The visual half of working memory
Working memory is not one thing. The standard model splits it into a phonological loop that holds verbal material, roughly what a digit span test measures, and a visuospatial sketchpad that holds where things are and what they look like. This test goes after the second one.
The two are related without being interchangeable. People are routinely much stronger on one than the other, which is exactly why a serious assessment samples several domains instead of trusting a single task to stand in for the lot.
What the comparison is, and is not
Elsewhere on this site, percentiles come with normative data behind them. The digit span test is scored against published adult norms, and the percentile calculator works from the standardised IQ distribution.
This format has no such data. Grid memory games vary in tile count, grid growth, flash duration and how many mistakes they forgive, so a level here is not comparable with a level elsewhere and nobody has standardised it.
What is consistently reported is the shape of the results: most people finish between level 5 and level 10, and getting past 12 is uncommon. The comparison you get at the end is built from that shape, which is why it reads better than about 84% of people rather than the 84th percentile. The first is an estimate and says so. The second would be a claim about a normative sample that does not exist.
It is a real distinction rather than a pedantic one. A percentile implies someone measured a representative sample and placed you in it. Nobody has done that for this task, and presenting a modelled figure as though they had is the easiest way for a site like this one to lose the thing that makes its other numbers worth reading.
Frequently asked questions
What is a good score on a visual memory test?
Most people finish somewhere between level 5 and level 10 on a test of this kind. Reaching level 10 puts you ahead of roughly 84 percent of people and getting past level 12 is genuinely uncommon, at somewhere near the top 4 percent. Those figures are estimates from how people typically perform on this format rather than a standardised distribution, because no normative sample exists for grid memory tasks.
What does this actually measure?
Visuospatial working memory, which is your capacity to hold spatial information in mind briefly and act on it. It is the visual counterpart to the verbal capacity measured by a digit span test, and the two are related but not identical. People are routinely much better at one than the other.
Is visual memory the same as photographic memory?
No. Eidetic imagery, the ability to retain a detailed visual scene as though still looking at it, is rare, largely confined to children, and disputed. What this measures is ordinary visuospatial working memory, which everyone has and which holds only a handful of items at a time. Scoring well here does not indicate a photographic memory.
Why does the grid grow instead of just adding more tiles?
Because difficulty on one axis alone plateaus. Adding tiles to a fixed 3 by 3 grid quickly becomes a matter of remembering which few squares were empty, which is an easier task than remembering which were lit. Growing the grid as the pattern grows keeps the proportion filled roughly stable, so each level is genuinely harder than the last.
Does visual memory relate to IQ?
Working memory contributes to general cognitive ability and forms one of the four index scores on the Wechsler scales, so there is a real connection. It is a contribution rather than a substitute. Holding a pattern in mind is a different operation from reasoning about one, and reasoning is most of what a full-scale score reflects.