Flanker test online: the Eriksen arrow task

A row of 5 arrows appears and you answer for the middle one only. When the outer arrows point the other way you slow down a little, and that slowdown, in milliseconds, is your interference effect.

Start the Flanker test

80 trials · about 2 min · arrow keys or tap

Your result shows once the 80th row is done, and a free account keeps it.

Middle points right: press rightConflicting row · ignore the outer 4

How it works

A short practice with feedback, then 80 scored rows with none. Each row needs one quick answer, so the whole task takes about 2 minutes.

  1. 1 · Practice

    8 rows with feedback

    Each of the 4 kinds of row appears twice, in random order. After every answer a tick, a cross or a dash for no answer stays up for 0.8 seconds.

  2. 2 · Half the rows

    All arrows agree

    All 5 arrows point the same way, left or right. 40 of the 80 scored rows look like this.

  3. 3 · The other half

    Outer arrows disagree

    The 4 outer arrows point one way and the middle one points the other. These 40 rows are shuffled in among the matching ones.

  4. 4 · Every row

    Answer within 1.5 seconds

    Press the left or right arrow key. On a phone or tablet, tap the left or right half of the screen. No answer within 1.5 seconds counts as a miss.

Before each row a small + flashes in the center for 0.1 seconds to hold your gaze there. The row then stays until you answer or 1.5 seconds pass, and a half-second blank follows. The middle arrow is drawn at the same size and line weight as the others, only darker, so its place is what tells it apart. After the practice, a 3, 2, 1 countdown starts the scored rows and a thin bar at the top fills as you go.

Barbara and Charles Eriksen introduced the task in 1974 with letters: people named a target letter while other letters, which they called noise letters, sat on either side. Arrow versions keep that idea and are now common in research. An arrow Flanker was the task given to 800 adults aged 18 to 30 across three labs in a 2020 study (Imburgio et al., 2020), and the NIH Toolbox research battery includes a Flanker test as well, reported on its own score scale rather than in milliseconds (Zelazo et al., 2014).

What it measures

Every row asks the same simple question. What changes is whether the arrows around the middle one push you toward the right answer or away from it.

Rows where the arrows agree

  • Basic response speed: how fast you see a direction and press the matching side.
  • Steady attention across 80 quick rows, with no feedback to keep you on track.
  • Accuracy when nothing pulls against the answer.

Rows where they disagree

  • Selective attention: keeping your focus on the middle arrow and filtering out the 4 around it.
  • Interference control: resisting the pull toward the side the outer arrows show.
  • The cost of that conflict, measured as extra milliseconds and sometimes as extra errors.

Subtracting your average time on agreeing rows from your average on disagreeing rows takes away most of your basic speed. What is left, the interference effect, shows how much the conflicting arrows cost you. Smaller means the outer arrows pulled on you less.

Flanker is one of the 4 tasks in the ADHD check, where it stands for focus. A slower or less accurate run does not mean a person has a condition: sleep, a noisy room or an unfamiliar device move the number too, and task scores often do not match how attention feels day to day. In two samples of 133 and 142 people, task measures of impulsivity agreed with one another but not with questionnaires on related traits, ADHD symptom ratings included (Bernoster et al., 2019).

A test, not a diagnosis.

An interference effect describes one short session on one task. On its own it cannot confirm or exclude a diagnosis.

What you get

The result screen opens right after the last scored row.

On the result screen

  • A headline chosen by your interference effect from 3 fixed bands: under 80 ms, 80 to 120 ms, and above 120 ms. The bands are the same at every age.
  • Your average time on agreeing rows and on disagreeing rows, side by side, with the difference marked as interference.
  • Short lines on overall accuracy and on accuracy for disagreeing rows, plus a note to read the result with caution if overall accuracy fell below 85%.
  • A details panel with both averages, accuracy for each kind of row and the trial counts, and a CSV of every scored row.

When you keep it

  • Create a free account and the result is stored with your other tests, ready on any device you sign in from.
  • It sits under attention in your results, beside Go / No-Go and the other attention tests you take.
  • Another test is one click away from the result screen, so the picture can widen beyond attention.

How this version is scored

For every scored row the test records the kind of row, your key or tap and the time it took. The numbers on the result screen come from that record.

Correct answer
Your key or tap matches the direction of the middle arrow. A row left unanswered for 1.5 seconds counts as incorrect.
Response time
Milliseconds from the moment the row appears to your key press or tap. Only correct answers enter the averages.
Interference effect
interference = mean time on disagreeing rows minus mean time on agreeing rows, correct answers only. It can come out below zero when the conflicting rows were not slower for you.
Accuracy
Correct answers out of 80, shown overall and separately for the 40 agreeing and the 40 disagreeing rows.
Enough answers to read
The interference effect needs at least 20 correct answers in each kind of row. With fewer, the screen says the attempt cannot be scored, shows your counts and marks the interference effect as not available.

Why errors stay out of the averages. A wrong press can be quick because the outer arrows carried you along. Averaging correct answers only keeps the interference effect about the time it takes to resist that pull, and errors are reported on their own, as accuracy.

No percentile. The result screen does not compare you with other people. The published Flanker tables this site holds also fit its results poorly: people here tend to show smaller effects than those tables, so a percentile would put most of them near the top.

The same 80 rows, reshuffled. Every attempt has exactly 20 rows of each kind: right or left, agreeing or not. Only their order changes from one attempt to the next.

Before you start

Two minutes is short, so small things around you show up in the numbers.

For a cleaner attempt

  • On a computer, rest two fingers on the left and right arrow keys before the countdown ends.
  • On a phone, hold it in both hands or lay it flat so each thumb covers one half of the screen.
  • Turn off notifications: one banner over the arrows can cost a row.
  • Keep your eyes on the center, where the + appears, rather than on the edges.

Interpret cautiously when

  • You were tired, unwell, hurried or interrupted.
  • You use switch access, a screen reader or other assistive input, which changes the timing the score depends on.
  • You took it on a phone, where touch timing is less exact than a keyboard.
  • Your accuracy was low: a fast run full of errors makes the interference effect hard to read.

Questions

What does the Flanker test measure?

It measures selective attention and interference control: how well you answer for the middle arrow while the arrows around it point the other way. The main number is the interference effect, the extra time you need on those conflicting rows compared with rows where every arrow agrees.

What is a normal flanker effect?

Conflicting rows are usually somewhat slower, and the size of the gap depends on the task version, the device and the person. In the shared data of a large online study, 519 adults averaged about 72 ms on the conflict effect of a related arrow task, with wide differences between people (Eisenberg et al., 2019). The result screen here uses fixed bands at 80 and 120 ms, not a percentile or age norms.

Can I take the Flanker test on my phone?

Yes. On a phone or tablet you tap the left or right half of the screen instead of pressing arrow keys, and a faint divider with two arrows marks the halves. Touch timing is less precise than a keyboard, so compare a phone result with other phone results.

Is the Flanker task used in the ADHD check?

Yes. It is one of 4 short tasks in the ADHD check, alongside 6 questions, and it covers focus. A task result alone cannot answer whether ADHD is present; only a qualified specialist can assess ADHD.

Sources

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