You are told a colour, then the squares appear all at once. Find the one you were asked for and tap it. The round starts with two squares and ends with sixteen — and the growth is the point.
Two numbers, not one:
The split comes from a well-replicated finding known as the Hick–Hyman law: choice time grows with the logarithm of the number of options, not with the options themselves. Going from 2 squares to 4 costs about the same as going from 8 to 16.
The familiar version — wait for green, then click — measures conduction and little else. It is identical on trial one and trial fifty, there is nothing to find and nothing to decide, and a lucky early click can beat a careful one.
Here the target is named first, then eight or sixteen coloured squares appear at once. You have to search, recognise and choose before you move. The classic version is inside this one: the two-square round is close to simple reaction, and everything above it adds the decision.
Four rounds of growing difficulty — 2, 4, 8 and 16 squares — with several trials each. For every round we take the median rather than the mean, because one distraction should not decide your result. Then a straight line is fitted across the four rounds: where it starts is your base speed, and its slope is your decision cost.
Trials you answer before the squares appear are cancelled, not counted against you. Leaving the tab mid-trial cancels that trial too, because a browser stops drawing in a hidden tab and the timing would be meaningless.
Everything here is measured in the browser, so your screen and input device are part of the result. A 120 Hz phone screen reports faster times than a 60 Hz laptop; a mouse differs from a trackpad; a busy machine adds milliseconds of its own.
That is exactly why this site shows you no population percentile. The honest comparison is you against your own earlier attempts on the same device, and that comparison is the one we save for you.
Mostly the search: you learn where to look and your base speed drops a little. Decision cost moves far less, which is the interesting part — it behaves like a property of the task rather than a trick you can learn. And whether any of this transfers to reactions outside a screen is not something we claim; the evidence for that kind of transfer is weak.
It is free and runs directly in your mobile or desktop browser with nothing to install and no sign-up. Signing in is optional, and all it does is save your results so you can follow them from any device.