Sequence Memory Test

Most tests here give you two attempts at a length. This one gives you none, and that is what makes it feel different. One mistake ends the run. By round eight you are not only remembering. You are managing the risk of fumbling a step you already know.

That also makes the score jumpier than a span test, which is the trade. Expect a spread of two or three steps between runs on the same evening. The number worth keeping is your best of several, which is why the result card shows your record next to the current attempt.

Every round replays the whole sequence, not just the new step. So the task rewards holding one steady picture of the lot, rather than keeping a rolling window of the last few. It is the same feature that made the electronic memory games of the late 1970s hard. It is also why nine steps feels far more than one step harder than eight.

The other consequence is that a slip anywhere kills the run. Attention tends to fail at the join between the part you have practised and the part you have just seen. That is why so many runs die on the second-to-last step rather than the last one.

This task joins two things the other memory tests keep apart. Corsi block tapping scatters its blocks so you cannot name them. Visual memory drops the order entirely. Sequence memory keeps place and order both, on a tidy grid. So most people quietly turn it into words or a rhythm: top left, centre, right, right again.

That is why scores here usually beat Corsi spans by several units. Your spatial memory did not improve between the two tabs. This layout simply lets you talk yourself through it, which Corsi is built to prevent. How working memory works goes into the two-system account behind the gap.

Grouping works and is expected. Splitting the sequence into runs of three, or humming it as a rhythm, will add steps to your score. So will pausing for a beat before you start clicking instead of rushing the first step.

Playback speed is a setting because it moves your score. Faster playback squeezes the sequence into a smaller window for repeating it back to yourself, and usually costs you a step or two. Runs at different speeds cannot be compared, and the reference distribution covers the standard speed only.

If you want the word-based equivalent, what counts as a good digit span is the place to start. It has a century of comparison data behind it.