The chimp test shows numbers scattered across a screen, hides every one of them the moment you touch the first, and asks you to tap the rest in ascending order from memory. It is named after a set of experiments at Kyoto University in which young chimpanzees outperformed human adults on a task of this shape — a result that is real, frequently overstated, and genuinely interesting either way. You can play it on the [Chimp Test](/games/chimp-test).
Researchers at the Primate Research Institute at Kyoto University, working with chimpanzees who had been taught Arabic numerals, ran a task in which numbers appeared briefly on a screen and then turned into blank squares. The chimpanzees had to touch the squares in numeric order.
At very short display durations, young chimpanzees outperformed human adults — faster and more accurate at capturing the whole layout in a glance. That is the finding, and it has been replicated within the same research programme.
What it does not show
It is not a claim that chimpanzees are more intelligent than people, and the researchers did not make one. It is a result about one specific task at one specific presentation speed, and the leading interpretation is a trade-off: the human specialisation for language may have come at some cost to this kind of rapid whole-field capture.
Notably, the effect is strongest in young chimpanzees, and adult chimpanzees do not show the same advantage. Any version of the story that flattens this into 'chimps beat humans at memory' has lost most of what makes it interesting.
Simultaneous spatial memory: capturing a whole layout in one glance rather than learning it item by item. It is a different capacity from sequence memory and from digit span, and being good at one predicts the others only loosely.
The distinguishing feature is that you get no second look. The moment you touch 1, everything else hides, so the entire task is decided during the brief period before your first tap.
Do not read the numbers
This is the part that surprises people. Reading the numbers one at a time is a verbal, sequential process, and it is far too slow for the display time. Let your eyes go soft and take the whole field in at once without naming anything.
Trying harder, in the deliberate verbal sense, reliably makes people worse at this task. Relaxed glancing scores higher, which is exactly the point the Kyoto results were making.
Lock the first four
Most runs end in the first four taps, not the last. The numbers you tap early are the ones you have held longest under the least interference, so if the run is going to fail it usually fails there. Spending your encoding on 1 through 4 is worth more than spreading it evenly.
Group nearby numbers into shapes
Three consecutive numbers that happen to form a triangle is one memory, not three. The same chunking principle that works on visual memory works here, and it is what carries people past the level where raw position memory runs out.
Tap at a steady rhythm
Rushing the early taps is the single most common cause of a wasted round. The display time is over; there is nothing to gain from speed during the recall phase.
Chimp Test — the task itself.
Visual Grid Memory — position memory without the ordering demand, which isolates the harder half.
Star Map — positions in open space with no grid to anchor to.
Color Sequence — order memory without the spatial capture, the other half of the task.
Flash Memory — the same speed pressure applied to recognition rather than recall.
Young chimpanzees in the Kyoto University studies outperformed human adults at the very short display durations used there. It is a real result about one specific task, not a general claim about intelligence, and adult chimpanzees do not show the same advantage.
Simultaneous spatial memory — capturing a whole layout in a glance rather than learning it item by item. It is separate from digit span and from sequence memory.
Because deliberate, verbal effort interferes with whole-field capture. Reading the numbers one at a time is too slow for the display. Relaxed glancing genuinely scores higher.
Compare against your own previous runs. Published human figures vary enormously with display time and grid size, so quoting one would be misleading.
No. It measures one narrow capacity and has no established relationship to general intelligence.
The chimp test is the rare benchmark where trying harder makes you worse. Take the whole field in without reading it, lock the first four positions, group what you can into shapes, and tap at a steady rhythm. [Try it](/games/chimp-test) and see where your ceiling sits.