The Landmark Kyoto University Primate Research Institute Experiments
In 2007, cognitive primatologist Sana Inoue and Professor Tetsuro Matsuzawa at Kyoto University published a shocking study in Current Biology titled Working memory of higher visual task in chimpanzees. The results challenged the long-held assumption that human cognitive capability is universally superior to that of non-human primates in all domains.
The researchers presented chimpanzees (including a young male named Ayumu) and university students with numbers 1 to 9 randomly scattered on a touchscreen. As soon as the subject touched the number 1, all remaining numbers (2–9) were immediately masked with solid white squares. The subject had to remember which square concealed which number and touch them all in exact ascending order.
The 65-Millisecond Flash Trial: Human Defeat
When numbers remained visible until touched, humans and chimps performed with comparable high accuracy (~90%). However, Matsuzawa then reduced the initial viewing time to extreme flash speeds: 650ms, 430ms, 210ms, and finally a staggering 65ms (roughly the duration of a single eye blink).
At 65ms—far too brief for human saccadic eye movements or subvocal rehearsal—human university students’ accuracy collapsed below 35%. Ayumu, however, maintained an astonishing 80%+ accuracy rate, tapping out all 9 numbers correctly in under 2 seconds!
The Cognitive Trade-Off Hypothesis
How could a chimpanzee possess superior working memory to adult humans? Matsuzawa proposed the landmark Cognitive Trade-Off Hypothesis:
1. Ancestral Primate Baseline: In dense jungle canopies, survival demanded rapid, high-bandwidth spatial snapshots of the environment—tracking predators, food branches, and rival chimps in sub-second intervals without requiring slow linguistic processing.
2. Human Evolutionary Divergence: As hominids evolved on the savannah, evolutionary pressure selected for complex symbolic language, social cooperation, and syntax. To allocate cortical real estate in the prefrontal cortex and left hemisphere (Broca’s and Wernicke’s areas) for language, humans sacrificed the raw photographic high-capacity working memory buffer of our primate ancestors.
Eidetic Memory vs. Serial Subvocal Rehearsal
Humans process the Chimp Test serially: we look at "1", move our eyes to "2", find "3", and chant "1-2-3-4" in our phonological loop. Because eye saccades take ~200ms and articulatory loop cycling takes ~250ms per word, humans cannot encode 9 items in 65ms.
Chimpanzees, by contrast, appear to retain an immediate eidetic after-image in their visual working memory, treating the entire screen as a single parallel spatial map.
Can Humans Train to Beat Chimps on the Chimp Test?
On the Human Benchmark Chimp Test, most untrained humans reach Level 5 to 7 before striking out. However, with intensive deliberate practice and specialized spatial tracing techniques (forming geometric constellation shapes across numbers), top human players can achieve Level 12+ scores.
While humans cannot match the chimp’s 65ms flash speed without years of specialized eidetic training, our ability to apply flexible symbolic chunking allows us to conquer much larger grids over longer exposure times.
- Kyoto University experiments proved young chimpanzees (Ayumu) outperform adult humans on rapid 65ms spatial working memory tasks.
- The Cognitive Trade-Off Hypothesis proposes humans traded raw eidetic spatial memory capacity for complex symbolic language and syntax.
- Humans rely on slow serial eye saccades and phonological rehearsal; chimps utilize rapid parallel visual working memory snapshots.
- Untrained humans average Level 5–7 on the Chimp Test; top benchmark performers use geometric spatial grouping to surpass Level 12.