Stillness

Try this. Find a point and fix your gaze on it — a spot on the wall, a period at the end of a sentence. Hold it there. Don't move your eyes. After a few seconds, anything sitting in your peripheral vision will begin to fade. Not dramatically. A slow bleed, like a color left in sunlight. The image dissolves, and then the brain fills the gap with whatever's around it.

Ignaz Paul Vital Troxler noticed this in 1804. A Swiss physician, apparently observant enough to document something that had been in front of human eyes for the entirety of human existence without being formally described. Which suggests either that it's subtle, or that we don't pay much attention to what we're not looking at.

The reason it happens is adaptation. Neurons that receive an unchanging signal begin to quiet down. No change, no report. This is a general property of the sensory system — of most sensory systems. Mechanoreceptors in skin stop firing during sustained pressure; you stop feeling your clothes within minutes. Olfactory receptor neurons adapt within half a second to a sustained odor; you walk out of your house smelling of whatever you ate for breakfast, oblivious. All of these systems are change detectors pretending to be state detectors. They report the edge, not the surface. The stable world you perceive is continuously reconstructed from a signal that only tells you what just changed.

In the 1950s, researchers built devices to test how much change the visual system actually requires. They mounted tiny optical systems on contact lenses that moved with the eye, keeping an image perfectly stabilized on the same photoreceptors regardless of any eye movement — a perfectly still image on the retina. The result: it disappeared entirely within one to three seconds. Alfred Yarbus, in 1967, wrote the conclusion plainly: in any test field unchanging relative to the retina, all visible differences disappear, and do not reappear.

So seeing requires that the image not hold still on the retina. Something must keep moving it.

Your eyes aren't still when you look at something. Even during careful fixation, they drift continuously, and several times a second they produce a small involuntary jerk — a microsaccade, less than a degree of arc, invisible to an outside observer. In 2006, Susana Martinez-Conde and Stephen Macknik tracked subjects' eyes while they reported Troxler fading. Before fading events, the rate and magnitude of microsaccades dropped. Before recovery — before the image came back — microsaccades increased. The inference was that these tiny movements keep refreshing the retinal signal, interrupting adaptation before it can take hold.

Eight years later, a closer analysis distinguished between preventing fading and reversing fading once it's started. Slow drift between microsaccades — the wandering, not the jerk — turns out to account for more than half of prevention events. Drift holds things at bay. Microsaccades are more important for recovery: rescuing the image after it's already gone. The system has two mechanisms doing different parts of the same job, and neither one alone is sufficient.

But here's the part that keeps sitting with me: the visual system suppresses its own movements. During every microsaccade, there's a brief neural suppression — the brain edits out the motion it just caused, so you don't experience the world lurching. This is why you can't catch your own eyes moving in a mirror: the motion happens and is removed before it becomes perception. The same movements the system depends on for visual continuity are simultaneously hidden from the system's output.

So the mechanism runs below the level of the experience it produces. You see a stable world because the system is continuously and invisibly disturbing itself. Stop the disturbance and the world fades. But you can't perceive the disturbance — that's been edited out. What you get is the product, not the process.

And when the process fails — when adaptation wins anyway — the brain doesn't produce blankness. It fills in. During Troxler fading, surrounding texture and color are actively projected into the gap by higher cortical areas. In subjects with stabilized retinal images, the image didn't stay gone; it kept cycling, fragments reassembling briefly before dissolving again. Something in the system, lacking a real signal, kept generating an answer.

I'm not sure what to call that except: prior. When input fails, the system falls back on expectation. Which might mean that the stable world we perceive is always partly that — expectation, corrected by signal when signal is available, running on its own when it isn't. The microsaccades carry information. The filling-in generates content from somewhere else. Under normal conditions, these two contributions are blended so thoroughly that you can't find the seam.

What Troxler found in 1804 is still there to find. Look steadily at a point and something in your peripheral field will eventually vanish, and then the brain will paint over the gap with what it thinks should be there. The disturbance that prevents it is happening right now, edited out before you can observe it.

I don't know what it would mean to actually see clearly what you see.