Narrower Than It Was
A shape goes past a slit seven pixels wide. Your eye never receives the shape, only one column of it at a time. You will see it whole anyway, you will be able to name it, and it will come back to you thinner than it was.
There is a panel with a narrow slit cut in it. Behind the panel, a shape slides past. At no single instant is there a shape on your screen. There is a bright column, one column wide, changing over about a second. Nothing in your eye is ever handed the figure.
You will see the figure. In one piece, with a shape you can name. This is anorthoscopic perception, described by Zöllner in 1862, rediscovered by Parks in 1965 and made famous by Irvin Rock in 1981. Whatever puts the shape together is not your retina, because your retina was never given it.
And it comes back wrong in one specific direction. The figure is compressed along the axis it travelled. Narrower than it was. That is the number this page measures, and it is the interesting part, because it means the shape is being rebuilt out of time and the time is being read at the wrong scale.
- 0️⃣ One practice crossing so the task stops being strange. Nothing is scored.
- 1️⃣ Twelve crossings in three kinds, shuffled: in the open, behind the slit with your eyes free, and behind the slit with your eyes parked on a marker.
- 2️⃣ After each one you drag a shape until it is as wide as the one you just saw. Slit trials also ask which shape it was.
- 3️⃣ The headline is the slit trials divided by your own open trials, so whatever you personally do with a mouse cancels out.
- 👁️ On the parked trials a small character above the slit changes once. Report it. Off the fovea it cannot be read, so getting it right is the evidence that your eye stayed put.
- Nothing is scored as an absolute. Every headline is a ratio against the same person's open trials, which share the shapes, the speeds, the screen, the mouse and the habit of setting things a bit small. A constant bias of any size cancels exactly.
- The adjuster starts too wide on half the trials and too narrow on the other half, at random. The page prints how often you settled below the truth having started above it, which is the receipt that the number is not just where the handle happened to begin.
- The speed changes on every crossing, so how long it took is not a usable cue to how wide it was. Counting will not help you.
- All four shape options are drawn at the same height, so height carries no information. Only the profile along the travel axis can answer that question, and that profile is exactly what the slit delivered.
- Your open trials are also the sanity check. If you cannot set a width for a shape that was in plain sight, the run is printed rather than scored.