What this test times

Reaction time is the gap between a stimulus appearing and your response to it. Here the stimulus is a panel turning green (simple mode) or a dot appearing on one side (choice mode), and your response is a click or a key press. The headline figure is the median of your completed trials in milliseconds — the middle value once your presses are sorted, so one distracted 900 ms press cannot drag the summary the way it would drag an average. Simple reaction time asks only that you detect something and respond; choice reaction time inserts a decision — which side lit up — and the selection of a matching hand, which is why choice times sit reliably above simple times.

How a trial runs

1. Each trial waits a random 1.2–3.2 seconds before the go signal, so counting your way to the answer does not work.
2. The panel turns green, or a dot appears in one half. The clock starts as the stimulus is drawn.
3. Click anywhere or press Space in simple mode; click the lit half or press F / J in choice mode.
4. Your press stops the clock. A wrong-side press in choice mode is logged as an error and the trial is re-run instead of recorded.

Clicking before the go signal is a false start: it is counted and the trial re-armed, so a lucky guess never enters your data as an impossibly fast time. Results report the mean, fastest trial, standard deviation, valid trials and false starts. The verdict bands are coarse — under 250 ms fast, under 400 ms ordinary, beyond that slower than typical. For adults, simple visual reaction time generally falls in the 200–300 ms region, and a two-way choice usually adds something like another 100 ms.

A worked run

Say ten simple-mode presses come out at 214, 231, 208, 240, 219, 262, 226, 205, 233 and 218 ms. Sorted: 205, 208, 214, 218, 219, 226, 231, 233, 240, 262. With an even number of trials the median averages the two middle values, so (219 + 226) ÷ 2 = 222.5 ms, displayed as 223 ms. The mean of the same ten is 225.6 ms. Had the 262 ms trial instead been a 600 ms lapse, the mean would rise to 259 ms while the sorted list becomes 205, 208, 214, 218, 219, 226, 231, 233, 240, 600 — and the median would not move at all. That insensitivity is why the headline is a median. Your fastest trial, 205 ms, estimates the floor your hardware allowed that day, not your typical speed.

Limits of a browser test

• Latency that is not yours. Every measurement swallows the time to paint the stimulus, your device's polling interval and the browser's event handling — a 60 Hz screen refreshes every ~16.7 ms. Compare runs on one device, never across devices.
• Few trials. Chance alone can move your median by 20 ms or more between runs.
• Practice and state. Later trials are often faster once you know the rhythm, and fatigue or caffeine moves the number more than any stable trait.

This demonstrates a method; it is not a medical or neurological assessment, and a slow number on a laptop says nothing about your health.

Frequently asked

Why the median rather than the mean?

Reaction time distributions have a long right tail. Occasional slow trials usually reflect a lapse in attention rather than your speed, so the middle value represents a typical press more honestly.

Why is my choice time so much slower?

The decision and response selection must finish before the movement begins. Classic work on response selection found adding alternatives lengthens reaction time roughly with the logarithm of how many there are, so two options already cost real time.

Does a false start ruin the run?

No — the trial is discarded and re-run, and false starts are counted separately so you can see whether you were jumping the gun.

Can I compare my score with someone else's?

Only loosely. Their screen, mouse and browser add their own delay to every press, and the gap between two people is often smaller than the gap between two machines.