Binocular rivalry

Binocular rivalry is a form of visual competition in which incompatible images presented to corresponding regions of the two eyes produce changing percepts. One image can dominate awareness while the other is suppressed, followed by a reversal, although the physical images remain unchanged. A conventional demonstration presents a different picture or grating orientation to each eye.

Reviewed 21 October 2026. Ten original perception papers, including study exclusions and reporting-condition limits About review dates.

The phenomenon concerns how the visual system selects between conflicting inputs. For AR displays, the visibility of information can depend on whether it is supplied to one eye or both eyes and how it overlaps the background. VR also provides a way to study rivalry in controlled three-dimensional scenes.

Perceptual experience

Rivalry does not always alternate between two complete pictures. Different parts of a larger image can dominate at the same time, producing a changing patchwork called piecemeal rivalry. Dominance can also be coordinated across separate regions, so that the visible pattern reflects relationships between local image features and their eye of origin.

Rivalry is one possible outcome of binocular vision, rather than a name for every difference between the eyes' images. The following distinctions matter when describing a headset or experiment.

Term	Meaning and distinction
Binocular fusion	Signals from the two eyes contribute to a single percept. In a study of blurred edges, fusion, suppression and double vision were treated as different perceptual outcomes, with their occurrence depending on disparity, blur and contrast.
Diplopia	Both non-fused images remain visible as separate images. This differs from suppression of one input.
Interocular suppression	An input from one eye becomes less visible or unavailable to awareness. A suppressed input is not the same as a physically absent image; alternating dominance is the characteristic rivalry pattern.
Monocular rivalry	Superimposed patterns can compete even within one eye's image. A see-through display can therefore involve competition within an image as well as competition between the eyes.

A monocular display describes which eye receives the added imagery. It does not mean that its wearer views the entire scene with only one eye: in a monocular see-through arrangement, both eyes can see the background while only one receives the overlay.

Dynamics and measurement

Experiments distinguish the length of an individual dominance period from predominance, the proportion of viewing time assigned to a percept. Mixed states need separate treatment. For example, a 2018 experiment recorded exclusive horizontal or vertical gratings, patchwork percepts and superimposed percepts, rather than assuming that one complete image was always visible.

Contrast-based relationships often called Levelt's propositions describe how stimulus strength affects predominance, dominance durations and alternation rate. Their applicability depends on the stimuli. Skerswetat and colleagues found that the expected relationships did not describe competition between their luminance-modulated and contrast-modulated gratings; the latter varied local contrast rather than mean luminance. Simply making one pattern more detectable did not guarantee that it would dominate.

Eye tracking can provide complementary measures under suitable conditions. In a 2014 study with 20 observers, Frässle and colleagues used the slow phase of optokinetic nystagmus to track perceived motion direction and pupil size to track rivalry between stimuli of different luminance. These signals were compared with button reports and with physical replay of image changes. The measures were imperfect and depended on the chosen stimuli; they were not a general decoder for arbitrary visual content.

The distinction between perception and reporting also affects interpretation. In that study, active reporting altered rivalry dynamics for the stationary gratings, and some rivalry-related brain activity was reduced when participants viewed passively. Thus a button-tracking experiment also involves attention, monitoring and action.

Competition between eyes and patterns

An exclusively eye-based account cannot explain every result. In 1996, Kovács and colleagues divided rival images into complementary patchworks. Observers could sometimes perceive coherent forms assembled from pieces delivered to both eyes, rather than only the complete patchwork supplied to either eye. Their color experiments likewise produced grouped percepts across the eyes.

Eye of origin nevertheless remains relevant. Stuit and colleagues' 2011 experiments with pairs of rival gratings found stronger grouping for targets shown to the same eye. For horizontal and vertical gratings, shared orientation also affected grouping durations; the same orientation effect was not established for their oblique gratings. Their results support contributions from both eye of origin and image content, with effects depending on the arrangement.

Computational models formalize possible mechanisms. Li and colleagues' 2017 model combined relatively fast mutual inhibition between eye-specific signals with slower, feature-selective attentional feedback and adaptation. Depending on its inputs and parameters, it reproduced conventional alternation, different outcomes when images were rapidly swapped between eyes, and contrast-dependent dynamics. These were model simulations evaluated against experimental observations, rather than identification of a single anatomical switch responsible for every rivalry percept.

Research in XR

Naturalistic geometry in VR

Han and colleagues used an Oculus Quest 2 and a virtual corridor to compare conflicting gratings placed at the same depth, separated in depth, or arranged as a plausible partial occlusion. Their 2024 study involved 17 people: 14 in the first experiment and eight in the second, including five who took part in both. Participants tracked static or moving gratings during two-minute trials. Rivalry persisted under the plausible occlusion geometry.

The study isolated geometry rather than reproducing unrestricted everyday viewing. Participants did not navigate the scene, the second experiment stabilized their heads with a chin rest, fixation accuracy was not quantified, and differential optical defocus was not simulated. The findings establish rivalry under those controlled naturalistic geometric conditions; they do not establish its frequency during ordinary activity.

Information visibility in wearable displays

Kim and colleagues simulated three peripheral display configurations in an HTC Vive Pro Eye: monocular opaque, monocular see-through and binocular see-through. Thirteen people were recruited; the reported analyses used 11 after eye-tracking collection problems. In their moving-corridor task, binocular presentation produced higher target visibility than either monocular condition. Guided saccades and smooth pursuit also increased visibility relative to central fixation.

These were simulated display configurations and a specific peripheral grating task, rather than tests of physical optical combiners or real-world walking and driving. The monocular see-through condition also changed the target's effective contrast, limiting attribution of the whole difference to one rivalry mechanism.

Bayle and colleagues' 2021 study used stereoscopic projectors and active glasses to reproduce a monocular AR overlay for 18 participants. Eye-tracking analyses of latency and event fixation included 17 after one participant was excluded because of measurement noise. Background characteristics and overlay presentation affected target tracking and event-detection measures, but reported losses of information were uncommon. The authors noted that their small ring-and-circle stimuli may have limited suppression, and that larger symbology needed further testing. Interocular conflict, measured task performance and reported perceptual suppression therefore should not be treated as interchangeable outcomes.

Presenting two virtual environments

The 2026 ParaSights research system intentionally presented different environments to the two eyes using a Meta Quest 3. Yamamoto, Morimoto and Watanabe compared it with a split-field presentation while 12 experienced VR users performed tracking and ball-catching tasks. They reported higher task success and presence for ParaSights under that comparison. However, the baseline allocated a smaller field of view to each environment, and the study included neither an overlay baseline nor a single-environment control. It did not directly measure rivalry alternation rates, dominance durations or suppression depth. The results therefore concern a particular dual-task presentation experiment, rather than demonstrating that rivalry generally improves multitasking or establishing its perceptual mechanism.