Average bear could be pretty smart

Computer tests of solitary species reveal animals’ ability to learn concepts

10:33am, August 24, 2012
Sponsor Message

American black bears that take computerized tests by pawing, nose-bumping or licking a touch screen may rival great apes when it comes to learning concepts.

Using three zoo bear siblings as classroom subjects, comparative cognitive psychologist Jennifer Vonk of Oakland University in Rochester, Mich., and her colleagues presented pairs of pictures to the bears on a rugged computer screen and gave them food treats for pawing the image from a certain category. To demonstrate learning a concept, bears had to figure out what kind of picture would earn a treat and then pick that kind of image from a new set.

One challenge, picking the portrait of a black bear instead of an image of a person, could be mastered by relying on a mix of visual clues such as furriness or snout shape. But picking out all the animals from non-animals — cars or landscapes, for example — required finding more abstract connections among pictures that didn’t look much at all alike.  

At least one of the three bears showed some capacity at each of the five levels tested, Vonk and colleagues report in an upcoming Animal Behaviour.

Bear behavior has been “very underappreciated,” says comparative ethologist Gordon Burghardt of the University of Tennessee at Knoxville. “They’re very smart and they have large brains.” They also live relatively solitary lives, which make them an important contrast to the mostly social animals tested for complex mental capacities to date.

Researchers have proposed that social living may favor the evolution of mental capacities different from those of solitary life, Vonk says. Highly social animals, such as chimpanzees, may need more skills for cooperating or spotting clues to other animals’ intentions. For loner species, evolution might hone more physical powers such as spatial memory for finding food or the ability to grasp the mechanisms behind tool use.

The tests in the new study were aimed at homing in on the mental capacities of these not particularly social bears. The bears differed considerably in the challenges they met, possibly because of the order in which researchers presented the tests. One animal learned to tell animals from non-animals but didn’t quite get the “bears vs. humans” challenge.

All three bears rose to the challenge of computer use, though. When the biggest, Brutus, realized that someone would give him treats for nosing a screen pressed against the fence of the enclosure, “he wouldn’t let anyone else near the computer,” Vonk says. She spent a lot of time throwing cookies and yelling to herd various bears toward or away from the computer cart. Even with her best efforts, “they would find ways of opening a little door and barging in on each other’s sessions,” she says.

To understand what the results mean will take a lot more analysis of the tasks themselves, says Lars Chittka at Queen Mary, University of London. The new study suggests that bears’ ability to categorize images may be on par with that of honeybees, which don’t have particularly big brains. Yet the bees can learn to categorize diverse images such as flowers, landscapes or green vegetation.

“The real question cognitive scientists need to ask is, what is the computational nature of the task, and what neural circuitry might be required?” Chittka says. “Once we have answers to such questions, we can then ask questions about their evolution, or the required brain size, or the ‘social intelligence hypothesis.’ But not before.”

More from Science News