Memory
Letters appear one at a time. Press Match when the current letter is the same as the one shown N letters earlier. Simple to explain, famously hard to do.
Results stay in this browser
Choose a mode
Example
2-Back
Letter 3 matches letter 1, two steps back. You press Match on letter 3 and nowhere else in this sequence.
Every trial lasts exactly 2.5 seconds, so sessions at the same level are comparable.
The N-back task is one of the most widely used working memory measures in neuroscience.
Press Space on a keyboard or tap the large Match button.
Results stay in your browser only.
Your result uses d prime, which rewards catching matches and penalizes false presses fairly.
Exactly 30 percent targets, with few near-miss lures, so every session is fair.
Your sensitivity: how well you separate matches from non-matches. The main score.
How it is calculatedz(hit rate) minus z(false alarm rate), with log-linear correction
The share of scored letters you handled correctly, pressing or not pressing.
How it is calculatedHits plus correct rejections divided by scored trials
Hits are matches you caught. False alarms are presses on non-matches.
How it is calculatedCounts of each response type
How quickly you pressed on correct matches.
How it is calculatedAverage time from letter onset to press, hits only
These bands are site-defined guidance for adults on a desktop computer or a recent phone. They are not population norms and not a diagnosis.
| Band | Range | What it means |
|---|---|---|
| Excellent | 3.0 or more | Matches and non-matches clearly separated. |
| Good | 2.0-2.9 | Reliable performance at this level. |
| Average | 1.0-1.9 | Some confusion, common at 2-Back and 3-Back. |
| Needs practice | below 1.0 | Try a lower N level first. |
The psychologist Wayne Kirchner introduced the N-back task in 1958 to study short-term memory. It became popular decades later in brain imaging research.
Today it is a standard way to load working memory in a controlled, adjustable way.
N-back measures working memory updating. You must hold the last N letters, compare each new one, and drop the oldest, again and again.
That constant update is what makes 2-Back and 3-Back so demanding.
Pressing on every letter catches all matches but creates many false alarms. Never pressing avoids false alarms but misses every match.
d prime balances both, so it reflects real discrimination rather than a strategy of pressing often or rarely.
People improve at the N-back with practice. Whether that improvement spreads to other skills, such as reasoning, is debated, and many large studies found little transfer.
Treat the N-back as a measure and a mental workout, not a guaranteed way to boost intelligence.
Every scored letter falls into one of four groups. A hit is a match you pressed for. A miss is a match you let pass. A false alarm is a press on a non-match, and a correct rejection is a non-match you correctly ignored.
Looking at these four numbers together tells you whether your errors come mainly from missing matches or from pressing too often.
1-Back is mainly a test of attention and quick comparison. 2-Back adds real working memory load, and 3-Back pushes most people to their limit.
A good rule is to stay at a level until your d prime is reliably above 2, then move up. If d prime drops below 1, step back down for a few sessions.
After the first N letters, exactly 30 percent of the scored letters are matches. Non-matching letters avoid accidental matches, and near-miss lures are kept rare, so each session at the same level is equally fair.
Results are compared only within the same N level, so a 3-Back session never counts against your 2-Back best. The history page lets you filter by mode and see your trend over time.
Expect your d prime to rise over the first sessions at a new level, then level off. Short, regular sessions in a quiet setting usually give the clearest picture.
If a session feels unusually hard, check how you slept and how many distractions were around before reading too much into the number.
N-back tasks are used in brain imaging, aging and attention studies to apply a controlled memory load.
A clear demonstration of working memory limits and how they rise with load.
A demanding focus exercise with adjustable difficulty.
Researchers use standardized versions. This online test is not a diagnostic tool.
Silently say the last N letters as a short chant, updating it with each new letter.
Learn the rhythm at 1-Back, then move up when your d prime is above 2.
False alarms hurt d prime as much as misses. Do not guess.
Each trial is exactly 2.5 seconds. Settle into the beat rather than rushing.
It is a working memory task. You watch a stream of letters and press when a letter matches the one N steps earlier.
Each letter shows for 0.5 seconds, followed by a 2 second blank. Press Space or tap Match during that time if it is a match. Doing nothing counts as no match.
Practice takes about a minute, Standard about two minutes and Challenge about three.
Yes. Press P, use the pause button, or switch tabs. The interrupted letter is shown again when you resume.
Any modern browser, with a keyboard or touch screen.
d prime is your main score: higher means you separate matches from non-matches better. Accuracy and hit RT add detail.
On this site a d prime of 2 or more is Good and 3 or more is Excellent. Compare only within the same N level.
Use History in the menu. Results are grouped by N level.
Only in your browser.
Reload the page. If Space scrolls the page, click inside the test first.
A match needs a letter N steps earlier to compare with. The first N letters have nothing to compare against, so they are never matches and are not scored.
d prime. Accuracy can look high simply by never pressing, because most letters are not matches. d prime rewards real discrimination between matches and non-matches.
If you have never tried the task, start with 1-Back in Practice mode to learn the rhythm and controls. Move to 2-Back once you feel comfortable and your d prime is above 2.
Results depend on age, device, sleep, language, mood and practice. Average several sessions before drawing conclusions. This is not a medical or psychological diagnostic tool.
It takes about a minute. Your results stay private in this browser.