Behavioral researchers at Duke University announced "the strongest evidence to date" that monkeys and humans share a basic system for comparing numerical values. The finding could help illuminate the evolution of mathematical skill.
"There's a general misconception that numerical abilities are very culturally specific and dependent on language," said Elizabeth Brannon, an assistant professor of psychological and brain sciences and the co-author of a study published in the May issue of Psychological Science. "We're tapping a representational system that is very primitive and very natural."
The system is based on ratios, which researchers believe play a more important role than absolute values in discriminating between numerical representations. For example, 50 peanuts look just like 51 at a quick glance, but three peanuts are easily distinguishable from two, even though each set is one number apart. Brannon and the study's lead author, graduate student Jessica F. Cantlon, were able to demonstrate that this ratio-dependency exists in both humans and monkeys.
Advertisement
In their experiment, two monkeys—Feinstein and Mikulski, named after the female US senators from California and Maryland, respectively—ranked pairs of even values between two and 30 by touching a video screen. (The value of four, for example, was represented on-screen by four dots inside a box.) Human subjects performed the same task but were told to do so as quickly as possible and not to count out each set of dots to themselves—basically, to behave like monkeys.
The humans were slightly more accurate than the monkeys—87% vs. 80%—but the monkeys completed the task faster. When controlling for speed, the results show that humans and monkeys share an equally accurate, nonverbal system of numerical representation.
"Adult humans, you'd think, would be flawless at this task," said Brannon.
But when forced to react quickly, humans access a more instinctual method of counting, she added.
"We can talk about the fact that 29 and 30 are equally different from two and three," she said, "but we don't actually treat them that way."
According to Cantlon, this primitive system forms a foundation for the more complex mathematical abilities humans develop later in life.
"People assume that language is what allows humans to be so complex, and people have said for a long time that math is dependent on language," she said. "By showing that there are these commonalities between monkeys and humans, we can look at how brains might change as a function of language."
In addition to the system for comparing values, Cantlon and Brannon also showed that monkeys may not have an upper limit to their numerical abilities. While previous studies had stopped at a value of nine for the sake of simplicity, Feinstein and Mikulski, the monkeys in this study, correctly ranked the values 10, 15, 20 and 30, without additional training.