What this measures
Two groups of dots, briefly. Which side has more? No time to count — and that is the point.
This tests the approximate number system, a capacity for judging quantity without counting that humans share with infants and with many other animals. It operates in a fraction of a second and it is entirely separate from arithmetic.
Why some comparisons are easy and others are not
Accuracy here follows a clean law: what matters is the ratio between the two quantities, not the difference. Telling 10 from 20 is trivial. Telling 20 from 30 is harder — same difference, worse ratio. Telling 100 from 110 is nearly impossible at a glance.
This is Weber's law, and it applies to brightness, loudness and weight as well as number. Your personal threshold — the smallest ratio you can reliably call — is what this game narrows in on as it gets harder.
The traps built into the display
Dots are jittered in size and spacing on purpose, because the obvious shortcut is to judge total area rather than count. If every dot were the same size, more dots would always mean more ink and you could answer without engaging number at all.
- Do not judge by which side looks "fuller". That is area, and the display is built to make area misleading.
- Do not fixate on one side. Take both in with a single centred glance — the comparison is faster than sequential inspection.
- Go with the first impression. Deliberation does not improve accuracy here and costs you items. Wrong answers subtract, but hesitation subtracts more.
- Expect to be wrong on close ones. Near the threshold you are guessing, and that is the correct behaviour.
What it does and does not predict
Sharper approximate number sense correlates with mathematics achievement in children, and that finding has generated a lot of interest and a lot of overstatement. The correlation is modest, the direction of causation is disputed, and adult performance tells you very little about mathematical ability.
What it does show reliably is a genuine perceptual limit that is roughly stable within a person and varies between people — which is why the task keeps appearing in assessment batteries as a measure of rapid perceptual judgement.
Frequently asked
Q. Can I count if I am fast?
A. Up to about four items, yes — that is subitising, a separate and very accurate
mechanism. Beyond that the display is too brief and the counts too large.
Q. Does it improve with practice?
A. Slightly and quickly, then it plateaus. The underlying threshold is fairly fixed.
Q. Why does it feel like guessing at the end?
A. Because it is. Near your ratio threshold, guessing is the correct strategy — the task
has found your limit, which is what it was looking for.
For raw speed without a perceptual judgement, try reaction time.