What the n-back task measures
Wayne Kirchner introduced the n-back task in 1958 to study memory for rapidly changing information. At 1-back you compare each square with the one just before. At 3-back you hold three positions at once, check the new one against the oldest, drop it, and take in the next, every 2.5 seconds without a pause.
That constant updating is what sets it apart from the digit span test, which asks how much you can hold, and the visual memory test, which asks how big a pattern you can recall. N-back asks how well you keep a moving list current.
Does n-back training raise your IQ?
This is why most people have heard of n-back. In 2008 Susanne Jaeggi and colleagues published a study in PNAS reporting that training on dual n-back raised fluid intelligence, with bigger gains the more people trained. It was a striking claim, and it launched a wave of brain-training apps.
The follow-up research was far less encouraging. A 2013 randomised trial by Redick and colleagues, with a placebo control group, found no improvement in intelligence at all. A meta-analysis by Melby-Lervåg and Hulme the same year found no convincing evidence that working memory training transfers to reasoning. Meta-analyses that did find a small benefit leaned on studies whose control groups did nothing, which lets expectation and practice at testing pass for improvement.
The pattern is a familiar one: the better controlled the study, the smaller the effect. What training reliably does is make you better at n-back itself. For what does and does not move a score, see can you improve your IQ.
What your level means, and does not
Your score is the highest level you cleared, with three errors or fewer across 20 squares. Missed matches and presses on non-matches both count, so tapping Match on everything gets you nowhere.
There is no normative sample for this format. Pace, grid size, block length and the rule for clearing a level all differ between n-back versions, which changes how far people get. The labels on your result describe the level rather than rank you against other people.
Frequently asked questions
What is a good n-back score?
On a single position task like this one, 2-back is the level most people can hold and 3-back is where most start to strain. Clearing 4-back is uncommon and 5-back rarer still. There is no normative sample for this format, because pace, grid size and the rule for clearing a level all vary between versions, so treat those as a rough guide rather than a ranking.
Does n-back training increase IQ?
Probably not in any meaningful way. A widely publicised 2008 study reported that dual n-back training raised fluid intelligence, but later, better controlled studies mostly failed to find the effect. Meta-analyses that found a small benefit relied heavily on studies without an active control group. What practice reliably does is make you better at n-back.
What is the difference between n-back and dual n-back?
Dual n-back runs two streams at once, usually a square on a grid and a spoken letter, and you track matches in both. This is single n-back on position only. Dual needs spoken letters timed precisely against the squares, which browsers cannot guarantee, so it is not offered here rather than offered badly.
What does the n-back task measure?
Working memory updating. You have to hold the last few positions, compare the new one against the right one, then drop the oldest and take in the next, every 2.5 seconds. That is a different demand from a digit span test, which asks how much you can hold, not how well you keep a moving list current.
Why do false alarms count as errors?
Because without them the task could be beaten by pressing Match on every square, which would catch every real match. Counting presses on non-matches as errors is what makes the score reflect memory rather than button mashing.
Is n-back related to IQ?
Working memory and fluid reasoning are correlated, and n-back performance tracks both to a degree. The link is real but moderate, so a high n-back level says your working memory updating is strong without telling you your IQ. Reasoning is most of what a full-scale score is made of.