Monkeys add up like we do
Rhesus monkeys master basic addition in a similar way to humans.
A mathematical competition between two rhesus macaques and fourteen undergraduates has revealed a new similarity between monkeys and college students: their ability to handle basic addition.
In the battle of man versus macaque, students bested the monkeys for overall accuracy at 94% to 76%. But response times during a computerized test of addition were approximately the same in the two groups. Both groups were more likely to stumble as the magnitude of the sums increased.
Such similarities, researchers say, suggest an evolutionary continuity between basic mathematical skills in humans and other primates. The results are published this week in PLoS Biology1.
The fact that monkeys can handle basic arithmetic is not in itself new, as it had been suggested by previous work. If monkeys watch as lemons are placed behind a screen, for example, they will stare longer at the fruit if the screen is lifted to reveal an incorrect sum of lemons2. Their apparent surprise when the number of lemons revealed isn't what was expected suggests the presence of rudimentary mathematical ability, says Jessica Cantlon of Duke University in Durham, North Carolina.
"It’s not math in the sense of a symbolic procedure, the way that humans typically think of math," says Cantlon. Monkeys won't be doing full-blown algebra anytime soon. "It’s a more primitive form."
Cantlon and her colleague, neuroscientist Elizabeth Brannon, tested the extent of these abilities by training monkeys to watch as two groups of dots were displayed briefly on a computer screen. Subjects were then presented with two new sets of dots and asked to select the one that represented the correct sum of the previous set (see video). The task didn't get too hard: the highest sums were always less than 20.
The dots were flashed onto the computer screen for only half a second – too quick for humans to use their verbal skills to count them. "If you make them respond really rapidly, they don’t have enough time to perform some sort of verbal counting algorithm," says Cantlon. "If you really push the system, you can get a performance out of it that’s much more primitive."
Cantlon and Brannon also varied the size and position of the dots, so that study participants could not use estimates of surface area or pattern recognition to help in choosing the correct sum.
The monkeys were rewarded for correct answers with a drink of Kool-Aid; college students received $10 for participating in the study. Students had a handful of trials as practice before launching straight into the test.
Both groups needed about a second to come up with an answer. Accuracy in both groups declined as the numbers of dots got larger, and when the two possible answers were close together in value (determining if 10 + 10 dots = 19 or 20, for example, would be much harder than working out if 10 + 10 dots = 12 or 20).
"When the two [possible answer] numbers get larger and closer together, the probability of confusing them increases," says Cantlon. "That was the case in both humans and monkeys."
The study beautifully extends previous findings, says Marc Hauser, a cognitive scientist at Harvard University. But the two monkeys performed thousands of trial runs before being tested. A critical next step will be to test the animals in the absence of training, Hauser adds. “This will be important,” he says, “especially since the human–animal comparison often breaks down because the former are tested without training and the latter with training.”
- Cantlon, J. F. & Brannon, E. M. PLoS Biol. 5, e328 (2007).
- Flombaum, J. I., Junge, J. A., Hauser, M. D. Cognition 97, 315-325 (2005).