Reaction time, memory span, aim: what's normal?
Updated 2026-08-29 · about 9 minute read
Benchmark tests for reaction, memory and aim are compulsively shareable — and the context for the numbers usually goes missing. Here is what the psychology actually says about normal, good and suspicious scores.
Reaction time: the 250 ms baseline
Median simple visual reaction time — see a change, press a button — sits around 250 milliseconds, with most adults between 200 and 280. Under 200 is genuinely quick; elite esports players and sprinters test around 150–180; and anything under about 120 ms is anticipation, not reaction — which is why athletics rules any start within 100 ms of the gun a false start, full stop.
Two structural facts worth knowing: auditory reactions run ~40 ms faster than visual (nerve path length — it is why races start with a gun, not a light), and reaction time lengthens gradually from the mid-twenties onward, a normal ageing curve rather than a defect. The reaction test averages five rounds with randomised delays, because single rounds are noise and fixed delays measure timing, not reacting.
Memory span: seven, plus or minus two
George Miller's 1956 paper gave psychology its most famous number: short-term memory holds about seven items, plus or minus two. On the digit span test that means 7–8 digits is squarely normal, 10+ is strong, and the people reciting fifty are not holding digits at all — they chunk, converting digit groups into images and stories. You do it too: a phone number is three chunks, not ten items.
The spatial versions tell the same story in different clothes. The sequence test (repeat a growing Simon-style pattern) typically tops out at levels 7–10; the visual test (recall which tiles flashed) at 8–11 — and the escape trick in both is the same: see one shape, not many positions. Some people are markedly better at one flavour than another, which is itself a finding worth having about yourself.
Aim: Fitts's law in practice
Pointing speed obeys one of the few genuinely reliable laws in human-computer interaction: Fitts's law — time to hit a target grows with distance over size. On the aim trainer that cashes out as 600–800 ms per standard target for most people, 400–500 for the sharp, and sustained sub-350 averages for FPS veterans, who are spending their advantage exactly where the law says: less correction at the end of each movement.
Your plain reaction time is the floor under every aim score — no amount of mouse practice removes the ~250 ms of noticing.
What actually moves the numbers
Mostly state, not traits. Sleep deprivation slows reactions on the scale of legal alcohol impairment; fatigue, illness and alcohol all show up immediately; caffeine buys a few milliseconds back; age moves everything slowly. This is the honest use of these tests — not ranking yourself against the internet, but noticing your own variance. A morning score and a 1 a.m. score are different people.
Technique moves scores more than raw ability: chunking for digits, path-seeing for patterns, smoothness over twitch for aim. That these tricks work so well is the real lesson — the brain benchmarks measure strategy at least as much as wetware.
The hardware tax on every score
Browser benchmarks measure you plus your equipment. A 60 Hz display adds up to 16 ms before a stimulus even appears (the refresh test shows what you are running); wireless mice and Bluetooth add more; a busy machine drops frames. The rule: compare only scores from the same setup, and treat cross-device comparisons as entertainment. Gamers chasing milliseconds should check the display before the reflexes — 144 Hz shaves ~9 ms per frame off everything, which is more than most training buys.
Does brain training work?
At the tasks themselves, yes and quickly — scores climb for days to weeks as strategy and familiarity build. Transfer to general memory, attention or intelligence is where the evidence gets thin: the large reviews consistently find that training a task improves that task and little else. The honest framing is that these are measurements and amusements, not gym equipment for the mind.
What does move the underlying machinery, boringly, is sleep, exercise and not being stressed — the same list as everything else. Meanwhile the benchmarks remain excellent for settling family arguments. The full test bench lives in the Tests category, the games that exercise the same circuits (2048, Minesweeper) are one shelf over, and Pikkit has everything else.