Dot Comparison test online: which side has more dots

2 groups of black dots flash for a fifth of a second, too quickly to count. You press the arrow key toward the side that had more. Your accuracy across 8 ratios gives a Weber fraction, a measure of how finely you sense quantity.

Start the Dot Comparison test

128 trials · about 4 min · arrow keys

When the last trial ends you see your Weber fraction reading and can keep it with a free account.

128 trials · about 4 min · arrow keys

This task needs a keyboard with arrow keys and a wide screen. Open this page on a computer to take it.

Start the Subitizing test

Subitizing works on a phone: a few dots flash and you type how many you saw.

8 or 12: the right arrow200 ms · bigger dots, fewer of them

How it works

8 practice trials come first, each answer followed by a green or red flash, then a 3, 2, 1 countdown and 128 scored trials. After each flash the screen waits for your key, so it takes about 4 minutes.

  1. 1 · Each trial

    2 clouds, 200 ms

    After a + for half a second, 2 groups of black dots appear side by side, split by a thin grey line, for 200 milliseconds. Then they are gone.

  2. 2 · Your answer

    Arrow toward more

    The screen shows the 2 arrows. Press left if the left group had more dots, right if the right one did. Go with your first impression; there is no time limit.

  3. 3 · 8 levels

    From easy to nearly equal

    The 2 groups are set at a ratio from 1:2 to 8:9. Each ratio is used 16 times, with 3 to 21 dots per group, in shuffled order.

  4. 4 · Size controls

    Dot size varies

    In half the trials both groups cover the same total area, so the group with more dots has smaller ones. In the other half every dot is the same size. No single visual cue gives the answer every time.

After practice, nothing tells you whether you were right. A bar at the top shows progress, and a short pause follows each answer before the next +. Dot positions, group sizes and the side that holds more are drawn at random on every attempt.

The quick, inexact sense of how many is often called the approximate number system. Its precision is summed up by the Weber fraction, w: the smaller the difference you can reliably detect, the lower your w. Szűcs and colleagues showed that visual features of the display can strongly affect w (Szűcs et al., 2013), which is why tasks like this one vary dot size and area.

What it measures

Each answer is a snap judgement of quantity. The spread of right and wrong answers across the ratios shows how fine that judgement is.

Across all 128 trials

  • Approximate number sense: judging quantity without counting.
  • The Weber fraction, fitted to your accuracy at all 8 ratios.
  • Overall accuracy.

Easy and hard ratios

  • Accuracy at the 4 easier ratios, 1:2 to 4:5.
  • Accuracy at the 4 harder ratios, 5:6 to 8:9, where the groups are close to equal.
  • Accuracy with equal total area and with equal dot size, kept apart.

Mazzocco, Feigenson and Halberda (2011) linked weaker acuity on this kind of task to mathematical learning disability in children. Halberda and colleagues later measured it in an internet sample of 10,548 people aged 11 to 85 (Halberda et al., 2012), and Norris and colleagues found no reliable difference between a younger and an older adult group (Norris et al., 2015).

Dot comparison is one of the 4 tasks of the dyscalculia check, which runs the same 128 trials and compares your Weber fraction with a published adult figure. It looks at one ingredient of number skills, not at maths ability as a whole.

A test, not a diagnosis.

A Weber fraction from one session is an estimate with a margin around it. On its own it cannot show whether someone has a learning difficulty.

What you get

The result appears as soon as the 128th answer is in.

On the result screen

  • A headline reading from your Weber fraction, with a gauge from sharper to less precise, a shaded zone marked as most adults and a marker for you.
  • Your accuracy at each of the 8 ratios as bars, grouped into easy and hard.
  • A data panel with your Weber fraction to 3 decimals, the typical range it is read against (w up to 0.29), and your accuracy with equal area and with equal dot size.
  • A CSV file of every trial.

When you keep it

  • Saved to your results with a free account, on whichever device you sign in from.
  • Grouped with your other number tasks, so quantity and written-number tasks sit side by side.
  • A suggested next test right after, to keep building the picture.

How this version is scored

Each trial stores both dot counts, the ratio, the size condition, your key and whether it was right. The scores use the 128 main trials.

Correct answer
You pressed the arrow toward the group with more dots. Speed is recorded but not scored.
Accuracy by ratio
The share right at each of the 8 ratios, 16 trials each.
Weber fraction
The value of w between 0.01 and 0.50 whose predicted accuracy curve best matches your 8 accuracies, found by trying values in steps of 0.005. The curve is the standard one used by the published adult figures, in which each group's perceived size varies in proportion to its size. Lower means finer discrimination.
Size controls
Accuracy with equal total area and with equal dot size, 64 trials each, shown apart so you can see whether one visual cue was carrying your answers.
Screen reading
Your w is placed against a published young-adult figure (mean 0.22, SD 0.07). Up to about 0.29 (above the 16th percentile) gets the first reading, up to about 0.36 (above the 2nd percentile) the middle one, anything higher the third: the same bands as the dyscalculia check. The gauge marker sits at a fixed point for each reading, not at your exact value.

Read against the published average. A meta-analysis of 47 adult samples aged 18 to 30 put the mean w at 0.22 (Guillaume and Van Rinsveld, 2018). The result screen reads your w against a young-adult figure with that mean, as the dyscalculia check does, so an adult close to the average reads as typical.

w depends on the dots. The same meta-analysis found that w shifts with the way dot arrays are generated. A value from this task compares best with other results from this task.

Feedback in practice only. The 8 practice trials flash a colour for right or wrong. The 128 scored trials do not.

Before you start

The dots are gone in a fifth of a second, so set up before the countdown.

For a cleaner attempt

  • Use a computer with a keyboard and a window wide enough to show both groups at full size.
  • Sit square to the screen so both halves are equally easy to see.
  • Rest your fingers on the 2 arrow keys and leave them there.
  • There is no time to count, so answer from the overall look of each side.

Interpret cautiously when

  • You have low vision or use magnification or a screen reader, which change what a brief display can measure.
  • Glare, a dim room or a very small screen made the dots hard to see.
  • Your attention drifted or you got tired over the 128 trials.
  • You took this task not long ago and knew what to expect.

Questions

What does a dot comparison test measure?

It measures the precision of your approximate number sense: how well you judge which of 2 groups has more items without counting. The result is a Weber fraction, w. A lower w means you can tell apart groups that are closer in number.

What is a normal Weber fraction for adults?

A meta-analysis of 47 adult samples aged 18 to 30 found a mean of about 0.22, with a 95% confidence interval of 0.19 to 0.26. Values depend on how the dots are generated, so treat any single figure as context. The result screen here reads your w against a young-adult figure with that mean (SD 0.07), so up to about 0.29 reads as typical.

Can I take the dot comparison test on my phone?

Not on this page: it needs arrow keys and a wide screen, so the tests hub asks for a computer. The dyscalculia check runs the same trials with on-screen buttons and includes a result taken on a phone with a caution. On a phone you can also start with the Subitizing test.

Should I count the dots?

No. The dots show for 200 milliseconds and a group can hold up to 21 of them, far too many to count in that time. The task is built to catch your first impression of quantity, so answer from the overall look of each side.

Sources

By Cristian Dominguez Rein-Loring, Creative Technology Lead · Updated October 12, 2026