Ghost Ball

Your brain keeps rolling the ball

The moment the ball slides behind the cover, your eyes lose it — but something in your head keeps it moving. You track an object that is not on screen, at a speed you are only remembering, and press at the moment your internal copy of the ball crosses a real line. It is the closest thing to watching your own simulation run.

You use this constantly without noticing: stepping off a curb while a car disappears behind a truck, reaching for a dropped phone, catching keys tossed slightly behind you. The seeing is optional. The model is not.

The lab version has rails and lights

Sports scientists measure this with a device called a Bassin anticipation timer: a long rail of small lights that flick on in sequence, imitating something travelling toward a target, while the subject tries to press at the exact instant the run of light “arrives”. The skill it measures has an unlovely official name — coincidence anticipation timing — and it is the quiet ingredient in every sport where things approach you: batting, goalkeeping, return games, anything with a net.

Ghost Ball is that task with the rail swapped for a rolling ball and the trick made explicit: the cover. Take the cover away and this would be easy — you would track the ball to the line and press, which is really just Lights Out wearing a costume. Hiding the ball forces the switch this test exists to catch: from tracking to predicting. The moment vision cuts out, you are running the ball on memory alone — its speed, its distance, its schedule — and pressing on your model’s word, not your eyes’.

The number to watch is the sign

Everyone misses by some amount. The interesting part is the direction. Call it early over and over and your inner ball runs fast; late over and over and it runs slow. That lean is remarkably stable for each person — and memory research has a related quirk called representational momentum (Freyd & Finke, 1984): our memory of a moving thing tends to drift a little further along its path than it really was. Moving objects are, in a small way, remembered into the future.

Within about 40 ms of the line is a sharp call. Under 100 ms either way is solid. A consistent 150 ms lean in one direction is not a worse score so much as a fact about your simulation — and unlike raw speed, it is something you can actually watch yourself correct.

Getting better here is actually possible

Raw reaction speed barely trains. This does. Anticipation is a calibration problem, and calibration responds to feedback — which is why every miss on this page tells you its sign instead of just its size. The routine is almost embarrassingly simple: if your misses keep coming up early, your inner ball runs hot, so hold the press a breath longer than feels right. If they come up late, release on the first whisper of “now”. Most people can watch their lean shrink within a single session.

What tends not to shrink is the direction itself. Squeeze the early bias out of your calls and, weeks later, it is still an early bias you are squeezing — the lean seems to be a property of how you simulate, not a mistake you unlearn once. Which makes this the one test in the battery where the honest goal is not a bigger number. It is knowing exactly which way you are wrong, and by how little.

Further reading: representational momentum. This is the only test here where reacting fast is worthless — for the opposite skill, start with Lights Out.