A visual memory test shows you a pattern of lit cells on a grid, hides it, and asks you to click the cells that were lit. It sounds like the simplest possible task and it is one of the most revealing, because almost nobody scores near their ceiling on the first attempt — the strategy gap between a naive attempt and a practised one is enormous. You can take it on [Visual Grid Memory](/games/visual-grid) in about ninety seconds.
Visuospatial working memory: holding a set of positions rather than a sequence. It is a genuinely separate capacity from the memory you use for a phone number, and many people are noticeably stronger at one than the other.
The crucial detail is that it measures capacity in chunks, not in cells. A naive attempt tries to hold seven individual squares, which is right at the edge of what anybody manages. A practised attempt holds two shapes, which is trivially easy. That difference is why scores climb so steeply once the strategy clicks.
How it differs from the other memory tests
Sequence memory (Color Sequence, Simon) presents items in order and asks for them in order. Order information is held separately from identity, and it is more fragile.
Number memory (Number Memory) is verbal — you can rehearse digits by saying them, which visual patterns do not allow as easily.
The chimp test (Chimp Test) shows positions and numbers together and asks for them in numeric order, which combines position memory with sequencing.
Visual memory is the purest of the four: a static pattern, no order required, nothing to say out loud.
Encode shapes, not cells
Three cells in a row is one memory, not three. A square is one memory. A diagonal is one memory. Almost every strong score on a visual memory test is someone holding two or three shapes rather than seven or eight positions, and the shift is available to anyone who tries it deliberately.
Use the grid edges as anchors
'Second from the left, top row' survives the delay far better than a purely visual impression, because it converts a picture into something you can rehearse in words. Coordinates are slower to encode than shapes, so use them for the cells that do not fit a shape.
Count during the display
Knowing there were six lit cells tells you when to stop clicking. Without a count, people either guess an extra cell — which is penalised — or stop one short of what they actually remembered.
Do not move your eyes
Take the pattern in from a fixed point rather than scanning it. Scanning costs you the frames you need for encoding, and a grid small enough to fit in one glance should be taken in one glance.
Visual memory is less sensitive to sleep than reaction time is, but it is not immune. The bigger day-to-day variable is how much deliberate encoding you do, which drops off when you are tired or bored even if your capacity has not changed.
Screen size matters more than people expect. A grid that fits comfortably in one glance on a laptop may require eye movement on a phone, which turns a single-glance task into a scanning one. Compare only runs made on the same device — the same principle that applies to every cognitive benchmark.
You will get better at visual memory tests with practice, and quickly. Most of that gain is strategy rather than capacity, which is a real improvement in what you can do even if the underlying store has not grown.
Whether it spreads to anything else is the contested question, and the honest answer is that the evidence for far transfer is weak. Practising a grid pattern will not make you better at remembering where you left your keys, and anyone telling you otherwise is selling something.
What is well supported is that staying mentally and socially active is associated with better cognitive ageing across large observational studies. That is a good reason to play something you enjoy regularly, and a bad reason to pay a subscription for it.
Visual Grid Memory — the standard test, positions on a grid.
Brick Collapse — the same idea with a larger board, which forces chunking sooner.
Bomb Finder — position memory with a penalty for over-clicking, so counting matters.
Star Map — positions in open space with no grid to anchor to, which is considerably harder.
Pattern Copy — reproduce rather than recognise, which is a finer-grained demand.
The whole set lives on the memory games and spatial games hubs.
A grid where some cells light up, disappear, and you click the ones that were lit. It measures visuospatial working memory — holding a set of positions rather than a sequence.
Track your own rather than comparing. Scores depend heavily on grid size and display time, so a number from one site does not transfer to another. Most people gain several levels in the first week just from learning to chunk.
Encode shapes rather than individual cells, use the grid edges as anchors, and count the lit cells so you know when to stop clicking. That combination is worth more than any amount of untargeted practice.
No. Photographic memory in the popular sense — recalling a page in detail at will — has never been convincingly demonstrated in an adult under controlled conditions. Visual working memory is a small, ordinary capacity that everyone has.
Yes. Every game on this site is free with no account, no email and no download.
A visual memory test measures something small and real, and the fastest way to improve it is not more practice but a different strategy. Look for shapes, count the cells, and take the pattern in without moving your eyes. Then [take the test](/games/visual-grid) and see where you land.