A square lights up and a letter is spoken, every three seconds. Press Position when the square is where it was n turns ago, Letter when the letter is the one from n turns ago. The level moves by the 2008 trial’s own rule — and beside it is what the trials since actually found.
Dual N-Back
Level 1-back Play a block and your score, and your next level, appear here.
The level rule is the 2008 trial’s: fewer than 3 mistakes in each stream and n goes up by one; more than 5 in either and it goes down.
Your blocks (this browser only)
What the trials found
A training drill, not a test of intelligence or a clinical test, and it does not diagnose anything. This is not medical advice: if your memory or concentration has changed in a way that worries you, that is a conversation with your doctor.
Dual n-back is a working-memory drill: every three seconds a square lights up in one of eight places and a letter is spoken, and you press when either matches the one from n turns back. Practice makes you better at it. Whether that makes you smarter is disputed: the 2008 trial said yes, and the larger meta-analyses since found a small effect, a very small one, or none.
How to play
Press Start. Each turn, a square lights for half a second and a letter is spoken (or shown, if you choose). At 1-back, press Position (or the A key) when the square is in the same place as the turn before, and Letter (or L) when the letter is the same as the turn before — both, if both match. At 2-back you compare with two turns ago, and so on. A block is 20 + n turns with six position matches and six letter matches in it. When it ends you see your matches and false presses in each stream, and the level for the next block.
The drill, exactly as the 2008 trial ran it
Jaeggi et al. (2008) used “squares at eight different locations … at a rate of 3 s (stimulus length, 500 ms; interstimulus interval, 2,500 ms)” and “one of eight consonants” through headphones. Each block had “six auditory and six visual targets per block (four appearing in only one modality, and two appearing in both modalities simultaneously)”. The level moved by one rule: “If the participant made fewer than three mistakes per modality, the level of n increased by 1. It was decreased by 1 if more than five mistakes were made”. A mistake is a missed match or a press when nothing matched. A session was 20 blocks, about 25 minutes, and the groups trained for 8, 12, 17 or 19 days. The paper does not say whether “more than five” counts per stream or in total; this drill counts per stream, as its first rule does.
| Level | Turns | Position matches | Letter matches | Time |
|---|---|---|---|---|
| 1-back | 21 | 6 | 6 | 63 s |
| 2-back | 22 | 6 | 6 | 66 s |
| 3-back | 23 | 6 | 6 | 69 s |
| 4-back | 24 | 6 | 6 | 72 s |
| 5-back | 25 | 6 | 6 | 75 s |
What the trials found
The 2008 paper: “Here, we present evidence for transfer from training on a demanding working memory task to measures of Gf” (fluid intelligence). Its comparison groups “were not trained between the two testing sessions” — they did nothing, which later studies changed. Redick et al. (2013) compared adaptive dual n-back with an active placebo — an adaptive visual search program — and found that “despite a high level of statistical power, there was no positive transfer to any of the cognitive ability tests”. Melby-Lervåg, Redick and Hulme (2016), 87 publications: “no convincing evidence of any reliable improvements when working memory training was compared with a treated control condition”. Au et al. (2015), 20 n-back studies, found “a small but significant positive effect of n-back training on improving Gf” — two of its authors ran the 2008 trial. Soveri et al. (2017), 33 randomized controlled trials: “A medium-sized transfer effect was observed to untrained n-back tasks. For other WM tasks, Gf, and cognitive control, the effect sizes were of similar size and very small.”
What every one of them agrees on: people get better at the drill they practise. That is the honest way to use this page — as a measure you can repeat, beside the memory and Stroop tests and the reaction time test, not as a promise.
Worked examples
- A 1-back block with every match caught and no false presses: 12 of 12 matches, 0 mistakes in each stream — fewer than three, so the next block is 2-back.
- A 2-back block with 4 position mistakes and 2 letter mistakes: not under three in each stream, not over five in either — the next block is 2-back again.
- A 3-back block with 6 letter mistakes: over five in one stream, so the next block is 2-back.
Common mistakes
- Pressing on every turn to be safe. A press with no match is a mistake, the same as a missed match.
- Reading a higher level as a higher IQ. Your level measures this drill. The meta-analyses above are the evidence on the rest.
- Playing in a background tab. The timing stops being three seconds, so a block stops when you leave the page.
- Comparing yourself with a stranger’s level. We found no published norm for a browser version; compare your own saved blocks over the weeks.
Frequently asked questions
What is dual n-back?
A working-memory drill with two streams at once — a square’s position and a spoken letter. You press when either matches the one n turns earlier. Jaeggi et al. used it in their 2008 training study; it is the version most apps copy.
Does dual n-back increase IQ?
The 2008 trial reported gains in fluid intelligence. Later studies with active control groups found no transfer (Redick et al. 2013), and the meta-analyses found either a small effect (Au et al. 2015), a very small one (Soveri et al. 2017) or no convincing evidence (Melby-Lervåg et al. 2016). Gains on the drill itself are not in dispute.
How long should you train dual n-back?
In the 2008 trial a session was 20 blocks, about 25 minutes a day, for 8, 12, 17 or 19 days. That is a description of the study, not a recommendation: Soveri et al. (2017), across 33 trials, found “no effects of age, training dose, training type (single vs. dual)” on transfer.
What is a good dual n-back level?
We found no published norm for a browser version, so the drill does not compare you with anyone. It saves your blocks in this browser so you can compare yourself over time.
Is the n-back a working memory test?
Researchers use n-back tasks to measure and train working memory. This page is a training drill, not a clinical or diagnostic test, and it does not measure intelligence.
Why are the letters spoken?
The 2008 trial presented the letters through headphones while the squares were on screen — one stream for the ears, one for the eyes. You can show the letters on screen instead, which is easier with the sound off but is a different drill.
Read next
- Do nootropics help with memory?
- 3 meditation techniques for focus, concentration and memory
- Best nootropic stack for studying and learning
- How can I focus better without Adderall?
This is not medical advice, and nothing here diagnoses or treats anything. It is a training drill; if your memory or concentration has changed in a way that worries you, that is a conversation with your doctor. Some links on this site are affiliate links: I buy what I review at full price, and a commission never changes a verdict.