Monkeys understand numbers across senses

Monkeys can match the number of voices they hear to the number of faces they expect to see, Duke University scientists have found. The finding indicates that numerical perception is truly an abstract concept and not just a function of a particular sense, said the researchers. The experimental approach also will lead to further studies exploring whether human infants, before they have a verbal capacity, understand similar abstract numerical concepts, they said.

The researchers, led by Elizabeth Brannon of Duke University and Asif Ghazanfar of Princeton University, published their findings in the June 7, 2005, issue of Current Biology. Brannon is in the Center for Cognitive Neuroscience and the Department of Psychological and Brain Sciences. Lead author on the paper was graduate student Kerry Jordan in Brannon’s laboratory; and Nikos Logothetis of the Max Planck Institute for Biological Cybernetics in Germany was a co-author. The research was sponsored by The National Institute of Child Health and Development, the John Merck Fund, the Max Planck Society and a National Science Foundation graduate fellowship.

In their experiments, the researchers played rhesus monkeys the sound of natural “coo” calls made by unfamiliar monkeys, either with two or three animals making the calls. At the same time they gave the monkeys a choice to look at video images of either two or three monkeys. The researchers found that the monkeys overwhelmingly chose to look at video images that matched the number of monkeys they were hearing. This result is consistent with previous studies that both animals and infants tend to look preferentially at a visual stimulus that matches the sound they are hearing.

According to Brannon, previous studies had yielded conflicting results regarding whether perception of numerical values by nonhuman animals or human infants was tied to the sense used to perceive the number. To resolve the question, said Brannon, the researchers decided to design an experiment that used socially relevant stimuli.

“Our approach really derives from thinking about why a monkey would need to represent numbers across sensory modalities,” she said. “In the wild, a monkey might hear different monkeys vocalizing and not see them, yet need to know how many animals there are. For example, in a territorial dispute, you could imagine that an animal would want to know, ’Well, how many animals are really about to encroach on our territory?’”

Also, said Brannon, the researchers chose to test each monkey only once and not reward them in any way. This experimental design meant that the monkeys could not learn anything within the experiment and instead were demonstrating their spontaneous numerical abilities. Other aspects of their experimental design avoided drawbacks in previous experiments whereby animals or babies might have used relative intensity or sound duration, rather than the numerical commonalities extending across sensory modalities.

“The results we obtained provide evidence that monkeys spontaneously detect a correspondence in number between two different sensory modalities, and this tells us that language is not necessary to represent number abstractly,” she said. “When we humans apply the word ’three’ to sounds or visual images, we’re using language to link these different sets from different modalities. And the question has been whether an animal without that kind of language based representation can still notice or represent these commonalities.”

According to Jordan, their research team is planning future studies that will use the same experimental design to explore whether human infants have the same cross-modal numerical ability. “The experiment with monkeys has given insight into the evolutionary origins of cross-modal number representations,” Jordan said. “And studies with infants will tell us whether this ability applies to infants before they have acquired language.”