Science Watch

In the heart of the Florida Keys, two dolphins have demonstrated in carefully designed experiments that they grasp "relative numerosity," the concept that fewer items is "less" than more of those same items. This original evidence that not only non-humans, but non-primates, grasp basic numerical concepts made a splash in the August issue of the APA's Journal of Comparative Psychology (Vol. 119, No. 3).

Studies have shown that land mammals and birds, particularly parrots, can understand various numerical concepts, and that in primates--chimpanzees, for example--these numerical concepts can be very advanced. Dolphins have long been known to be highly intelligent, clear in everything from playful sociability to their superb echolocation, and efficient teamwork to corral their prey. However, scientists have only recently explored whether dolphins--which are curious and relatively cooperative--have mathematical potential similar to that of primates, including humans.

Given what they knew about some land animals and birds, researchers were eager to extend this knowledge to creatures of the water. Dolphins make a particularly interesting case in that primates and marine mammals--dolphins, whales and porpoises--started to evolve in different directions about 90 million years ago, notes Kelly Jaakkola, PhD, senior research scientist at the Dolphin Research Center in Grassy Key, Fla. As a result, their brains are organized differently and it's been unclear whether dolphins would grasp numeracy.

Jaakkola and her colleagues hoped, she says, "to see exactly what cognitive capacities dolphins share with humans and other 'intelligent' animals and how they might differ."

Their efforts appeared to shed light on these questions, according to Stan Kuczaj, PhD, an experimental psychologist who studies marine mammal behavior and cognition at the University of Southern Mississippi. He says the study was better controlled than previous work with dolphins, enabling researchers to trust the conclusion that dolphins could apply the concept of "less" to a task. Kuczaj, an APA fellow, continues, "The more species that we understand, the better off we are in terms of understanding the evolution of cognitive abilities."

Numbers game

In their pursuit of pinpointing dolphins' abilities, staff at the Dolphin Research Center, an independent nonprofit education and research facility south of Miami, enlisted two dolphins from its pod. Talon, born in 1990 at the center, is a male Atlantic bottlenose dolphin (Tursiops truncatus). Rainbow, another male bottlenose, was collected from the ocean at about 4 years of age in 1983 by another facility. He was about 24 years old at the time training began.

First working with Talon and then with Rainbow (who, distracted by a new female, slacked off for months), the researchers employed a test apparatus (see photo) with two black number boards attached to a length of PVC pipe. When dockside, the dolphins could see these boards on either side of a canvas shelter that screened the trainer from seeing the boards, eliminating a potential source of bias. An experimenter standing behind the trainer's screen rotated the pipe backwards and, out of the dolphins' view, attached different arrays of white dots to the boards. For example, one board might have three dots; the other five.

The experimenters varied dot diameter and placement so that the dolphins could not respond on the basis of non-numerical cues. The design also dissociated discrete numerical quantities (more or less items) from more continuous quantities (volume or surface area), thus isolating the number concept only.

In daily sessions of 20 to 30 minutes, the researchers trained Talon or Rainbow to tap the board with fewer dots with his rostrum (extended beak) after a trainer said "less" and used a hand signal. The experimenter, who observed the dolphin through a periscope over the trainer screen, reinforced a correct response by blowing a whistle; the trainer then provided fish and social interaction. Both experimenter and trainer ignored incorrect responses.

The dolphin's initial training used just a few specific number pairs (for example, two versus six, one versus three, three versus seven). Once the dolphins consistently chose the lesser amount on these training pairs, the researchers started experimental sessions using brand new pairs of dot arrays.

In the test sessions, the researchers presented Talon or Rainbow with all possible pairs of numbers between one and eight, except the training pairs, which were chosen randomly. The experimenter coded the dolphin's choices during the sessions; a second experimenter later checked those choices on videotape.

Less is more

Talon chose the board with fewer dots--the right answer--on 83 percent of the generalization trials, indicating that he could recognize and represent numerosities on an ordinal scale. "Because every trial presented a new combination of dot sizes and positions," say the researchers, "the only way to succeed was to recognize the numerosity of the displays."

Rainbow was equally adept, choosing the board with fewer dots on 82 percent of the trials. What's more, both dolphins tended to make more errors on trials in which the ratio between the two presented numerosities was small--a finding predicted by Weber's law of psychophysics, which states that a smaller difference is harder to perceive, such as between two and three instead of between two and eight.

The researchers speculate that the dolphins' performance fits with the "analog magnitude model" thought to help animals (including humans) grasp amounts along a continuum, like a mental number line.

Kuczaj isn't sure. "They can do something akin to counting when the numbers are side by side," he notes, "but whether they see those numbers as part of the same continuum is hard to tell." Evidence of abstraction, such as that dolphins can learn number symbols and manipulate numbers, would address this problem head-on.

Why would dolphins have this conceptual capability? "Numerical concepts could be useful in things such as foraging situations or in assessing competition," Jaakkola speculates.

Kuczaj agrees. "More and less are important in foraging: It's better to swallow a big fish than a small fish. Or when corralling fish, you might want a good number but not so many that it's overwhelming."

"From what dolphins have been shown to be capable of with respect to artificial language training and echolocation tasks, it does not surprise me that they would have [this] ability," says Laela Sayigh, PhD, a marine biologist with the University of North Carolina at Wilmington and the Woods Hole Oceanographic Institute in Massachusetts. "In the wild it would be very useful to keep track of which areas were richer food sources."

School isn't out for Talon, Rainbow or their fellow dolphins--nor, hopes Jaakkola, for their fellow creatures. "We know extremely little about other cetaceans' cognitive abilities," she explains. "It would be really interesting to test a beluga or an orca." The Dolphin Research Center will continue to study numerical cognition, but Jaakkola refers to a wealth of cognitive topics that have been studied extensively in primates and other animals, that haven't yet been tested in dolphins. She says, "We would like to start filling that gap."

Rachel Adelson is a science writer based in Raleigh, N.C.

Further Reading

For more information about this research, visit