How Tiny Eye Movements Reveal the Fragility of Visual Perception

What does it take to not see something that is actually there? Our latest research, led by graduate student Max Levinson in collaboration with McGill Prof. Chris Pack, and now published in open-access in the Journal of Vision, explores the surprising dynamics of visual disappearance—and the unexpected role of tiny, involuntary eye movements in shaping what we consciously perceive.

In the study, we investigated a compelling visual illusion known as perceptual filling-in. When we fixate steadily on a boundary between two adjacent surfaces, that boundary can begin to vanish from view, making the two areas appear to blend into one. Crucially, this illusion does not arise from any change in the stimulus itself—it emerges entirely from the brain’s own processes. That makes filling-in a powerful window into how the brain generates conscious visual experience.

Our central question: What causes the boundary to disappear—and what interrupts that process?

Using high-resolution eye-tracking and a carefully controlled visual display, we found that microsaccades—the small, spontaneous eye movements we make even while trying to keep our gaze still—play a central role. We found that the more often a person makes microsaccades, the less likely they are to experience filling-in. In fact, we show that these tiny movements reset the illusion: the momentary disruption they cause reactivates boundary representations, keeping perception veridical—that is, aligned with the actual visual input rather than the brain’s internal predictions or illusions.

We also discovered that this interaction is temporally precise. The timing of microsaccades relative to the subjective onset of filling-in reveals that the visual system is continuously balancing between adaptation (which favors disappearance) and refresh (which favors stable perception). Simply put, whether a boundary fades or remains visible depends on a dynamic push-pull between cortical adaptation and oculomotor correction.

Microsaccade dynamics and their influence on filling-in time (FT). (A) Relationship between amplitude and peak velocity of individual microsaccades during main trials, demonstrating adherence to the classic main sequence relationship. (B) Average normalized microsaccade rate in the period leading up to participants’ report of perceptual filling-in. (C) FT as a function of color contrast, comparing trials with at least one microsaccade (red) to trials without microsaccades (blue). (D) FT as a function of boundary eccentricity, comparing trials with at least one microsaccade (red) to trials without microsaccades (blue). Each data point represents an individual participant. Boxplots show the median, 25th, and 75th percentiles of participants’ means.

These findings not only help explain why our percepts can fluctuate despite stable inputs—they also offer a compelling example of how motor signals influence sensory experience, even at the scale of unconscious eye movements. The study sets the stage for our subsequent MEG-based work, which uncovers the brain-wide dynamics underlying these perceptual transitions, and which we are currently preparing for publication—stay tuned!

—

The full paper is available open-access in the Journal of Vision:
"Microsaccades counteract visual adaptation and delay perceptual filling-in"
https://doi.org/10.1167/jov.25.8.8