decoder to report stereo depth in the stimulus, because the extra activations create extra MR-signals that are not relevant for near-far decoding.
These extra MR-signals, for example, might be driven by conjugate movements of the binocular stimulus across both eyes' retinas or by disconjugate movements that cause irrelevant changes in near/far neuronal activations. In the latter case, as these changes are not time-locked to the disparity changes in the random dot figures, they represent an intrusive signal to a near/far decoder. Further intrusive signals interfering with near/far decoding may arise when binocular tuned-inhibitory neurons become activated by small misconvergences of binocular eye position.
This discussion also highlights a point that the near/far decoder is sensitive only to particular organizations of stereo-sensitive units within the cortex. The decoder is most effective when different MR voxels are sensitive to near as opposed to far disparities. This means that the decoder essentially favours an odd symmetric sensitivity to disparity with respect to the zero-disparity fixation plane. Even-symmetric sensitivity (classically tuned-excitatory or tuned inhibitory sensitivity---(Fischer et al., 1979; Prince et al., 2002) is not probed with this decoder.
All these issues mean that many of the main claims of the paper within the current draft (importance of dorsal areas, differences based on interocular suppression, selectively weakened V3A-centered coordination) are called into question.
Literature Cited
Blake R (2001) A Primer on Binocular Rivalry, Including Current Controversies. Brain and Mind 2:5-38.
Fischer B, Poggio GF, Lennie P (1979) Depth Sensitivity of Binocular Cortical Neurons of Behaving Monkeys [and Discussion]. Proceedings of the Royal Society of London Series B, Biological Sciences 204:409-414.
Kovács I, Papathomas TV, Yang M, Féher Á (1996) When the brain changes its mind: interocular grouping during binocular rivalry. Proceedings of the National Academy of Sciences of the United States of America 93:15508-15511.
Logothetis NK, Leopold DA, Sheinberg DL (1996) What is rivalling during binocular rivalry? Nature 380:621-624.
Prince SJD, Cumming BG, Parker AJ (2002) Range and mechanism of encoding of horizontal disparity in macaque V1. Journal of Neurophysiology 87:209-221.
Sabrin HW, Kertesz AE (1980) Microsaccadic eye movements and binocular rivalry. Perception & Psychophysics 28:150-154.
Wilson HR, Blake R, Lee S-H (2001) Dynamics of travelling waves in visual perception. Nature 412:907-910.
-
Reviewer #2 (Public review):
Summary:
The authors used multivariate and multi-session fMRI approaches to address an important question: how the brain coordinates binocular rivalry with binocular integration of disparity information. Compared to binocular fusion, binocular rivalry stimuli presented in the central visual field disrupted the disparity representation of a surround stimulus composed of random-dot stereograms (RDS) in V3A and IPS of the dorsal visual stream. This disruption to binocular disparity was observed not only for correlated RDS, but also for anti-correlated RDS in V3A, which shows an inverted disparity representation. Finally, informational connectivity analysis suggests weakened information flow between V3A and other visual areas. The authors conclude that V3A and IPS of the dorsal visual stream serve as key regions …
Reviewer #2 (Public review):
Summary:
The authors used multivariate and multi-session fMRI approaches to address an important question: how the brain coordinates binocular rivalry with binocular integration of disparity information. Compared to binocular fusion, binocular rivalry stimuli presented in the central visual field disrupted the disparity representation of a surround stimulus composed of random-dot stereograms (RDS) in V3A and IPS of the dorsal visual stream. This disruption to binocular disparity was observed not only for correlated RDS, but also for anti-correlated RDS in V3A, which shows an inverted disparity representation. Finally, informational connectivity analysis suggests weakened information flow between V3A and other visual areas. The authors conclude that V3A and IPS of the dorsal visual stream serve as key regions coordinating interocular competition and integration in binocular vision.
Strengths:
The study used a factorial design to manipulate binocular disparity and binocular competition independently in the same fMRI experiment. A large amount of fMRI data was acquired for each participant from multiple sessions to improve the reliability of results at the individual level. Several multivariate methods, including multivoxel decoding and cross decoding, decision-angle analysis, and informational connectivity, were applied appropriately to address the same question from different angles. The results were quite clear, especially the increasing rivalry cost to disparity decoding along the visual hierarchy shown in Figure 2B. The manuscript is well written.
Weaknesses:
(1) The number of participants (n = 8) is extremely small for a conventional 3T fMRI study. While the Ng et al. 2021 study used the same number of participants, it is ultra-high-field fMRI that has much higher signal sensitivity and specificity than 3T fMRI. Thus, it cannot be used to justify the low number of participants in the current study (Line 339).
(2) While a large amount of data (5 hours of fMRI) has been acquired for each participant from multiple sessions, there is no demonstration of whether the results were reproducible across sessions or reliable within each participant.
(3) The retinotopic ROI for the annulus of the disparity stimulus was not accurate. While they used a retinotopic localizer (Line 450), the contrast between the annulus stimulus and fixation is not location-specific and will certainly activate brain regions corresponding to the central binocular grating stimuli. This contamination of signal from the central stimulus is more severe in higher-order brain regions like V3A and IPS, which could explain the disruption to disparity decoding in these regions.
(4) This is extremely worrisome because there was no behavioral effect or rivalry cost on RDS stimulus surrounding the rivalry stimuli. The rationale for studying the interaction between binocular rivalry competition and binocular disparity integration using two different stimuli at different locations is not explained in the manuscript.
(5) The no-report paradigm cannot be justified by the claim that anti-correlated RDS failed to produce depth perception (Line 76). Their behavioral results in Figure 1C clearly show that participants reported inverted depth to anti-correlated RDS.
(6) MRI data acquisition. The authors mentioned parallel imaging (GRAPPA = 2) to reduce distortion, but there is still missing important information about image distortion. Has partial Fourier been used? What is the pixel bandwidth and echo spacing?
(7) MRI preprocessing. The preprocessing section does not describe the method used for spatial alignment of functional data across sessions. This is important because all data were combined for multivariate analyses, and alignment errors arising from inaccurate methods would greatly impact the accuracy of the results.
(8) One-sided tests were not justified. Many important claims were supported by one-sided tests, including the rivalry effect in the correlated condition (CF vs. CR) and informational connectivity (vs. 0). This is suspicious and unusual. The authors must justify why they chose one-sided tests for their main results.
(9) 12 disparity conditions were combined into 2 conditions for analysis, but the authors did not explain why. They could potentially look at the rivalry cost to disparity tuning in these brain regions.
-
-
-
Interocular suppression weakens binocular disparity integration in the dorsal visual stream https://www.biorxiv.org/content/10.64898/2026.09.11.751085v1
-
Interocular suppression weakens binocular disparity integration in the dorsal visual stream https://www.biorxiv.org/content/10.64898/2026.09.11.751085v1
-
