What the n-back test asks you to do
Here the items are positions on a three by three grid. One square lights up at a time, and you choose Match when the lit square is in the same place as the square N steps earlier. At 1-back you compare with the square just before it. At 2-back you compare with two before. At 3-back, three before.
The first N positions are for watching only. Marking a position that is not a match takes a point off, and the score never falls below zero. A round runs 13 to 20 positions. Squares also repeat at other distances by chance, so a vague sense of having seen one before is not enough.
What it measures, and what it does not
It measures working memory under a steady load. You hold a short list, update it every second or two, and compare while you update. You also have to drop anything older than N on time.
It does not measure your memory for the day or for what you read. One round does not even measure your n-back very well, because scores move a lot between rounds and between sittings.
Where the numbers come from
The task comes from an experiment by W. K. Kirchner, published in the Journal of Experimental Psychology in 1958, on how well people of different ages kept up with information that changed quickly.
Its reliability is a known weakness. Jaeggi and colleagues reviewed the psychometrics and ran three experiments, and concluded that the n-back task is not a useful measure of individual differences in working memory, partly because of its insufficient reliability (Memory, 2010). The same paper says the task is still useful for experiments. Read your score as a record of one round, not as a measure of you.
A meta-analysis of 87 publications on working memory training found reliable gains on working memory tasks themselves, and no convincing evidence of transfer to nonverbal ability, verbal ability, reading or arithmetic when training was compared with a treated control group (Melby-Lervåg, Redick and Hulme, Perspectives on Psychological Science, 2016). A 2017 meta-analysis of n-back training found clear gains on other n-back tasks and only very small ones on other working memory tasks, reasoning and attention control (Soveri and colleagues, Psychonomic Bulletin and Review).
How to practice it here
Start at 1-back and move up only when you can finish a round with no false marks. Say each position to yourself as it appears, in a short word or a number. Keep a window of N and let everything behind it go, rather than trying to hold the whole sequence.
If the page is hidden during playback the round restarts from the beginning, so play it in one sitting. Start with the n-back practice. The Compare exactly N steps back lesson shows how to keep count. Other memory tasks and lessons are on the Memory page.
Related practices on this site
- Digit span. Hold a run of digits and give it back in order.
- Running span. Digits stop without warning, so you never know what to hold.
- Operation span. Judge a sum, hold a letter, repeat, then type the letters back.
Questions people ask
What is an n-back test?
An n-back test shows one item at a time and asks you to mark it when it matches the item from N steps earlier. At 2-back you compare each item with the one two before it. It is a working memory task. W. K. Kirchner introduced it in a 1958 experiment on how well people keep up with information that changes quickly.
Is the n-back a good measure of working memory?
Not for one person's score. Jaeggi and colleagues concluded in 2010 that the n-back task is not a useful measure of individual differences in working memory, partly because of its insufficient reliability. They also said it is still useful for experiments. Scores move a lot between rounds and between sittings, so read yours as a record of one round.
Does n-back practice carry over to other skills?
Mostly not. A 2016 meta-analysis of 87 working memory training publications, by Melby-Lervåg, Redick and Hulme, found reliable gains on working memory tasks themselves but no convincing transfer to nonverbal ability, verbal ability, reading or arithmetic against a treated control group. A 2017 meta-analysis of n-back training found clear gains only on other n-back tasks.
How do you keep up at 2-back or 3-back?
Say each position to yourself as it appears, as a short word or a number. Keep a window of N items and let everything older go, rather than trying to hold the whole sequence. Start at 1-back and move up only when you can finish a round with no false marks. Items also repeat at other distances by chance, so a vague feeling of having seen one is not enough.
How is the n-back scored here?
The items are positions on a three by three grid. You choose Match when the lit square is in the same place as the square N steps earlier. The first N positions are for watching only. Marking a position that is not a match takes a point off, and the score never falls below zero. A round runs 13 to 20 positions.