Flanker Test online
Five arrows in a row. Answer the middle one and pay no attention to the rest — which is easy right up to the moment the rest point the other way. A free online flanker task, the one Eriksen and Eriksen built in 1974: 20 rows, under a minute, no account, running entirely in your browser. It tells you how many milliseconds the conflicting arrows cost you, and says in plain words how large that cost is.
Take the test
Answer the direction of the middle arrow. The ← and → keys work too.
▶▶▶▶▶
0
ms of interference
The conflicting rows barely held you up. Over a run this short that is as easily a good patch as it is control — take it again before you read anything into it.
This is the ordinary outcome: the gap is what the task is built to produce, so having one means the test worked, not that you did badly.
The conflicting rows cost you real time. That is common when you are tired, hurrying, or meeting the arrows for the first time.
On most conflicting rows the neighbours got their answer in before you did. Going a little slower usually shrinks this more than trying to go faster.
Every conflicting row had to be argued with. One distracted stretch moves this a long way, so it is worth a second run before reading much into it.
Too few clean answers to compare the two kinds of row. The run needs a couple of correct, unhurried ones of each — worth another go.
ms = milliseconds, from the row appearing to your press. This is the interference cost: your average on conflicting rows minus matching ones. An answer under 150 ms was on its way before the row could be read, so it is counted as a miss.
- Matching
- —
- Conflicting
- —
- Clean
- —
The row has to be looked at — there is no version of this task that works without seeing it. Nothing is uploaded; the result stays in this tab.
That was a single run. In the app it becomes a short daily session, with a Focus Index to show for it.
Point your camera at the code — the App Store opens.
What the flanker task measures
Selective attention, in the narrowest sense the phrase has: how well one thing stays in focus while the things beside it are pushing a different answer. The middle arrow is the task; the other four are the pressure.
▶▶▶▶▶
All five point one way. The neighbours are already saying what the middle one says, so nothing is in the answer’s way.
faster◀◀▶◀◀
The four around it point the other way. You have taken them in before deciding to, and that has to be pushed aside first.
slowerSchematic, and not drawn to scale: the bars only show which way the difference goes. Its size is what the test above measures.
That delay is the flanker effect, and its size is the whole result. It is read as a measure of how narrowly attention can be aimed: everything outside the middle of the row is irrelevant to the task, and the cost of it is how much of it got in anyway.
Why you cannot simply ignore them
Because “ignore” is not an instruction the eye can follow. Attention narrows, but it does not narrow to a point: whatever sits close to what you are looking at is taken in with it, whether or not it is any use. The four arrows stand right beside the target, near enough that there is no keeping them out.
So on a conflicting row two answers are on their way at once — the one the target calls for and the one four arrows are voting for — and the wrong one has a head start of nothing at all. It arrives first, and it has to be held back before the correct press can happen. The milliseconds you measure are how long that holding-back took.
The other tests here that put an automatic answer in the way of a correct one are gathered in one place: Attention Tests lays out what the family measures, what each member does differently, and what none of them can tell you.
Where it comes from
- 1974
Eriksen and Eriksen crowd the target
Barbara and Charles Eriksen publish Effects of noise letters upon the identification of a target letter in a nonsearch task. The target always sits in the same place, so there is nothing to search for — and identifying it still takes longer when the letters flanking it call for the other response.
- 2002
It becomes one part of a larger measure
Jin Fan and colleagues publish the Attention Network Test, which puts a flanker task inside a cued procedure so that three aspects of attention can be read off a single sitting. The flanker rows are the part that measures the conflict.
- Since
Letters became arrows
The version most people meet now uses arrows rather than letters, which costs the task nothing and hands it to anyone regardless of the alphabet they read. It is a standard measure of selective attention in research on attention and on cognitive control.
How to read your result
One headline number and two supporting ones. They are only worth anything read together.
-
The gap
Your mean on conflicting rows minus your mean on matching ones. Subtracting removes how fast you happen to be in general, leaving only what the conflict cost.
-
The two means
Shown separately so you can see which side moved. Both high means you were taking your time; only the conflicting one high is the effect in its pure form.
-
Clean answers
Out of 20. An answer is clean when it named the right side and took at least 150 ms — a small gap earned by pressing before the row was read is not the same result as a small gap with everything right.
Under the number the test also names the size of the effect — from minimal to very strong, on five steps. That is a description of this run and nothing else. It says how big your own gap was; it does not say where you stand among other people, because nobody here has that number to give you.
And because the gap is a difference, a quick person and a slow one can produce identical interference — which is exactly why it, rather than raw speed, is the figure worth watching. A positive gap is the normal outcome: it is what the task is designed to produce, so seeing one means the test worked. A gap near zero on a run this short is more often noise than a finding, and a negative one almost always is.
What this version cannot tell you
-
It is short
20 rows is enough to feel the effect and get a rough number, not enough for a stable measurement. Research versions run several times as many, and often several blocks of them.
-
Your equipment is in the number
Touchscreen versus keyboard, display latency, whatever else the browser is doing — all of it lands in the milliseconds. Compare your runs to each other on the same device, never to someone else’s.
-
It is one conflict, not attention
What this measures is how much four irrelevant arrows got in the way on one afternoon. Staying with a dull task for an hour, or switching cleanly between two tasks, are different abilities that this run says nothing about.
-
It is not a diagnosis
Flanker tasks appear in research on attention, and in batteries used by people trained to interpret them alongside much else. On its own, a large gap identifies no condition. A slow run here means you had a slow run here.
Questions about the flanker task
What is the flanker task?
A row of arrows appears and you answer the direction of the middle one, ignoring the four around it. On half the rows those four agree with the target and on half they point the other way, and the difference between your average time on the two kinds is the whole result. It was introduced by Barbara and Charles Eriksen in 1974 with letters rather than arrows, and is usually called the Eriksen flanker task. The run here is 20 rows long.
What does the flanker task measure?
The flanker task measures how narrowly attention can be aimed at one thing while other things nearby argue for a different answer — selective attention, and the control it takes to keep an unwanted response out of the way. It does not measure how long you can concentrate, how fast you are in general, or how well you reason. It measures the cost of the neighbours.
What is a good flanker score?
The flanker task has no single good score, and anyone quoting one for a browser test is guessing. How large the effect comes out depends on how many rows you run, how far apart they are drawn, what you answer with, and how tired you are. What is meaningful is your own gap over several runs on the same setup: a smaller gap means the conflicting arrows cost you less that time.
Is the flanker task the same as the Stroop test?
The flanker task and the Stroop test are relatives, not twins. Both put an automatic response in the way of the correct one and time the override, and both report the same subtraction — conflicting minus matching — so the numbers mean the same kind of thing. What conflicts differs: in Stroop it is reading a word against naming a colour, which needs a reader and a vocabulary; here it is four arrows beside the one you were asked about, which needs neither. The two are worth taking together for exactly that reason.
How do you do the flanker test?
Press Start, then answer each row with the direction of its middle arrow — the two buttons, or the ← and → keys. There is nothing to set up and no practice block: a short blank sits between one row and the next so that every row has a moment it starts at, the run is 20 rows long, and the result appears as soon as the last one is answered.
Why do answers that are too fast not count?
Because a press that lands under 150 ms after a flanker row appeared was already on its way. Whichever side it named, it was not a response to what was on screen, and the headline number here is a difference — the one kind of score that gets better the less carefully it is earned. So those presses are counted as misses rather than quietly dropped, and the figure under the result says how many of the 20 rows were answered cleanly.
Is the flanker task used to diagnose ADHD?
No, and a version in a browser tab least of all. Flanker tasks are used in research on attention, and versions of them appear in batteries administered under controlled conditions and read by someone qualified, always alongside other measures — and even there a large gap on its own identifies no condition. If your attention genuinely worries you, this page is not where the answer is; a clinician is.
Do I need an account?
No. The flanker test runs entirely in your browser, nothing is submitted, and your result is shown to you and then gone when you close the tab.
One run is an afternoon. The trend is the point.
A page can hand you one number. What it cannot do is tell you whether that number is moving. NEURA on iPhone runs conflict tasks of this family, adapts them to you, and rolls every result into a single Focus Index you can watch over weeks — which is the reading that was worth having all along.
