Working on Working Memory

Several years ago, I remember being amazed by a video of a chimpanzee named Ayumu from Kyoto University’s Primate Research Institute.  Ayumu appeared to be an incredibly gifted individual with the ability to recall Arabic numeral sequences with remarkable speed and accuracy on a computer touch screen.  When Kyoto University primatologists released their study based off of Ayumu and other chimpanzees titled “Working memory of numerals in chimpanzees” (Inoue & Matsuzawa, 2007), they made headline news and stimulated intense debate in academic and popular science circles.  This debate was not centred around the specific memory capacities of the chimpanzees, but instead around the fact that the chimpanzees out performed all of the humans in the study.

In the study juvenile chimpanzees could recall the position and order of a random set of five Arabic numerals between one and nine with approximately 79% accuracy, after being exposed to the numerals for 210 ms.  None of the nine humans in the study reached this level of accuracy.

The feat was thoroughly impressive.  Like most people who watched the video, I thought to myself: “do chimpanzees have a better working memory than humans?”  Tetsuro Matsuzawa, one of the lead authors of the study, claimed that they did.  In fact, he further added that his study revealed that chimpanzees possessed a mental capacity that humans beings had lost over evolutionary time.

Some evolutionary theorists proposed that there had been a trade-off in our development between language and memory.  Others proposed that enhanced chimpanzee memory was likely an adaptation to their environment unnecessary for our hominid ancestors.  However, it was clear that more hypotheses needed to be tested before we could know for sure.

As a result, Inoue & Matsuzawa’s study was analyzed.  Primatologists at other research labs wanted to replicate their methodology to confirm the finding that chimpanzees possessed a stronger working memory than adult humans.  There would be massive implications for evolutionary theory and our understanding of human evolutionary history.

Unfortunately, it did not take long for controversy to surface.  Siberberg & Kearns (2009) discovered that the Inoue & Matsuzawa study was critically flawed.  Ayumu and the other chimpanzees had had several sessions of practice on their task and none of the human participants had any practice sessions.  This difference in the level of practice before the testing appears to explain the difference in skill.  Importantly, Siberberg & Kearns (2009) presented data that showed humans could perform at the same level as Ayumu when given even moderate levels of practice.

However, the damage had effectively been done.  The Inoue & Matsuzawa study had become so popular in both the academic and popular press that most people accepted the idea that chimpanzee working memory was superior to modern human working memory.  This affected future research studies, like the paper “Did working memory spark creative culture?” (Balter, 2010).  Within this paper it is proposed that a trade-off between language and working memory in our evolution sparked the cultural revolution between 200,000 and 40,000 years ago.  Of course, this theoretical proposition had no grounding in empirical fact since it reported “chimpanzees are better than humans at some memory tasks” based on the Inoue & Matsuzawa (2007) study (Cook & Wilson, 2010).

And the myth also continued to persist in the popular press.  In 2012, several different online news sources ran specials on the Kyoto chimpanzees, and a BBC documentary titled “super smart animals” included video of Ayumu and emphasized her “superior memory abilities”.

Back within academia, the studies methodological flaws were becoming well known thanks to two papers: “In practice, chimp memory study flawed” by Peter Cook & Margaret Wilson (2010), and “This chimp will kick your ass at memory games – but how the hell does he do it?” by Nicholas Humphrey (2012).

In the latter article, published in Trends in Cognitive Science, Humphrey reveals some critical backstory explaining the skills of the Kyoto chimpanzees.  Humphrey revealed that the Kyoto chimps had been trained to touch and memorize Arabic numerals in random position for nearly a decade before the now controversial 2007 study.  Ayumu had managed to reach perfection by 2011.  When shown nine numerals for just 60 ms she could now reach perfect accuracy.  As impressive as this is, a decades worth of practice enabled perfection, not a generalized superior chimpanzee working memory.

However, Humphrey proposed yet another interesting possibility for Ayumu’s working memory skill: synaesthesia.  In the most common form of synesthesia, letters and numbers are perceived as inherently coloured.  For humans, this form of synesthesia significantly improves an individual’s ability to recall the order of arbitrary symbols like numbers.  If Humphrey is right, then the chimpanzees in Matsuzawa’s lab may be touching what to them looks like colours belonging to a well-known sequence, as opposed to what they look like to us, blank white squares (Humphrey, 2012).  Future research still needs to be conducted to see if Humphrey’s hypothesis is correct.

Within a span of five years, research on chimpanzee short-term memory has become one of the most controversial areas of inquiry within science.  And that is why I was surprised to see that Tetsuro Matsuzawa decided to showcase Ayumu and the now infamous short-term memory experiment last week at the Association for the Advancement of Science (AAAS).  In the video demonstration, Ayumu performed the 1-9 number sequence experiment, and also was shown to have now learned numbers 1-19.  The popular press has once again used this demonstration to conclude that chimpanzees incredible short-term memory is superior to humans, and is vitally adaptive.  Of course, within academia this is not the current consensus, as I have explained above.

I think what I have learned from following and researching this ongoing drama is that the popular scientific press has an incredible power and responsibility.  For science writers and bloggers that care deeply about scientific discourse, we must make sure that we continue to do exceptional research and remain critical.  If we passively regurgitate we will only perpetuate myth and pseudo-science.  And that cannot benefit future scientific inquiry.


Balter, M. 2010. Did working memory spark creative culture?  Evolution of Behavior, 328: 160–163.

Cook, P. & Wilson, M. 2010. In practice, chimp memory study flawed. Science, 328: 1228.

Humphrey, N. 2012. ‘This chimp will kick your ass at memory games – but how the hell does he do it?’ Trends in Cognitive Science, 16: 353–355.

Inoue, S. & Matsuzawa, T.  2007.  Working memory of numerals in chimpanzees.  Current Biology, 17: R1004-R1005.

Siberberg, A. & Kearns, D. 2009. Memory for the order of briefly presented numerals in humans as a function of practice. Animal Cognition, 12: 405–407.

4 responses to “Working on Working Memory

  1. Pingback: Working hard or hardly working? | Inside Out·

  2. Your post needs a few clarifications. SIlberberg and Kearns trained themselves on the limited hold task for dozens of sessions before they reached the level that Ayumu displayed in his first 10 sessions of the limited hold task (While kearns reached the level almost right away, he had been debugging the program prior to starting). It’s true that the chimpanzees had lots of training in learning the order of numerals for several years ( for Ayumu not a decade, because he was born in 2000) prior to this experiment, but that was without a masking event. That’s not a confound. The humans received far more informal training in the concept and order of Arabic numerals throughout their lives prior to the the experiment, as numbers play a large role in the daily lives of humans. Matsuzawa has retested with shorter latencies, including 90 ms, which is a sign that he is continuing to test the cognitive trade-off hypothesis. I think you go to far in calling his work “myth and psuedoscience” and claiming the matter has been put to rest by Silberbeg and Kearns.. Also, the other things you cite are not additional replications, they are commentaries that are based on Silberberg and Kearns data.

    • 1) I did not directly call the Inoue & Matsuzawa (2007) study myth and pseudoscience. But when popular science pieces continually state that “chimps have better memory than humans” that is myth and pseudoscience. That is hardly proven. And Matsuzawa has not provided evidence to conclusively state as such.

      2) I said a “decades worth of learning” for Ayumu by 2011. I did not state that in reference to the 2007 study. Ayumu improved her abilities between 2007-2011.

      3) Informal training (by being familiar with arabic numerals) would only help humans marginally with the types of memory tests Matsuzawa employed (especially if synaesthesia is widespread among chimpanzees). Either way, it doesn’t matter because humans are able to master the memory-test games that Ayumu has mastered with little practice (comparatively).

      4) In short, I doubt Matsuzawa’s experimental approach will help us better understand the differences between chimpanzee and human memory. I would look forward to him and his team testing their hypotheses with a different methodology.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s