How does the Serial Reaction Time Test work?
Four boxes sit in a row and one lights up at a time. You press it as fast as you can, and 250 ms after a right press the next one lights up. There are 200 presses in 10 blocks of 20. All but one block follow a hidden 10-step pattern; block 9 is random. You aren't told either.
Tap the boxes, or press D, F, J, K for boxes one to four. A wrong press is marked with a cross and the box stays lit until you hit the right one. The whole run takes about two and a half minutes.
How is the Serial Reaction Time result calculated?
Your result is the learning effect: how much slower you were in the random block than in the patterned blocks either side of it, in milliseconds. A positive number means your hands had picked up part of the pattern and lost that advantage when it stopped. Zero or less means no sign of it this time.
Each press is timed from the animation frame that lights the box to the moment your finger or key goes down. A block's time is the median of its right first presses, so one glance away in 20 presses doesn't move it; presses that started wrong are left out and counted as mistakes. The chart on the result shows every block, with the random one marked, and the pattern itself is shown at the end.
What does the Serial Reaction Time Test show?
It shows learning you didn't set out to do. Mary Jo Nissen and Peter Bullemer (1987) had people press keys for lights in four places following a ten-step sequence. Their presses got faster as the sequence repeated, compared with lights in random places. When people had to count tones at the same time, the learning was much weaker: hence their title, “Attentional requirements of learning”.
Learning of this kind is called implicit or procedural: it shows in what you do, whether or not you can say what the pattern was. Before you look at your result, try writing the pattern down. Many people find their hands knew more than they could tell.
Why is the Serial Reaction Time pattern new every game?
Because Nissen and Bullemer's sequence is published, and knowing it would turn the test into a memory test. Each game draws a fresh pattern of the same make-up: 10 steps, two boxes three times each and two twice. It is also chosen not to stand out: no box twice in a row, no three steps straight along the row, no back-and-forth, and no shorter pattern repeated.
The random block uses each box exactly as often as the pattern does and never the same one twice running, and it contains no four steps in a row from the pattern. So the only thing that changes in that block is the order your hands had learned.
What can affect your Serial Reaction Time result?
Spotting the pattern on purpose usually makes the effect bigger, because you can start to move before the box lights. Slowing down as you tire works the other way, which is why the random block is compared with the blocks on both sides of it rather than with the start. Your device's delay adds the same time to every press, so it cancels out of the difference. For plain speed without a pattern, try the Choice Reaction Time Test; to remember a sequence on purpose, the Sequence Memory Test.