The flanker task explained, with an arrow version to play
The flanker task is a reaction time task where you respond to one target in the middle of a row and try to ignore the items on either side of it. Barbara and Charles Eriksen introduced it in 1974. They found that side items calling for the opposite response slow the answer down, even though people know to ignore them.
The game above is Arrow Focus, our version with arrows. Five marks sit in a row and you answer for the middle one.
What is the flanker task?
The flanker task asks for a fast choice about a single target that always appears in a known place, with distracting items next to it. The distractors are the flankers. Because the target’s place is known in advance, nothing has to be searched for, which is why the original paper calls it a nonsearch task.
Rows come in three kinds, and the names below are the ones in common use today:
| Row type | Flankers | Arrow example |
|---|---|---|
| Congruent | Call for the same response as the target | ← ← ← ← ← |
| Incongruent | Call for the opposite response | → → ← → → |
| Neutral | Call for neither response | - - ← - - |
The measure is the gap in speed and accuracy between incongruent rows and the others.
Who introduced the flanker task, and what did the 1974 paper find?
Barbara A. Eriksen and Charles W. Eriksen published it in 1974 in Perception & Psychophysics, under the title “Effects of noise letters upon the identification of a target letter in a nonsearch task”.
Their version used letters. A target letter appeared just above a fixation cross for one second. If it was an H or a K the person moved a lever one way, and if it was an S or a C they moved it the other way. The target was shown alone or with other letters, which the paper calls noise letters, on both sides. The noise letters could come from the same response set as the target, from the opposite set, or from a neutral group of letters. The spacing between letters also varied.
The abstract reports two main results. Reaction time got shorter as the spacing between letters got wider, in every noise condition. Noise letters from the opposite response set slowed people down more than noise letters from the same set.
The Eriksens concluded that a person cannot prevent the processing of noise letters that fall within about one degree of visual angle of the target. On their account the response has to be held back until the person has worked out which letter is in the target position.
Later versions swapped the letters for arrows, which need no rule to learn: the arrow points the way you answer. The task belongs to the same family as the Stroop task, where a word interferes with naming an ink color. We cover that one in the Stroop effect explained.
What does one trial look like in Arrow Focus?
A row of five marks fades in, and you swipe left or right, or tap one of the two arrow buttons, to match the middle arrow. A round lasts 30 seconds and each correct row is one point.
There is no time limit on a single row. The next row appears as soon as you answer correctly, so the pace is yours. A wrong answer shakes the row and locks the board for 800 milliseconds before the next row comes.
The game builds four kinds of row, which we read from its generator code:
- Same: all four flankers point the way the middle arrow points.
- Plain: all four flankers are dashes.
- Clash: all four flankers point the other way.
- Mixed: the two marks next to the middle arrow point the other way, and one of the two outer marks agrees with the middle. An example is
→ ← → ← ←, where the answer is right.
The mixed row has no counterpart in the three standard types. We added it so that a row with arrows pointing both ways still has to be read from the third mark.
How does Arrow Focus change as you play?
The mix of rows shifts toward clashes every five rows you answer, and the row starts to move. The game keeps a step number that equals your level plus one for every five rows answered in the current round, right or wrong. Your level starts at 0, goes up by one after two rounds in a row with at least 80 percent accuracy, and drops by one after a round under 50 percent.
The table gives the chance of each row type at each step. The numbers come straight from the formulas in the game’s code, and we confirmed them by generating 20,000 rows at each step with a short program. Every share came out within one percentage point of the formula.
| Step | Same | Plain | Clash | Mixed | Row position |
|---|---|---|---|---|---|
| 0 | 45% | 30% | 25% | 0% | Centered |
| 1 | 39% | 30% | 31% | 0% | Centered |
| 2 | 53% | 10% | 37% | 0% | Centered |
| 3 | 47% | 10% | 43% | 0% | Slides sideways |
| 4 | 41% | 10% | 49% | 0% | Slides sideways |
| 5 | 35% | 10% | 38.5% | 16.5% | Slides sideways |
| 6 | 29% | 10% | 42.7% | 18.3% | Slides sideways and up or down |
| 7 | 23% | 10% | 46.9% | 20.1% | Slides sideways and up or down |
| 8 | 17% | 10% | 51.1% | 21.9% | Slides sideways and up or down |
| 9 and up | 15% | 10% | 52.5% | 22.5% | Slides sideways and up or down |
The share of rows where at least the inner flankers disagree with the target rises by six points per step, from 25 percent at step 0 to a cap of 75 percent at step 9. A player at level 0 reaches that cap after 45 answered rows in one round.
Dashes are common only at steps 0 and 1. From step 2 the plain share drops from 30 to 10 percent, and the difference goes to rows where the flankers agree, which is why the Same column jumps at step 2 before it falls again.
When the row moves, it moves on two rows out of three. The sideways slide picks left, center or right with equal chance, and the vertical one does the same from step 6. The sideways distance is at most 64 points and the vertical distance is 0.6 of one arrow’s height.
Why does the row move?
The row moves so that the middle arrow cannot be found by staring at one fixed spot on the screen. In the 1974 experiment the target always appeared in the same place above a fixation cross. On a phone, a player could do the same thing with the center of the board and never look at the flankers at all.
Once the row slides, the middle mark is the third of five and has to be found in each row. That keeps the flankers close to where you are looking, which is the condition the Eriksens studied.
How do you play Arrow Focus well?
These tips follow from the numbers above.
- Answer from the third mark only. At step 9 the flankers agree with the target on 15 percent of rows and fully oppose it on 52.5 percent, so they are a poor guide in either direction.
- Do not guess. A guess between two directions is wrong half the time, and each wrong answer costs 800 milliseconds, so guessing costs 400 milliseconds per row on average before you count the lost point.
- Expect the row to shift from the fourth step. If your level is 0, that is after 15 answered rows. Count to the third mark instead of looking at the middle of the board.
- Watch for mixed rows from step 5. The two arrows touching the middle one always point the wrong way in a mixed row, and one outer arrow points the right way. Reading the row from its edge gives the wrong answer half the time.
- Use the dashes while they last. At steps 0 and 1 almost a third of rows have no arrows except the target, and those are the fastest points in the round.
What does a flanker score tell you?
In a study, the flanker effect is a measure of how much irrelevant items near a target slow a response under that study’s conditions. Your Arrow Focus score is the number of rows you answered correctly in 30 seconds, and it says nothing about your health or ability.
People get faster at this game with practice, as they do at most timed tasks. Whether practice on tasks like this carries over to anything else is a separate question, and the evidence for broad transfer is weak. We go through it in what the research shows about brain training games.
Other games on the site also put an irrelevant signal in the way of a simple answer. Simon Space places an arrow on the left or right of the screen and asks which way it points, and Color Clash is our Stroop game. Priority Switch puts a decoy tile beside the one that counts and moves the frame between them.
Common questions
What is the flanker task?
It is a reaction time task in which you answer about one target item while other items sit on either side of it. The side items, called flankers, can call for the same answer as the target or the opposite one, and people answer more slowly when they call for the opposite one.
Who invented the flanker task?
Barbara A. Eriksen and Charles W. Eriksen introduced it in a 1974 paper in Perception & Psychophysics, volume 16, pages 143 to 149. Their version used letters. Arrow versions came later.
What is the flanker effect?
The flanker effect is the difference in speed between rows where the flankers call for the opposite response to the target and rows where they call for the same response. In the 1974 study the opposite kind slowed people down more.