What your peripheral vision is actually for

The blur is not a lesser version of the sharp bit

Outside the small central patch where you see detail, vision does not simply get fuzzier. It becomes a different instrument with different jobs. Understanding what that instrument can and cannot do tells you which parts of a layout can act as landmarks and which parts are, for practical purposes, invisible until somebody looks straight at them.

Acuity falls off faster than most people expect. At about 10 degrees away from where you are looking — roughly a hand's width at arm's length — you resolve around a fifth of the detail you resolve at the centre. At 20 degrees it is closer to a tenth. On a laptop screen at normal distance, that means while you read this sentence, the top corners of the window are being seen at about the resolution of a road sign a hundred metres away.

Crowding is the real limit, not size

The more useful fact is this: in peripheral vision, things interfere with each other. A letter that is easily large enough to identify on its own becomes unreadable when other letters sit beside it. This is called crowding, and it follows a rough rule — interference comes from anything within about half the distance to the point you are fixating.

So at 10 degrees out, everything within about 5 degrees of a target is smearing it. A single large word in empty space is legible out there. The same word inside a paragraph is not. This is why isolation works as an emphasis channel, and why it works even better than size: isolation does not just make something bigger, it removes the flankers that were destroying it.

The practical version:

• A 16px grey label in a corner cannot be a landmark. Nobody will find it without a deliberate search.
• A 40px dark heading with 32px of clear space around it is a landmark from anywhere on the page.
• A 16px icon in a row of 16px icons is unfindable peripherally. The same icon at 24px, filled rather than outlined, with a gap either side, is findable.

Filled shapes beat outlined ones out there for the same reason. A 1.5px outline is exactly the kind of high-frequency detail the periphery throws away; a solid blob is exactly what it keeps.

What the periphery is good at

Three things, all of them coarse:

1. Luminance blobs. Large areas of light or dark, and their approximate position.
2. Motion. Peripheral motion detection is excellent, and largely involuntary.
3. Gross shape and large areas of colour. Enough to tell a circle from a bar, or a red region from a grey one.

The motion point deserves attention because it is the one designers most often get wrong on purpose. An animated banner in the corner of a page is not competing politely for attention; it is triggering a reflex evolved to notice things moving at the edge of your vision. That is why auto-playing carousels and looping adverts are so widely hated, and it is why operating systems ship a reduce-motion setting. If you make something move in the periphery, you have taken a fixation whether the reader wanted to give you one or not.

You cannot inspect your own periphery

Here is where most people get stuck. The obvious way to check whether a corner label is findable is to look at the corner — at which point it lands in the fovea and looks perfectly clear. The instrument destroys the measurement.

The usable proxy is a blur. Export the layout as an image and apply a Gaussian blur at roughly 1.5% of the image width: on a 1200px-wide screenshot, that is a radius of about 18 pixels. Free tools do this in one menu:

• GIMP: Filters > Blur > Gaussian Blur, set both radii.
• Photopea (runs in a browser, no install): Filter > Blur > Gaussian Blur.
• Krita: Filter > Blur > Gaussian Blur.

What you are looking for in the blurred image is which shapes still exist as distinct blobs and where they sit. Anything that has dissolved into the background is something a reader will only ever find by deliberate search. That may be fine — a legal footer should dissolve — but it should be your decision rather than an accident.

The blur test asks a different question from the squint test in the next lesson. The blur asks what can be found. The squint asks what comes first.

The honest limitation

This is a model of the optics and early visual processing, not of the person. A reader who has come to your page specifically to find the unsubscribe link will find a 12px grey link in the footer, because deliberate search overrides all of it — they will scan systematically until the fovea lands on it. Peripheral design governs the unguided glance, which is most first encounters and almost no second ones.

It also varies. Peripheral acuity and crowding both worsen with age, and a reader with macular degeneration has the opposite problem — the central patch is the one that fails, and the periphery is all they have. Neither is a reason to abandon the model; both are reasons to keep real text at real sizes rather than treating the whole design as a pattern of blobs.