Subitizing test online: how many dots at a glance
Black dots flash for 150 milliseconds, then a pattern covers them and you type how many there were. How your answer time grows from 1 to 4 dots shows whether small sets register at once or get counted.
Start the Subitizing testAfter the 90th trial you see your accuracy and 2 timing slopes, and can keep them with a free account.
How it works
6 practice trials open the task and are not scored. A 3, 2, 1 countdown follows, then 90 trials. You type each answer when you are ready, so the length depends on you: about 5 minutes.
- 1 · Each trial
A brief flash
A + holds your gaze in the centre for half a second. Then 1 to 9 black dots appear on a white square for 150 milliseconds, too briefly to count them one by one.
- 2 · Right after
The mask
A grey cross-hatched pattern replaces the dots for 200 milliseconds. It wipes the afterimage, so you answer from what you took in during the flash.
- 3 · Your answer
Type how many
A box asks how many dots you saw. Type a number and press Continue. Nothing hurries you, but the time until you press Continue is recorded.
- 4 · 90 trials
Every count from 1 to 9
Each count from 1 to 9 comes up 10 times, in shuffled order. Practice uses 1 to 5 dots. Dots land in new random places on every trial and never touch.
During the 90 trials nothing says whether an answer was right, and a thin bar along the top marks your progress. Because the dot positions change every time, there is no fixed shape to learn for each number.
Psychologists call the quick grasp of very small sets subitizing. Ashkenazi and colleagues describe it as a fast, accurate judgement of 1 to 4 dots, while sets of 5 or more go through a slower and less exact estimate (Ashkenazi et al., 2013). This task covers both ranges in one run, so they can be set against each other.
What it measures
The task looks at how your answer time changes as the number of dots goes up, separately for small and larger sets.
1 to 4 dots
- Quick quantity recognition: whether a small set registers as a whole, without counting.
- The subitizing slope: how many milliseconds each extra dot adds between 1 and 4.
- Accuracy on counts that are usually named almost without error.
5 to 9 dots
- Counting or estimating, which takes longer as the set grows.
- The counting slope, the same timing measure from 5 to 9 dots.
- Misses, which become more frequent once the flash is too short to count every dot.
The usual picture is an answer time that hardly moves from 1 to 4 dots and then climbs. That bend is why the task reports 2 slopes instead of a single average. A steep rise inside the small range suggests the dots were being counted rather than seen at once.
In a 2-year study of school children, Landerl found that those with dyscalculia had consistently larger slopes for dot enumeration in the subitizing range (Landerl, 2013). Work on the origins of dyscalculia often starts from foundational number capacities like this one (Butterworth, 2010). A slope from one sitting describes that sitting, not the person.
A test, not a diagnosis.
Your slopes describe one sitting of one task. By themselves they cannot confirm or rule out any diagnosis.
What you get
Your result is on screen as soon as the 90th answer is in.
On the result screen
- A headline reading of your 1 to 4 dot slope and a gauge in milliseconds per dot: under 80 reads as the expected flat pattern, 80 to under 100 as somewhat slower, 100 or more as a rise that looks like counting.
- Your overall accuracy and both slopes, for 1 to 4 and for 5 to 9 dots.
- A bar chart of your mean answer time at each count from 1 to 9.
- A data panel with accuracy, mean time and mean error per count, and a CSV file of every trial.
When you keep it
- A free account holds on to your accuracy and slopes, ready on any device you sign in from.
- Filed with your number tasks in your results, beside any others you try.
- A next test is suggested at the end, so your profile keeps growing.
How this version is scored
Each trial logs the count shown, what you typed and how long you took. Practice trials are left out of every score.
- Correct answer
- The number you typed equals the number of dots. A near miss is not counted as right; how far off you were is kept separately as the error.
- Answer time
- From the moment the answer box appears until you press Continue, so typing is included. Right and wrong answers both enter the mean time for each count.
- Subitizing slope
- A straight line fitted to your mean times for 1, 2, 3 and 4 dots. Its slope, in milliseconds per dot, is what each added dot cost you.
- Counting slope
- The same fit for 5 to 9 dots. It is shown next to the first slope but gets no reading of its own.
- Accuracy
- The share of the 90 trials you answered exactly right.
- Usable run
- At least 20 of the 40 trials with 1 to 4 dots must be right. Below that, the slopes show as N/A and the screen explains why, since timing built on guesses says nothing about recognition.
Typing is part of the time. The clock runs until you press Continue, so a slow keyboard adds delay to every count. The slope compares counts with each other, which means a steady delay matters less than an uneven one.
Fixed limits, not norms. The 80 and 100 milliseconds per dot limits are built into the task. They are not drawn from an age-matched sample, and the result screen gives no percentile.
A new set of dots each time. Counts are shuffled and dot positions are random on every attempt, so no 2 runs show the same displays.
Before you start
The flash lasts less than a fifth of a second, so small things shape the result.
For a cleaner attempt
- Sit at a normal reading distance with the screen bright enough to show black dots clearly.
- Keep your eyes on the + before each flash; a glance away costs that trial.
- Go with your first impression for small sets rather than trying to count them.
- On a phone, have the number keys showing, since typing time is part of every answer.
Interpret cautiously when
- You have low vision or rely on a screen reader or magnification, which change what a brief visual flash can measure.
- You were tired, rushed or distracted during the 90 trials.
- You typed slowly or often hit the wrong key, because typing sits inside every answer time.
- You have done this task before and already know the rhythm of flash and mask.
Questions
What does a subitizing test measure?
It measures how quickly and accurately you recognise small numbers of objects without counting. Here 1 to 9 dots flash for 150 milliseconds and you type how many you saw. The key score is how much your answer time rises from 1 to 4 dots: a flat line suggests the sets were seen at once, a steep one suggests counting.
How many objects can people subitize?
Researchers usually put the subitizing range at about 1 to 4 items; beyond that people count or estimate. The task follows that split: the first slope covers 1 to 4 dots and the second covers 5 to 9.
Can I take the subitizing test on my phone?
Yes. The dot square fits a phone screen and you answer by typing into a box. Hold the phone steady at a comfortable distance, and keep in mind that typing on a phone keyboard adds the same delay to each answer.
What is a good subitizing slope?
The result screen reads a slope under 80 milliseconds per dot as the expected flat pattern, 80 to under 100 as somewhat slower, and 100 or more as a rise that looks like counting. Those limits are fixed in the task, not norms from a published sample, and there are no adult norms for this version yet.
Sources
- Ashkenazi, Mark-Zigdon and Henik (2013): Do subitizing deficits in developmental dyscalculia involve pattern recognition weakness?
- Landerl (2013): Development of numerical processing in children with typical and dyscalculic arithmetic skills, a longitudinal study
- Butterworth (2010): Foundational numerical capacities and the origins of dyscalculia