What dot count measures
Each round shows between twelve and ninety dots for 650 milliseconds, then hides them. That is long enough to see the cloud clearly and far too short to count it. You type a number, and you are scored on how far off you were as a percentage of the true count.
This is a measure of the approximate number sense, the capacity to judge quantity without counting. It is one of the better studied perceptual abilities: infants have it, many animals have it, and in adults its precision is usually expressed as a Weber fraction, the smallest proportional difference you can reliably tell apart. Typical adult values sit between 0.15 and 0.25, which is why most people land within ten to twenty percent here.
The sense has a characteristic failure. Errors scale with the count, so being three out on a cloud of fifteen and twelve out on a cloud of sixty is the same performance. It also leans toward underestimation for large sparse clouds and overestimation for small dense ones, which is why the dots here are spaced so that density cannot do the work for you.
What counts as a good estimate
| Estimate on 40 dots | Error | Accuracy |
|---|---|---|
| 40 | 0% | 100% |
| 38 or 42 | 5% | 95% |
| 36 or 44 | 10% | 90% |
| 32 or 48 | 20% | 80% |
| 28 or 52 | 30% | 70% |
| 20 or 60 | 50% | 50% |
| Average over five clouds | Percentile |
|---|---|
| Below 70% | Bottom tenth |
| 70 – 80% | Below average |
| 80 – 88% | About average |
| 88 – 93% | Above average |
| 93 – 96% | Strong |
| Above 96% | Top few percent |
How to estimate more accurately
- Do not try to count. At 650 milliseconds you can count perhaps four items before the cloud vanishes, and a half-finished count is worse than a clean impression. Let the whole cloud land at once.
- Anchor to a reference you know. Most people have a stable sense of what twenty and fifty look like. Ask whether the cloud is more or less than each, then place it between.
- Judge by area covered, then correct for spacing. A cloud that fills the box is not necessarily large; the dots are spaced to prevent that shortcut. Note how much of the box is covered and how tightly, and combine the two.
- Watch for your own bias. After a few rounds you will see whether you run high or low. Most people underestimate clouds above forty. Once you know your direction, nudge every guess against it.
- Guess quickly. Your first impression is the number sense talking. Deliberating for several seconds tends to replace it with a worse guess reasoned from fragments.
Frequently asked questions
What does this test actually measure?
The approximate number sense: the ability to estimate how many things are present without counting them. It is a distinct capacity from arithmetic and is present in infants and many animals.
Why 650 milliseconds?
Long enough to see every dot clearly, short enough that counting is impossible past four or five. The point is to measure estimation, not counting speed.
What is a good score?
An average above 88% over five clouds is above average, and above 93% is strong. Errors scale with the count, so a guess of 55 on 60 dots is exactly as good as 14 on 15.
Why do I keep guessing low on big clouds?
That is the most common bias. Large sparse clouds are systematically underestimated by most people. If you notice it, correcting upward on anything that feels like more than forty usually helps.
Can I improve?
Somewhat. The underlying precision is fairly stable in adults, but the systematic biases are correctable, and calibrating against a few known quantities improves most people's average noticeably.
Is this an IQ test?
No. Number sense correlates modestly with maths ability in children, but this is a single perceptual measure, not an assessment of anything broader.