A letter appears, then another, then another. Press Match only when the current letter is the same as the one from two steps ago (2-back) or three steps ago (3-back). This is a working-memory updating exercise for education and entertainment, not a clinical exam.
Level
2-back
Progress
Ready
Accuracy
—
—
Choose 2-back or 3-back, then start. Press Match when the letter matches the one from n steps ago.Choose 2-back or 3-back, then start. Press Match when the letter matches the one from n steps ago.
The n-back test is a classic working-memory updating paradigm. Instead of holding a static list, you continuously replace old items with new ones and compare the current stimulus with the item that appeared n steps earlier. Cognitive psychology uses versions of this task to study attention, updating, and response control under time pressure.
CognitiveBench uses a single stream of letters. That is a visual identity n-back, not dual n-back (which often adds a second stream such as locations). Dual n-back training products are a different product category and are not claimed here. Read the explainer What Is the N-Back Test? if you want the background before playing.
In 2-back, the comparison is close: letter 5 matches letter 3, letter 6 matches letter 4, and so on. In 3-back, the gap is wider, so you must keep a longer moving window. People often slow down or raise false alarms when they jump to 3-back too quickly. Start with 2-back until the rule feels automatic, then try 3-back on the same device.
Production timing is 500 ms stimulus on, then a blank until 2500 ms from onset. A Match press during that window counts as a response. Hit reaction time is the integer millisecond difference between letter onset and the press, using performance.now(). After the window—or immediately after a press—you see a short correctness label, then the next letter. Twenty scored trials follow n warm-up letters. Hits require a match on a true target. Skipping a non-target is a correct rejection. Pressing Match on a non-target is a false alarm. Missing a target is a miss.
| Pattern | Practical reading | What to try next |
|---|---|---|
| High accuracy, few false alarms | The rule is under control | Try 3-back or a second 2-back for consistency |
| Many misses | Targets were late or overlooked | Watch the fading letter; respond earlier in the window |
| Many false alarms | The comparison is noisy | Stay on 2-back and skip unless you are sure |
| Fast hits, low accuracy | Speed is beating the rule | Wait for identity, then press |
These bands are descriptive product guidance, not published norms or diagnostic cutoffs. Browser delays, display refresh, and input method all sit inside the reaction time. Compare yourself to yourself on the same setup.
Working memory is the ability to hold and update a small amount of information while doing something with it. N-back is one laboratory style of that work. Doing well on letters in this layout does not prove a general memory upgrade, and a weak round does not mean a disorder. Sleep, stress, interruptions, and caffeine can move the score more than people expect.
Some commercial programs advertise n-back as brain training that raises IQ. The research record is mixed: people improve on the trained task, while transfer to unrelated tests is much less reliable. This page does not sell a training protocol and does not claim treatment effects for ADHD or any other condition.
Sequence Memory asks you to reproduce a growing spatial pattern. Number Memory asks for a digit string after a brief view. Both load short-term storage, but they do not require the same continuous updating rule. The Stroop Test measures inhibition of an automatic reading response rather than n-item comparison. Use several tools if you want a broader personal snapshot, and do not average them into a fake IQ.
Results are for education and entertainment, not medical advice. See the Disclaimer and Methodology pages for limits of browser measurement.
It is a working-memory updating task. You must keep a short stream of letters in mind, compare the current item with the item from n steps ago, and decide whether to respond. The score reflects this specific task, not overall intelligence.
In 2-back you compare the current letter with the letter two steps earlier. In 3-back you compare with the letter three steps earlier. 3-back usually feels harder because more items must be maintained and updated.
The first n letters are warm-up and are not scored. Each later trial is a hit, miss, false alarm, or correct skip. Accuracy is hits plus correct skips, divided by scored trials. Mean hit reaction time uses only successful match presses.
A useful score is one you can repeat on the same device when you are rested. Accuracy near chance with many false alarms means the timing or level is too hard for this session. Do not treat a single percent as a clinical cutoff or IQ estimate.
No. This browser version is for education and entertainment. It cannot diagnose ADHD, dementia, or any other condition, and it is not a substitute for a qualified professional assessment.
Getting better at this layout is common with practice. Claims that dual n-back training produces broad intelligence gains are mixed and often overstated. Treat improvement here as task familiarity unless a controlled study says otherwise.
Yes. Use the Match button with touch input. Keyboard shortcuts (Space, Enter, or M) are optional on desktop. Compare phone rounds with other phone rounds because touch latency differs from a keyboard.
Background tabs can freeze timers and distort reaction time. If the page is hidden during a run, CognitiveBench stops the round so a delayed callback cannot be counted as a real response.