If you have seen “n-back” in a psychology paper or a brain-training ad, the core idea is the same. A sequence unfolds. You do not wait until the end and recite the whole list. You update continuously: drop the oldest item in the window, add the newest, and compare. That updating demand is why researchers treat n-back as a working-memory paradigm rather than a simple span test.
Working memory, not a static list
Working memory is the short-lived workspace you use while doing something with information: holding a phone number long enough to type it, tracking the last few turns in a conversation, or keeping a rule in mind while a new letter appears. Capacity is limited. Load rises when you must both store items and transform them.
A digit-span task mainly asks “how much can you repeat?” An n-back task asks “can you keep the window current and still detect a match?” Those are related skills, not identical ones. Someone can have a decent span and still lose the 2-back rule when letters arrive every couple of seconds. The CognitiveBench N-Back Test is the updating version: letters, a fixed n, and a timed response window.
How 2-back and 3-back differ
| Level | Match rule | Typical feel |
|---|---|---|
| 1-back | Same as the previous letter | Mostly attention and pace |
| 2-back | Same as two letters ago | Default working-memory load |
| 3-back | Same as three letters ago | Heavier updating; more misses and false alarms |
Example for 2-back. Sequence: H, K, H. The third letter matches the first, so it is a target. The next letter is compared with K, and so on. For 3-back, you ignore the last two letters and look three back. The extra step sounds small. In a timed stream it is not: each new letter shoves the whole window forward, and a late decision spoils the next comparison.
CognitiveBench defaults to 2-back because it is learnable in one sitting without turning the first round into noise. 3-back is optional when 2-back accuracy is already stable. Jumping straight to 3-back often produces a score that mainly says “this was too hard today.”
How to read a score without overclaiming
A round usually reports accuracy, hits, misses, false alarms, and a mean reaction time for hits. Accuracy combines hits and correct skips. That matters: never pressing Match can look calm while missing every target. Always pressing Match inflates hits while flooding false alarms. Read the parts together.
Reaction time is only summarized for hits, so a fast average does not excuse missed targets. Browser tests also include display refresh, input delay, and whatever else the device is doing. Compare phone sessions with phone sessions. Do not treat a percent as a clinical cutoff, an ADHD marker, or a ranking of intelligence.
If you want a personal baseline, take two or three 2-back rounds on the same browser when you are reasonably rested. Use the median accuracy, not the lucky peak. Then try 3-back as a separate series. Mixing levels in one chart hides the fact that you changed the task.
Memory training claims, stated carefully
Practice on n-back almost always helps n-back. You learn the letter set, the button, and the rhythm. That is ordinary skill learning. Some studies of dual n-back (a visual stream plus another stream, often locations) reported transfer to other reasoning tests; later work and reviews have been mixed, and commercial claims often outrun the evidence.
CognitiveBench does not run a dual n-back trainer and does not sell a protocol for raising IQ, treating ADHD, or preventing cognitive decline. If a session feels useful, treat it as a short attention-and-updating exercise. Pair it with sleep, movement, and work that actually matters to you. Transfer is a research question, not a feature checkbox.
How this relates to other CognitiveBench tests
Sequence Memory grows a spatial path you must reproduce. Number Memory uses a whole digit string after a brief glance. Both store information; n-back continuously replaces it. The Stroop Test is about inhibiting an automatic reading response, which is a different control problem. For the color-word walkthrough, see Stroop Test Explained.
Try the test, then keep the disclaimer
When you want the interaction instead of the definition, open the N-Back Test. Letters, a Match control that works with touch or keyboard, and a round summary with accuracy and hit timing. Results are for education and entertainment, not medical advice. See the Disclaimer. If everyday memory or attention changes worry you, talk to a qualified professional rather than a browser score.
Frequently Asked Questions
What is the n-back test in simple terms?
You see items one after another and respond when the current item matches the item from n steps earlier. In a 2-back letter task, you press when the letter is the same as the letter two trials ago.
Is 3-back always better than 2-back?
No. 3-back is usually harder, not more valid. If accuracy collapses or false alarms pile up, the round mainly measures overload. A stable 2-back session is more informative than a chaotic 3-back.
Does a high n-back score mean strong intelligence?
No. The task samples working-memory updating under a specific timing scheme. It is not an IQ test and should not be used for school, hiring, or clinical decisions.
Can n-back training treat ADHD or raise IQ?
This site does not claim that. People often get better at the trained version. Broad transfer to intelligence tests or clinical symptoms is not something a browser game can promise.
How should I interpret misses versus false alarms?
Misses mean targets went by without a press. False alarms mean you pressed when the item was not a match. Both lower accuracy, but they point to different adjustments: later detection versus over-responding.