The study, led by Tristan Loria with colleagues including Kanji Tanaka, Joëlle Hajj, Katsumi Watanabe, and Luc Tremblay, exploited one of the most famous tricks in perception science: the sound-induced flash illusion. In its classic form, a single flash of light accompanied by two brief beeps is often perceived as two flashes, an error known as the fission illusion. The complementary version, the fusion illusion, occurs when two flashes paired with a single beep collapse perceptually into one. First described by Shams, Kamitani, and Shimojo in 2000, the illusion demonstrates that audition can literally rewrite what we think we see, and it has since become a standard probe for measuring how the brain combines information across the senses.

For decades, the dominant view held that such cross-modal binding occurs automatically whenever stimuli are spatially and temporally congruent, following principles such as near-optimal Bayesian integration, in which the brain weights each sense according to its reliability. But a growing body of evidence has complicated that picture. Attention, cognitive load, expectation, and even clinical conditions such as glaucoma, migraine, and autism spectrum disorder all modulate how strongly the illusion grips the observer. The new study adds a surprising factor to that list: the body’s own goal-directed movements.

The researchers designed two experiments. In the first, participants performed pointing movements toward visual targets while unimodal visual stimuli, congruent audiovisual pairs, or illusion-inducing audiovisual combinations were presented either at the target location itself or at adjacent non-target locations, both close to and far from the target. After completing each movement, participants reported how many flashes they had seen, giving the team a precise measure of perceptual accuracy for each stimulus type at each location. The logic was straightforward: if multisensory perception is uniform across space, accuracy should not depend on where the stimuli appeared relative to the movement goal.

That is not what happened. The results revealed a clear spatial gradient in perceptual accuracy. Bimodal, congruent audiovisual stimuli were processed more accurately at the target location than at either the close or far non-target locations, indicating that the perceptual benefits of combining sight and sound are concentrated where the action is directed. More intriguing still was the pattern for the fusion illusion. Participants were more susceptible to the illusion, meaning lower perceptual accuracy, at the target location, while accuracy was higher at the nearby non-target positions. In other words, the very spot the brain prioritizes for action became the spot where sound most powerfully distorted vision, collapsing two flashes into one.

The second experiment asked whether this modulation depends on actually moving, or whether merely preparing an action is enough. Participants again pointed at targets while the same battery of stimuli was delivered at the target location, but the researchers compared trials in which the movement was executed against a no-movement condition. The answer was unambiguous: the modulation of multisensory perception at the target location appeared only during movement itself, not when participants remained still. This temporal specificity suggests that the efferent signals and sensorimotor processes accompanying active movement, rather than attentional preparation alone, are what reshape the audiovisual percept.

Perhaps the most technically revealing aspect of the study is a double dissociation hiding in the data. Only the fusion illusion, in which two flashes merge into one, was sensitive to both the spatial gradient and the movement manipulation. The fission illusion, in which one flash splits into two under the influence of two beeps, remained remarkably stable across all locations and both action states. This asymmetry aligns with neurophysiological evidence that the two illusions, despite their mirror-image phenomenology, rely on distinct underlying mechanisms. Prior work using electroencephalography has traced fusion to early cross-modal interactions in auditory and visual cortex, while fission appears to engage different, possibly later, processing stages. The new behavioral results suggest that whatever mechanism goal-directed action recruits, it taps into the circuitry underlying fusion but leaves fission untouched.

The findings resonate with a broader literature on action-centered attention and peripersonal space. Research on selective reaching has shown that attention is organized around the body and its intentions rather than around retinal coordinates, and studies of saccade preparation have demonstrated that attention shifts to the goal of an eye movement before it lands. Earlier work by Tremblay and colleagues had already shown decreased sensitivity to an audiovisual illusion during goal-directed reaching, and a 2020 study by Loria, Tanaka, Watanabe, and Tremblay found that deploying attention to a pointing target modulates audiovisual processing even at non-target locations. The new experiments sharpen that picture by separating spatial proximity from movement execution, showing that the two factors interact: space defines where perception is modulated, and action defines when.

Why would the brain deliberately make itself more vulnerable to an illusion at the location it is reaching toward? One plausible interpretation comes from reliability-weighting models of multisensory perception. Peripheral vision is coarse and unreliable, and when visual information is degraded, the brain leans more heavily on the more precise auditory signal, which is exactly the condition under which fusion illusions flourish. During a goal-directed action, resources are funneled toward the target to guide the hand, and the resulting perceptual economy may amplify the relative weight of sound at that location. The paradoxical outcome is that the behaviorally most relevant spot in space becomes the spot where the senses merge most aggressively, a trade-off between sharpening an action plan and accepting perceptual distortion.

The implications extend well beyond the laboratory. Multisensory integration underpins everything from speech perception and virtual reality design to rehabilitation after stroke and the performance of surgeons, pilots, and athletes working in cluttered, noisy environments. If the fidelity of audiovisual binding shifts dynamically with our actions, then interface designers, clinicians, and human-factors engineers may need to account for a moving, action-dependent perceptual landscape rather than a static one. The authors note that the raw data will be made available upon reasonable request, supporting verification and secondary analysis. For now, the message is vivid and a little unsettling: the world you perceive is not a fixed rendering but a construction that your own movements edit in real time, bending sight around sound precisely where you intend to act.

Subject of Research: Modulation of audiovisual multisensory integration and the sound-induced flash illusion by goal-directed action

Article Title: Goal-directed action modulates audiovisual multisensory perception

Article References: Loria, T., Tanaka, K., Hajj, J., Watanabe, K., & Tremblay, L. (2026). Goal-directed action modulates audiovisual multisensory perception. Attention, Perception, & Psychophysics, 88(7), Article 201. https://doi.org/10.3758/s13414-026-03337-1

Image Credits: AI Generated

DOI: 10.3758/s13414-026-03337-1

Keywords: multisensory integration, sound-induced flash illusion, audiovisual perception, goal-directed action, pointing movements, spatial attention, peripersonal space, fusion illusion, fission illusion, sensorimotor processing, psychophysics, attention

The task the researchers chose is the additional singleton paradigm, one of the most widely used tools in attention research. Participants search for a uniquely shaped target, such as a pear among apples, while a salient distractor, here a red object among green ones, appears on half of the trials. In traditional versions, people respond more slowly and make more errors when the distractor is present, a phenomenon known as attentional capture. The salient stimulus grabs attention automatically, even though it is irrelevant to the task. What has been missing is a way to watch that capture unfold in three dimensions, in movements that resemble the reaching and looking we do every day.

Fifty-four participants aged 18 to 26 wore an HTC Vive Pro Eye headset fitted with integrated eye tracking and held a motion-tracked controller in their dominant hand. The system recorded hand positions at roughly 90 hertz and gaze direction from the headset’s infrared eye tracker, allowing the team to capture reaching and looking behavior at the same time in the same trial. Participants completed the task in two contrasting virtual environments: a sparsely detailed, untextured room designed to mimic a bare laboratory, and a richly detailed virtual greenhouse filled with visual clutter. The order of the environments was counterbalanced across participants.

On each trial, four fruit-shaped stimuli appeared on a virtual tree at the top, left, right, and bottom locations. The target was always the unique shape, and its color and identity were randomized from trial to trial, encouraging participants to search for any singleton rather than a specific feature. When a distractor appeared, it was always the uniquely colored object. Participants initiated each reach by holding the controller inside a central starting sphere for two seconds, then had three seconds to move to the target. Correct responses earned a pleasant 600-hertz tone; errors or timeouts produced a low 300-hertz buzz.

The results were striking in their consistency. When a distractor was present, participants were slower to initiate their movements, slower to complete them, and slower overall. Their reach trajectories curved more, their hands spent a greater proportion of each movement being pulled toward the distractor location, their eyes traveled greater angular distances, and they took longer to fixate the target for the first time. Effect sizes were substantial, with distractor presence explaining between roughly a quarter and nearly two-thirds of the variance in individual measures. In short, the attentional capture effects documented over thirty years of button-press experiments generalize robustly to three-dimensional immersive environments.

But the study’s real methodological contribution lies in how the movements were analyzed. Traditional reach-tracking studies collapse three-dimensional trajectories into two-dimensional summary statistics such as maximum perpendicular deviation from a straight path. The problem is that a large deviation does not necessarily mean the hand was drawn toward the distractor; it could also reflect a delayed decision, with the hand drifting toward the center of the display before committing. Gurr and colleagues instead computed continuous target attraction and distractor attraction scores: for every time point in a movement, they determined which object the hand was moving closest toward, and calculated the proportion of the trajectory devoted to each. These measures revealed capture effects that the curvature measure missed entirely, particularly on trials where the distractor appeared far from the target.

Those attraction scores also allowed the team to reconstruct the time course of capture using cluster-based permutation analyses. When the distractor was adjacent to the target, the hand was pulled toward the distractor from the very first moments of movement, with the effect persisting for up to 78 percent of the trajectory before the movement was corrected. Even when the target and distractor sat on opposite sides of the display, the hand still showed significant early attraction to the distractor, though the effect was briefer and weaker. This graded pattern mirrors earlier findings that capture costs shrink as the distance between target and distractor grows, and it extends them into a fully three-dimensional movement space.

The comparison between the sparse room and the rich greenhouse produced a subtler story. Participants were generally slower and moved their eyes more in the detailed greenhouse, consistent with prior work showing that cluttered environments impair visual search. Yet the size of the attentional capture effect itself did not differ between environments. Bayesian analyses provided moderate evidence for this null result on several measures, though the evidence was mixed on others. One complication emerged: participants who started in the sparse room performed nearly identically in both environments, while those who started in the rich greenhouse were markedly faster in the sparse room afterward, hinting at practice or adaptation effects that future studies will need to disentangle.

The authors are candid about the limitations. Many participants adopted a rigid strategy of looking first at the top location, which complicated the analysis of initial eye movements. The task’s fixed layout may have allowed participants to learn that the bottom object revealed the target’s color, potentially shrinking the distractor effect. And VR hardware brings its own caveats: the headset’s eye tracker is less precise than professional laboratory systems, and the motion-to-photon lag between a real hand movement and its virtual rendering, estimated at a few tens of milliseconds, means VR-derived timing should be compared cautiously with data from electromagnetic or optical trackers.

Even so, the study makes a compelling case that a commercially available VR headset, costing a fraction of traditional tracking rigs, can deliver laboratory-grade measurements of how distraction plays out simultaneously in the hand and the eye. Because the equipment is portable and the software tools are increasingly accessible, the approach could spread quickly through labs studying attention, perception, and motor control. The data and analysis scripts have been released openly, and the team plans to publish user-friendly processing software. The gap between the psychology laboratory and the cluttered, three-dimensional world we actually inhabit may finally be starting to close.

Subject of Research: Measuring attentional capture dynamics in hand and eye movements using virtual reality

Article Title: Measuring the dynamics of attentional capture in hand and eye movements with virtual reality

Article References: Gurr, M. D., Moher, J., Egbert, M. D., & Erb, C. D. (2026). Measuring the dynamics of attentional capture in hand and eye movements with virtual reality. Behavior Research Methods, 58(11), Article 302. https://doi.org/10.3758/s13428-026-03177-9

Image Credits: AI Generated

DOI: 10.3758/s13428-026-03177-9

Keywords: attentional capture, virtual reality, eye tracking, hand tracking, additional singleton task, visual attention, reaching movements, distractors, Behavior Research Methods, sensorimotor processing, visual search, immersive environments