Glance from one eye to the other in a mirror and the movement goes missing, though anyone watching you sees it plainly, because the brain turns down part of vision while the eyes are in flight.

Look into a mirror and look at one of your own eyes. Then look at the other. The eye makes the trip. You will not see it happen. Anyone standing beside you will see it without any difficulty at all.

That asymmetry is the whole phenomenon, and it is stated plainly in an overview of saccadic masking: "The eyes can never be observed in motion, yet an external observer clearly sees the motion of the eyes."

Saccadic masking, also known as saccadic suppression, is described in that overview as the brain selectively blocking visual processing during eye movements, in such a way that neither the motion of the eye and the resulting blur, nor the gap in visual perception, is noticeable to the viewer.

What remains open is how much of vision actually goes quiet while the eye is in flight.

The paper trail starts a long way back. The overview records that the phenomenon was first described by Erdmann and Dodge in 1898, when it was noticed during unrelated experiments that an observer could never see the motion of their own eyes. The mirror comes in as the easy way to duplicate the observation, by looking from one eye to the other.

The movement involved is a saccade, a fast eye motion. Because the motion is optimised for speed, the overview notes there is inevitable blurring of the image on the retina as the retina sweeps the visual field. Blurred retinal images are not much use. On that account the eye has a mechanism that cuts off the processing of retinal images when they become blurred.

One timing detail has been taken as a clue about where the cutting off is organised. Saccadic suppression starts before the actual onset of the saccade. Because of that, the overview argues, it cannot be triggered by retinal motion and must be centrally activated by the brain. It points to a supporting finding: a significant reduction of the cortical signals encoding stimuli presented immediately before a saccade, found as early as primary visual cortex.

So the missing movement is not simply a case of something happening too fast to register. Something is turned down in advance.

What exactly gets turned down was tested directly. In 1994, D. Burr, M. Morrone and J. Ross published evidence in Nature that saccadic suppression is real but selective. In their results it was selective for patterns modulated in luminance at low spatial frequencies. Patterns of higher spatial frequency, and equiluminant patterns modulated only in colour at all spatial frequencies, were not suppressed during saccades but actually enhanced.

That pattern of results, they argued, suggests the suppression is confined to what they call the colour-blind magnocellular stream, which provides the dominant input to motion centres and areas involved with attention. There, in their words, it could "dull the otherwise disturbing sense of fast low-spatial-frequency image motion".

Put in everyday terms, the signals most affected were coarse light-and-dark patterns, which is the kind of signal the authors link to fast image motion. Fine detail and pure colour did not drop out of their data. On this evidence the effect is a dimming of part of vision, not a shutter closing over all of it, and calling it a blackout claims more than the measurements do.

There is also a straightforward way to defeat the mirror, and it is worth stating because it marks the edge of the claim. Hold a smartphone close and watch your own face on the front camera, and it does show your eyes jump. The demonstration belongs to the mirror specifically, not to every way of looking at your own eyes.

The overview makes the point that saccadic masking is not fully related to the saccade itself. The same overview also states that the masking starts with the onset of the saccadic motion and the associated blur, and that it finishes as soon as the image on the retina has stabilised, whether or not the saccade has actually finished.

In the laboratory, the way this is shown is specific. A striped pattern is moved too fast to be seen, so that when the eyes are still it looks like a homogeneous surface. When the participant makes an eye movement in the same direction as the pattern, the velocity of the eye briefly matches the velocity of the pattern, the pattern is stabilised on the retina for a moment, and it becomes visible. The overview calls this intrasaccadic perception.

It also happens outside the laboratory. The example the overview gives is a train window. Look straight out at the adjacent track while moving fast enough and the track is just a blur, because its angular speed on the retina is too fast for the eye to track. Then flick your gaze left and right along the track. If the peak velocity of the saccade is at least as large as the angular velocity of the track, there is a moment when the eye and the track are moving at the same speed, the retinal image goes still, the blur stops, and the suppression switches off. What you end up seeing afterwards, according to that account, is a freeze-frame of the gravel and the dirt between the tracks.

Within-saccade detection has also been demonstrated. The overview cites a paper by Schweitzer and Rolfs, published in Science Advances in 2021, as proving and detailing within-saccade movement detection.

Older framings have been called into question. The overview records that there were two major types of saccadic masking claimed, flash suppression and saccadic suppression of image displacement, and that testing since then has revealed that these two theories may not be correct.

Saccadic masking is often used to help explain a timing illusion called chronostasis. It is the experience, reported by some observers, of glancing at a clock and feeling that the second hand has paused.

K. Yarrow, P. Haggard, R. Heal and colleagues measured the effect and published it in Nature in 2001. When voluntary saccades are made to a silently ticking clock, they report, observers sometimes think the second hand takes longer than normal to move to its next position. Their explanation is that the brain extends the percept of the saccadic target backwards in time to just before the onset of the saccade, and that this happens every time the eyes move but "is only perceived when an external time reference alerts us to the phenomenon".

They suggest the temporal extension is one of the mechanisms that fill in the perceptual gap during saccadic suppression. One of their manipulations sharpens the picture. If the target is moved unpredictably during the saccade, breaking perception of its spatial continuity, the illusion disappears.

Some limits are worth carrying into the mirror. The demonstration establishes that the movement is not perceived. It does not by itself establish why, and accounts of the mechanism have been questioned. The 1994 Nature result is evidence about how patterns of different spatial frequency and colour are seen around the time of a saccade, and the authors' own localisation is hedged. They write that masking studies suggest the suppression precedes the site of contrast masking and may therefore occur early in visual processing, possibly as early as the lateral geniculate nucleus. The chronostasis work reports on perceived duration, and the filling-in account is offered there as a suggestion rather than a settled finding.

The observation itself has outlasted the framings attached to it: two major types of saccadic masking once claimed may not be correct, according to the overview. It was noticed during unrelated experiments in 1898, it can still be repeated with nothing more than a mirror, and what has been called into question since is whether those claimed types of masking hold up.

Study: D. Burr, M. Morrone and J. Ross, "Selective suppression of the magnocellular visual pathway during saccadic eye movements," Nature 371, 511-513 (1994), https://www.nature.com/articles/371511a0. Related work cited: K. Yarrow, P. Haggard, R. Heal and colleagues, "Illusory perceptions of space and time preserve cross-saccadic perceptual continuity," Nature 414, 302-305 (2001), https://www.nature.com/articles/35104551.

Sources:

• "Saccadic masking," overview article, https://en.wikipedia.org/wiki/Saccadic_masking
• Burr, D., Morrone, M. & Ross, J., "Selective suppression of the magnocellular visual pathway during saccadic eye movements," Nature 371, 511-513, 1994, https://www.nature.com/articles/371511a0
• Yarrow, K., Haggard, P., Heal, R. et al., "Illusory perceptions of space and time preserve cross-saccadic perceptual continuity," Nature 414, 302-305, 2001, https://www.nature.com/articles/35104551