N-Back Test
Squares light up one by one. Press when this one is where it was N steps ago.
Squares light up one cell at a time. Same cell as 2 steps ago? Press!
- Cells light up one at a time. Remember where they were.
- Same cell as 2 steps ago → press the button below. When it is not, do nothing at all.
- A wrong press costs the same as a missed one. About three in ten are matches.
First time? Try 8 practice turns first — they are not scored.
Space or A also presses it
How to read your score
| Score | What it means |
|---|---|
| 1-back · d-prime below 1.40 | An N-Back Test 1-back d-prime below 1.40 (20 scored squares) means matches and non-matches weren't clearly told apart; higher is better. At 1-back you only compare with the square just before, so this range usually comes from pressing on near-misses or pressing before the square registered, not from forgetting. It's a range on this test, not a ranking. |
| 1-back · d-prime 1.40 to 2.15 | An N-Back Test 1-back d-prime between 1.40 and 2.15 (20 scored squares) is a middle band at the easiest level. Check your false alarms: at 1-back, the gap between this band and the next is usually presses on the two near-misses, one step off the square you're comparing against. |
| 1-back · d-prime 2.15 to 2.85 | An N-Back Test 1-back d-prime between 2.15 and 2.85 (20 scored squares) means you caught most of the 6 matches and let most of the 14 non-matches go. This band's boundaries were set around the averages reported in published 1-back studies - it helps you read your score; it isn't a pass mark. |
| 1-back · d-prime 2.85 or more | An N-Back Test 1-back d-prime of 2.85 or more (20 scored squares) is close to the top score this test can give, about 3.30. If you keep landing here, 1-back isn't much of a challenge any more - try 2-back; the result card has a button for it. |
| 2-back · d-prime below 0.50 | An N-Back Test 2-back d-prime below 0.50 (20 scored squares) means matches and non-matches were hard to tell apart; higher is better. 2-back means holding two positions while a third arrives - a real step up from 1-back - so landing here on a first round is nothing to worry about. Try a 1-back round and compare. |
| 2-back · d-prime 0.50 to 1.25 | An N-Back Test 2-back d-prime between 0.50 and 1.25 (20 scored squares) means you were telling matches apart, with plenty of mix-ups left. Check your bias: the same d-prime can come from missing matches or from pressing on near-misses, which are opposite problems. |
| 2-back · d-prime 1.25 to 2.00 | An N-Back Test 2-back d-prime between 1.25 and 2.00 (20 scored squares) is around the averages reported in published 2-back studies, which this band's boundaries were set to. Twenty squares vary on their own, so one round places you in this band, not at a precise point in it. |
| 2-back · d-prime 2.00 or more | An N-Back Test 2-back d-prime of 2.00 or more (20 scored squares) means you caught nearly all 6 matches with few false alarms among the 14 non-matches. The top score is about 3.30 and needs a perfect round, so there's still room above - and 3-back is where the challenge comes back. |
| 3-back · d-prime below -0.15 | An N-Back Test 3-back d-prime below -0.15 (20 scored squares) means you pressed on non-matches at least as often as on matches; higher is better, and negative values are possible. That's not a glitch at 3-back: holding three positions in order while a fourth arrives is genuinely hard. 2-back is a good place to work from. |
| 3-back · d-prime -0.15 to 0.45 | An N-Back Test 3-back d-prime between -0.15 and 0.45 (20 scored squares) shows some ability to tell matches apart, near the bottom of what this level can show. Here the two near-misses do the most damage: with three positions in your head, a square one step off is very hard to resist. |
| 3-back · d-prime 0.45 to 1.05 | An N-Back Test 3-back d-prime between 0.45 and 1.05 (20 scored squares) is around the averages reported in published 3-back studies, which this band's boundaries were set to. It's much lower than the 2-back band of the same name - d-prime can't be compared across levels, which is why this page keeps three separate tables and records. |
| 3-back · d-prime 1.05 or more | An N-Back Test 3-back d-prime of 1.05 or more (20 scored squares) is a strong round at the hardest level; higher is better. 3-back is as far as this test goes, so compare several of your 3-back rounds instead, and treat a change smaller than one band as no change. |
These are score ranges for this site's version of the n-back game, in three tables because d-prime at 1-back, 2-back and 3-back can't be read on one scale. The boundaries are set around averages reported in published n-back studies, rounded to 0.05. They aren't percentiles or pass marks and don't say how many people you beat - this site has no reference data of its own. Twenty scored squares place you in a band, not at a point, so look at several rounds at one level and treat a change smaller than one band as no change.
How to play
- Pick your n-back level: 1, 2 or 3: The ready screen offers 1-back, 2-back and 3-back; 2-back is the default. The number is how far back you compare: at 2-back you ask whether this square is in the same cell as the one two squares ago. Your choice is remembered, and each level keeps its own records because their scores can't be compared.
- Run the n-back practice round first: Practice is eight goes that don't count, and you're told after each press whether you were right. When it ends, a screen says scoring is about to start - press Continue when you're ready.
- Let the first few squares pass without pressing: The first N squares of each round are just a lead-in: there's nothing N steps before them, so they can't match and don't count. The counter at the top always shows where you are.
- Press only when the position repeats: Each square shows for half a second, then the grid is empty for two seconds; you can answer during either, and only your first press counts. Press when this square is in the same cell as N steps back, and do nothing when it isn't - there's no second button, and not pressing is your no.
- A worked 2-back example, step by step: Number the cells 1 to 9, left to right and top to bottom, and say the squares arrive at 5, 1, 5, 3, 1. The first two are the lead-in and can't match. Step 3 is cell 5, and two steps back was cell 5 - same, so press. Step 4 is cell 3, and two steps back was cell 1 - different, so do nothing. Step 5 is cell 1, and two steps back was cell 5 - different again, so do nothing. Three answers, and only one of them was a press.
- Read your d-prime, not your hit count: Your big number is d-prime: how well you told matches from non-matches. Pressing on everything or on nothing both score near zero. Below it: hits out of 6, false alarms out of 14, your median hit time (the middle of your correct-press times) and whether you leaned toward pressing or holding back.
Full rules
This is a non-standardised adaptation — one square every 2.5 seconds, where most published n-back work uses 3 seconds — so the result cannot be exchanged with a clinical scale. A round is N lead-in squares plus 20 scored ones. The 3x3 grid stays on screen; one cell fills for 500 milliseconds, the grid is empty for 2000, and a new square arrives every 2.5 seconds. Press the button when the filled cell is the one that was filled N squares ago.
The 20 scored squares are always 6 matches, 2 near-misses and 12 ordinary ones, every round and every level. The 2 near-misses land where the square was N-1 or N+1 steps back - close enough to feel like a match if you're tracking loosely. Every round has exactly 2, so your rounds stay comparable.
In every round, no cell appears three times in a row, matches never come three in a row, and at 1-back two matches are never next to each other. So a streak you notice is real. If a round can't be built to these rules, you're told before the first square and it isn't scored.
Your score is d-prime, worked out from your 6 matches and 14 non-matches. It takes your false alarms off your hits, so it rewards telling them apart, not pressing a lot. If you press on every square or on none, there's nothing to score: the round finishes, tells you so, and isn't recorded.
The bias figure, C, shows which way you leaned. Positive means you held back and missed some matches; negative means you pressed on some non-matches. Bias isn't a mistake - two people with the same d-prime can get there from opposite directions.
A press within 200 milliseconds of a square appearing is a false start - too fast to have compared anything. It still uses up that go and never counts as a hit; on a match it's recorded as a miss. It can only lower your score.
Four habits cost points in early rounds: counting how many squares have passed instead of rehearsing the last few positions; pressing because a cell feels familiar (that's exactly what the two near-misses catch); pressing twice to be sure (only the first press counts); and pressing during the lead-in squares, which can't match.
There's no feedback during the 20 scored squares - only practice tells you if you were right. The counter at the top is the only thing that changes. Your score appears after the final pause, even if you answered early on the last square.
Stay on this page until the round ends. Switching tabs, refreshing, stopping, or the page stalling badly enough to throw off the timing means the round doesn't count - no score, no record, no resume.
Each level keeps its own best and history, and the chart shows only your current level. 2-back and 3-back d-prime can't be compared - on one line, a good 3-back round would look like a bad 2-back one. Only a few numbers per round are saved, and they stay in your browser.
Compare several rounds at the same level, not just two. Twenty scored squares is short: catching one more of the 6 matches moves your d-prime by about 0.4 on its own, more if you were already near 6 out of 6 - so two rounds less than one band apart haven't told you anything. The middle of your last several rounds is a steadier guide than your latest or your best.
FAQ
What is the n-back test, and what does it measure?
The n-back test is a working memory game: items arrive one at a time and you say whether the current one matches the one N items earlier. Here the items are positions in a 3x3 grid - one cell fills every 2.5 seconds, and you press when it's the cell filled N squares ago. It shows how well you keep a short list of positions in mind while updating it as new ones arrive. A round is 20 scored squares at the level you chose.
What is a good n-back score, and what does d-prime mean?
d-prime shows how well you tell matches from non-matches, so it's only high when you catch matches and let non-matches go; higher is better. On this test, 1.25 to 2.00 at 2-back is around the averages reported in published 2-back studies, and the top score is about 3.30. It can't tell you where you stand among other people, because this site has no reference data.
How is 2-back different from 1-back and 3-back?
The number is how many items back you compare, so 2-back means holding two positions while a third arrives, and 3-back means holding three. At 1-back you only remember the square you just saw, so the step up to 2-back is a big one. Scores can't be compared across levels, which is why each level has its own record and band table.
What is the difference between the n-back test and a digit span test?
A digit span test shows you a list, waits for it to end, and scores the longest list you can repeat; an n-back test never ends the list, and scores how well you judge each item against the one N steps back. Span is mostly about how much you can hold; n-back adds constant updating, since every new square goes in as the oldest drops out. Published comparisons find the two agree less than you might expect, so a long digit span doesn't guarantee a high n-back score.
How do you get better at the n-back task?
Rehearse the last few positions in order and drop the oldest as each new square arrives, instead of counting squares - naming the cells to yourself often works better than picturing the grid. Then check your bias: positive means you're holding back and missing matches; negative means you're pressing on squares that only looked close. Practice reliably makes you better at this game.
Does n-back training transfer to IQ or general intelligence?
Whether n-back training carries over to general intelligence is disputed, and the evidence for it is weak. Early dual n-back studies reported gains on reasoning tests; later meta-analyses found the effect shrank or vanished once active control groups and publication bias were accounted for (Melby-Lervag and Hulme; Simons et al., 2016), and the question isn't settled. What isn't disputed is that people get better at n-back itself. Treat this page as practice and self-knowledge, not a treatment.
What percentile is my n-back score in?
This n-back test shows no percentile, because that needs reference data and this site doesn't have any. Showing a rank anyway would mean inventing the people you're compared with, and a made-up number that looks official is worse than none. You get your d-prime, its band, and your own earlier rounds at the same level - that's what can honestly be shown.
What is dual n-back, and how is it different from single n-back?
Dual n-back runs two streams at once, usually a position and a spoken letter, each judged separately; single n-back, which this page runs, uses positions only. Dual versions are harder and their scores noisier, and adding sound would make the timing less clean, so this version stays visual.
Why was my n-back round not scored?
An n-back round here gets no score in three cases: you pressed on every scored square or on none, so there's nothing to judge; you left the page, refreshed or stopped; or the page stalled badly enough to throw off the timing. The screen says which one happened and offers you another round.
Related games
MemorySequence Memory TestSquares light up one at a time on a 4×4 grid. Tap them back in the same order and see how long a sequence you can reproduce.Play1-2 min
MemoryVisual Memory TestMemorize the lit cells on a 5×5 grid, tap the same spots once they disappear, and see how many positions you can hold.Play1-2 min
MemoryChimp TestTest number-position memory online: memorise the numerals on a 6×6 grid, tap 1 to hide them, then recall every position in ascending order across eight scored rounds.Play1-2 min
MemoryCard MatchTurn two cards over at a time and clear the board. Your score is the number of flips it took, so fewer is better.PlayAbout 1-3 minutesThis test is for self-knowledge and practice only. It reports one run of one task and is not a clinical or intelligence assessment.
Last updated: 2026-11-05 · Scores stay on this device