What this test measures
Keep a phone number in your head while you hunt for a pen and you are using verbal working memory. This game runs on its spatial counterpart: a short-term store for where things happened and in what order. Psychologists call it visuospatial working memory. You use it when you watch someone tap a door code and try to repeat it. Or when you glance at a map and hold the next few turns in mind.
Two demands stack up on every level. You have to remember which squares lit up, and you have to remember when each one did. A player who recalls all the positions but scrambles the order still fails. That order component separates this task from plain location memory, and it is why each added step feels so much heavier than the last.
The science
The direct ancestor of this game is the Corsi block-tapping task. Philip Corsi invented it in 1972 for his research at McGill University. Nine wooden blocks fixed to a board. A tester taps some of them in order. The person being tested taps the same order back. The longest sequence a person could reproduce, their Corsi span, became one of the most widely used measures of visuospatial working memory. In 2000, Roy Kessels and colleagues published a standardised version, because decades of use had splintered the task into dozens of slightly different lab variants.
For the percentile you see after a run, we use recent normative data. Facchin and colleagues (2024) tested 340 healthy Italian adults aged 21 to 89. The average span was 4.97. Most people sat within about one step of that, and the full range ran from 2 to 8. Your score is placed against that distribution.
How our version differs from the lab task
Three things separate this page from the tester's table. Each one matters for how you read your result.
The layout is the first difference. The lab task uses nine blocks scattered irregularly across a board, and we use a regular 3×3 grid. A regular grid is easier to describe to yourself, since every square has an obvious name like "top-left" or "centre", and that probably helps recall.
The pacing is the second. An examiner taps blocks by hand, whereas here each square lights for half a second (a little longer if your device asks for reduced motion) and you click your answer back at whatever speed suits you. Self-paced responding removes the pressure of a person waiting across the table.
The third is the step numbers. Every lit square shows its position in the sequence alongside the color change, so the pattern stays readable if you cannot distinguish the colors. The numbers also hand you something to rehearse verbally, which wooden blocks never did.
Each change makes the screen version a bit easier to memorise than the original. And the reference group is Italian adults spread across nearly seventy years of age. So treat the percentile as a reference point against a published dataset, not as a clinical score. It tells you roughly where you sit. It does not measure you the way a neuropsychologist would.
What a typical score looks like
In the Facchin study, the average span was 4.97. In plain terms: the typical adult can reliably repeat a five-step sequence. The standard deviation was 1.03, so spans of 4 and 6 are both ordinary. The full range across all 340 participants was 2 to 8, and nobody went past eight.
So what counts as good? Level 5 puts you at the adult average. Level 6 sits a full standard deviation above it. Level 7 or 8 matches the best people in the study. Anything past 8 beats every tested participant — though on our friendlier grid, such scores come a little cheaper than on Corsi's board. Bear in mind that the sample covers ages 21 to 89, so "average" here averages across a wide stretch of adult life.
How to remember longer sequences
Somewhere around level 5 or 6, raw memorisation stops working for most people and strategy takes over.
Chunk positions into shapes. Three squares forming an L, a diagonal or a row are one object to hold instead of three separate locations, so a nine-step sequence becomes three shapes chained together. It also helps to trace a path. Imagine a line travelling from square to square as the pattern plays, because a single continuous route replays more easily than a list of isolated points.
Name the squares as they light. The grid has obvious labels and each flash carries a number, so muttering "one top-left, two centre, three bottom-right" recruits verbal memory as a backup for the spatial trace. And keep your eyes on the grid. Glancing at a notification hurts more than you would expect, because the spatial trace fades quickly and re-anchoring your gaze costs attention you needed for the next flash.
An honest note on practice. When your score climbs over repeated runs, much of the gain comes from sharper strategy, not from growth in underlying capacity. Skilled memory in every domain works through structure like this, so the improvement is real. It simply means a practised level 8 and a first-try level 8 are different achievements.
Frequently asked questions
Is this the same as the Corsi test?
No. It borrows the core idea: a sequence of places that grows by one step per level. But the neat grid, the relaxed clicking pace and the visible step numbers all differ from the official tester-run task. Results from the two are related without being interchangeable, which is why we present the percentile as a reference, not a formal Corsi span.
Why did the pattern not play?
Almost certainly the tab was hidden. Browsers slow their clocks in background tabs to save power. That would scramble the timing of the flashes, so the game waits instead of playing a garbled sequence. Bring the tab to the foreground and keep it visible, and the pattern will run.
Why do the squares light up more slowly for me?
Each square normally stays lit for 500 milliseconds. If your system reports a reduced-motion preference, the game extends this to 700 milliseconds per square so the playback stays comfortable. An optional hard mode shortens the flash to 300 milliseconds and keeps its own record. Scoring is unchanged: your result is still the last level you completed.
What exactly is my score?
The last level you finished. Reproduce a six-step sequence, then click one wrong square during the seven-step one, and your score is 6. A single wrong click ends the run, so there is no partial credit for a nearly correct sequence.