Why You Can't See Your Own Blind Spot

The popular version has your visual system inventing content. The evidence supports something more conservative and, in one respect, stranger: it may not be generating anything at all.

Where the hole is

A million or so axons from your retinal ganglion cells have to exit the eye and head for the brain. They do it through a single opening in the retina, the optic disc, and that opening cannot contain photoreceptors because the nerve fibres are occupying the space.

In each eye the resulting blind region sits about 12 to 15 degrees temporally from where you are looking — away from your nose — and roughly 1.5 degrees below the horizontal. It spans about 5.5 degrees horizontally and 7.5 degrees vertically. That is close to three thumb-widths across, held at arm's length. It is not a pinprick.

Edme Mariotte described it in the 1660s and demonstrated it to the court of Louis XIV and to the Royal Society in London, which tells you something about how counterintuitive a finding it was: the fact that you have a substantial hole in each visual field was a courtly entertainment, because nobody had suspected it.

Find yours in thirty seconds

Draw two marks on a sheet of paper about eight centimetres apart — a cross on the left, a solid dot on the right. Close your left eye. Fix your right eye on the cross and hold it there. Move the paper slowly towards and away from your face.

At some distance the dot vanishes. Not dims, not blurs — vanishes, and the page where it was looks like plain page. Keep moving and it reappears.

That replacement is the whole puzzle. If the blind spot simply produced no signal, you might expect something like darkness, or a smudge, or a sense that information is missing. You get uninterrupted paper instead.

The explanation that is almost right

The standard account says the brain fills in the gap by guessing what belongs there. The verb is doing too much work. What the system does is far more constrained, and the constraint is the point.

Whatever happens in the blind spot is copied from the immediate surround. Put the blind spot on a red wall and you see red. On a striped shirt, the stripes continue through. Run a straight line through it and the line appears unbroken. Every one of those is the local neighbourhood extended inwards.

What never happens is the supply of actual content. Your blind spot will not show you a face that is not there, a word, or an object. If something sits entirely inside it, you lose it completely — and you lose it without any sense of having lost it. Interpolation, not invention. The system extends what surrounds the gap; it does not hypothesise about what the gap contains.

That is a meaningful difference. "Your brain makes up what you see" invites the conclusion that perception is mostly fabrication. What the blind spot actually demonstrates is a narrow smoothing operation over a region with no data — closer to how you fail to notice a substantial change in a scene than to hallucination.

The argument about whether anything is filled in at all

Here is where the field splits, and the split has not closed.

One camp holds that there is genuine neural completion: activity somewhere in visual cortex represents the blind spot region as containing the surrounding surface, as if the gap had been painted over.

Daniel Dennett argued the opposite, and the argument is harder to dismiss than it first appears. Perhaps nothing is painted in. Perhaps the visual system simply has no mechanism for reporting an absence of data, so it never flags one — what Dennett called a thrifty producer principle. On this account you do not see paper where the dot was. You fail to see anything there, and failing to see is not an experience you can have. There is no border, no edge of the data, nothing to notice.

Put starkly: does your brain fill the hole, or does it merely never tell you there is one? Those are different claims with different predictions, and both have evidence.

What the neurons say

Fiorani and colleagues recorded in 1992 from V1 neurons whose receptive fields corresponded to the blind spot of the contralateral eye. About twenty per cent of them responded to stimuli presented across it. A subset — "completion neurons" — responded well to bars longer than the blind spot's diameter but poorly to bars confined to one side of it.

That is the signature of interpolation rather than silence. A cell that fires for a bar spanning the gap, and not for either half, is responding to something the retina did not deliver.

What perception says

The most elegant demonstration is Murakami's, from 1995. Motion visible only by virtue of blind-spot completion produced a motion aftereffect — and that aftereffect transferred between the eyes, implicating the same early binocular pathway used by real motion. Filled-in motion, in other words, behaves downstream like motion.

Timing also argues for a process rather than a non-event. Ramachandran and Gregory showed that an artificial gap in a textured field fills in over roughly two to three seconds — slow enough to watch happen. De Weerd and colleagues found the time to completion increases as the hole grows from 1 to about 12.8 degrees, which is what a spreading process looks like and not what a labelling decision looks like.

The most defensible position today is that both mechanisms are real, operating in different cases. Some completion is genuine neural interpolation with measurable dynamics. Some apparent completion is just the absence of any signal that something is missing. The blind spot probably involves both, and sorting out which is which is still live work.

Why you never notice it in ordinary life

Four things conspire, and the filling-in debate is only one of them.

• Your two blind spots do not overlap. Each eye's gap sits in a region the other eye covers, so with both eyes open nothing is missing at all.
• Your eyes move constantly. Several saccades a second mean no part of the scene stays in the blind spot long.
• It is far from the centre of gaze. Whatever you are attending to is on the fovea, nowhere near it.
• There is no edge. A region with no receptors does not produce a boundary, and without a boundary there is nothing to draw attention.

Point four is the one that generalises. Your visual system has no error message for missing data. That is why the only way to find your own blind spot is a deliberate trick with one eye closed — and why it took until the seventeenth century for anyone to notice.

What it is good for

The blind spot is a working clinical instrument. Its position is known precisely, which makes it a fixed landmark in visual field testing: perimetry uses it to confirm that a patient is fixating where they are supposed to be, and an enlarged or displaced blind spot can be a sign of optic nerve disease.

For researchers it is better than that. It is a region where the retina provides nothing and perception provides something, which makes it one of the few places where you can study what the visual system contributes without having to subtract out the input.

There is also an honest general lesson, which is not the one usually drawn. Your experience of the visual field as complete is not evidence that it is. Absence of evidence does not arrive labelled — the same reason a dream can vanish without leaving any sense of something lost.

What it does not mean

The blind spot gets recruited for a much larger claim: that all perception is hallucination, or that you never see the world, only a model of it. The gap between the evidence and that conclusion is wide.

What the blind spot establishes is that a small region with no input is smoothed over using the surrounding input, without notification. That is a local interpolation over about five degrees of a field that spans close to two hundred. Extending it to the whole of vision is an argument that needs to be made separately, on its own evidence — not something the blind spot delivers.