Who runs this
One independent developer builds and runs this site. If something is broken, wrong or unfair, write to ysc9550@gmail.com and we will fix it.
Every figure here was checked against the paper it came from before it went up. If we could not verify a source, the number is not on the site. In our view, a missing number is better than a made-up one.
How the percentiles work
Your score is compared with results from published research, not with other visitors to this site. People who find a reaction time test by searching for one are not a fair sample of everyone.
The rule is simple. If a published norm fits the task, you get a percentile, and that test's page names the study. If none fits, you get your score and your own history instead. Inventing a distribution would be easy and would look more impressive. It would also be a lie.
Reaction Time — reference study
Our reference is Woods, Wyma, Yund, Herron & Reed (2015), Factors influencing the latency of simple reaction time, Frontiers in Human Neuroscience 9:131. It tested 1,469 adults aged 18 to 65. Their mean simple reaction time was 231 ms, and the spread between people (the standard deviation) was 26.8 ms.
That study used lab equipment. A browser adds its own delay on top of your reaction. Tests of online experiment platforms put that delay at roughly 71 to 87 ms (Anwyl-Irvine, Dalmaijer, Hodges & Evershed, 2021, Behavior Research Methods 53(4):1407–1425).
So a fair comparison would take about 80 ms off your time. On the study side, it would use the authors' own device-corrected mean of 213 ms instead of the raw 231. That puts both numbers on the same device-free scale. Comparing a corrected score with an uncorrected mean would make everyone look faster than they are.
The catch is that your own device may add more or less than 80 ms. A correction that is right on average can still be wrong for you. That is why reaction time gets no percentile here. The study and the 80 ms figure stay on this page as reference and never turn into a "top N%" on screen.
Sequence Memory — percentile shown
Your span is compared with the forward Corsi block-tapping span in Facchin et al. (2024), Regression-based normative data for Corsi Span and Supraspan learning and recall among Italian adults, Neurological Sciences 45(12):5707–5718. That study tested 340 healthy adults aged 21 to 89. Their mean span was 4.97, with a standard deviation of 1.03 and a range of 2 to 8.
In the classic task, an examiner taps nine blocks placed at odd spots on a board. A 3×3 grid clicked on a screen is a close relative of that task, not the same test.
Aim Test, Number Memory, Visual Memory, Chimp Test, Verbal Memory, Color Match, Focus Test, Stroop Test, 10 Second Challenge, Typing Speed — compared with your own history
No published norm matches these tests as they are played in a browser. So your score is compared with your own history on the same device, and with the leaderboards.
- Aim Test: your time depends a lot on what you point with, such as a mouse or a trackpad.
- Number Memory: in the clinical version, an examiner reads the digits aloud. Here you see them on screen and type them back. The well-matched studies we could verify also covered only narrow age ranges.
- Visual Memory: the closest norms come from the Corsi block-tapping task. There, an examiner taps real blocks one at a time. This test shows every square at once.
- Chimp Test: the research studied trained chimpanzees who saw the numbers for a fixed time. Here, people play at their own pace in a browser.
- Verbal Memory: no published norm uses this stream of SEEN and NEW words. The Korean pages also use their own word list.
- Color Match: the result mostly reflects your screen, not your eyes.
- Focus Test: the grade lines lean on a tablet study of the same 5×5 Schulte table by Khramova and colleagues (2021, Sensors), and on a Russian Ministry of Health guideline that treats 40 to 50 seconds as normal. The tablet study tested only ten adults, so we will redraw the lines from this site's own scores once there are enough.
- Stroop Test: the grade lines are provisional. They come from per-word times in Zurrón and colleagues (2009, a two-button lab task) and Crump and colleagues (2013, typed answers online). Neither used four on-screen buttons, so we will redraw the lines from this site's own scores.
- 10 Second Challenge: the upper lines use human reaction time and counting rhythm. The lower ones use timing research such as Bartholomew and colleagues (2015, PLoS One), where 647 adults timed 1 to 15 seconds. No study uses a 3-second preview and three tries, so these lines are provisional too.
- Typing Speed: the task changes with keyboard layout and language, and no published norm matches typing sentences the way you do here.
Each test's own page explains the reasoning in full.
What we do not claim
None of these tests diagnose anything. They do not screen for color blindness, attention disorders, dementia or brain injury, and a low score on a web page is not medical information. If you are worried about your eyesight or your memory, please see a doctor, not a browser tab.
Accessibility
- 11 of the 13 tests never use color as the only signal. Text, icons or numbers mark each state as well. Color Match and the Stroop Test are the exceptions, because color is what they measure, and each page says so. We also picked the Stroop colors to stay as far apart as possible for the most common kinds of color blindness.
- Animation follows your operating system's reduced-motion setting.
- Anything you can click is at least 44 pixels, with one exception. On narrow phones, the largest memory grids cannot give every tile 44 pixels and still fit, so tiles there shrink to 30 pixels. That still clears the 24-pixel minimum in WCAG 2.2 success criterion 2.5.8. We measured the other option too: forcing 44 pixels made tiles overlap and steal each other's clicks.
- The aim test's optional hard modes shrink targets to 32 and 24 pixels on purpose, because hitting a small target is the challenge you chose. The default mode keeps the 48-pixel target.
How this is built and checked
Plain code, on purpose
The site uses no framework and no UI libraries. It is plain JavaScript modules and plain CSS, uploaded as static files, so pages load in well under a second on a phone. Speed was not the main reason, though. A test that measures the gap between two moments should have nothing extra between your click and the code that records it. Every library in that path adds a delay nobody can see.
Outside scripts
Three outside scripts load here: Google Analytics, Microsoft Clarity, and the Google AdSense loader, which sits idle while our application is reviewed. All three load in the background, so none of them sits in the measurement path. Clarity is not loaded on the test pages or the Full Brain Test at all. On the Korean pages, the KakaoTalk share buttons (on a result, in a group room and in the footer) also load Kakao's script, but only when you press one.
Analytics and Clarity show us which pages people use and where they get stuck. When you finish a test, Analytics also gets an event with the test name and your score, but nothing that says who you are. Your bests and history stay in your browser. The privacy page explains what Clarity records and how to switch these tools off.
Checks that stop a bad release
We do not rely on good intentions to keep things correct. Automated checks stop the build instead. The build refuses to publish when:
- a page's meta description runs too long
- an internal link points at a page that does not exist
- a distribution in the norms file has no source link
- an English phrase is hard-coded into a screen that must also work in Korean
- a test is missing a start button you can reach with the keyboard
We added each of those checks after that exact mistake reached the live site.
Before anything goes live, a script plays the 12 main tests from start to finish in both languages. It clicks through real rounds and reads the result panel, because unit tests can pass even when a result screen never shows up. That script is how we learned that browsers slow down timers in background tabs enough to break the timing. Now a hidden tab gets a notice instead of a bad number.
Your data
Your personal records (bests, streaks, achievements) are kept in your own browser's local storage. They disappear if you clear your browser data.
Only two things go to our server. When a test finishes, an anonymous statistics record helps build this site's own percentile baseline. It holds the game, mode, score, device type, language, date and whether it was the daily, plus accuracy on the aim test and voided clicks on reaction time. Nothing in it can identify you. And if you choose to join a leaderboard or a group room, the nickname and score you submit are stored there, along with whether you played by mouse or touch. The privacy page has the full detail.
Changelog
- 14 October 2026: Group rooms now have a big-screen view. Put it on a projector or TV and it shows a QR code, the room code and the live room ranking in large type. You can also flip the ranking to show last place first. Seven hidden achievements are in too. Finding them is up to you.
- 12 October 2026: Added the 10 Second Challenge, bringing the site to thirteen tests. The clock hides after 3 seconds, and you try to stop it at exactly 10.00, three times. The header now folds every test into one Tests menu.
- 11 October 2026: Added the Stroop Test (twelve tests). Added achievements, kept in your browser and listed on the rankings page. Added group rooms: share one link and everyone plays the same board for 24 hours, with a room ranking. Also new: a dark and light theme switch, and a home page that compares your records with your last visit.
- 8 October 2026: Split aim records and leaderboards into mouse and touch. Every game page gained a statistics table measured on this site. The Full Brain Test now plays in one screen and ends with a five-area chart.
- 3 October 2026: Added the Focus Test, a 5×5 Schulte table where you find 1 to 25 in order against the clock. The Full Brain Test now covers five brain areas (processing speed, focus, working memory, spatial memory, verbal memory) with one test each, worth 200 points apiece. The summary shows a profile by area.
- 28 September 2026: The rankings page now brings the daily challenge and every game's leaderboard together. The Full Brain Test gained a 1,000-point score with its own board, and each test page shows a short play demo.
- 4 August 2026: Added F1 Reaction as a themed version of the reaction test. It follows the real Formula 1 start procedure: five red lights, a random hold, then go when the lights go out. Its records are kept apart from the standard reaction test, and it stays out of the Full Brain Test and the daily rotation.
- 4 August 2026: Added challenge links. You can send any finished run to a friend as a duel on the identical board, with the same targets in the same order. Each player gets one attempt, used up the moment it starts. Also added: win and lose verdicts, a rematch loop, and link previews that show the challenger's score.
- 4 August 2026: Leaderboards now build up per game instead of clearing daily, and season changes will be announced in advance. Difficulty modes and typing languages keep separate boards. Signed tokens, minimum play times and rate limits protect submissions.
- 4 August 2026: Removed the reaction-time percentile from the result screen. The Woods et al. study and the 80 ms browser offset stay documented as reference, because delay differs from device to device and no single correction can be honest for everyone.
- 4 August 2026: Added Verbal Memory as the ninth test. Words appear one at a time, and you call each one SEEN or NEW.
- 3 August 2026: Added Typing Speed as the eighth test, with a play screen styled after Korean typing practice programs, key sounds and a pace graph. Every game got its own daily leaderboard, and results gained tier badges.
- 3 August 2026: Added the daily challenge, difficulty modes (chimp and aim), one-line result verdicts, social sharing and the chimp logo.
- 2 August 2026: Added the Chimp Test after a close look at the category leader, bringing the site to seven tests. Also added three written guides.
- 2 August 2026: Added Visual Memory, bringing the site to six tests. Grid tile sizing moved to one place after we found tiles overlapping their neighbors on narrow screens.
- 2 August 2026: First release, with five tests, published-norm percentiles for reaction time and sequence memory, local scores and streaks, and share cards.