A new study has shown that chimpanzees may be able to determine whether their partners know they are in

danger. This suggests that these primates are able to decide how ignorant or informed their peers are about

an unexpected situation.

The finding, made by a team of researchers at Ohio State University's Comparative Cognition Project,

suggests that chimps share with humans the ability to perceive the knowledge state of a peer, and perhaps

the intention to protect that peer.

Earlier experiments with both rhesus and Japanese macaque monkeys failed to show the same abilities in

those animals. These new results strengthen the argument that in some ways, chimpanzees are closer to

humans than they are to other primates.

The studies were presented Aug. 16 in Chicago at the annual meeting of the American Psychological

Association. Sally Boysen, associate professor of psychology at Ohio State and director of the project, said

the fundamental question for the test was whether one chimpanzee could tell if another was ignorant of a

specific situation, in this case, of a threat or a reward.

Boysen and her colleagues tested three pairs of chimpanzees at the Ohio State colony. Two adult males,

Kermit and Darrell, who had been together for 18 years, were tested, along with a pair of females, Sarah

and Abagail, and a male and female -- Bobby and Sheba.

For the tests, Boysen modeled both a treat and a threat to the chimps. She used grapes, a food the chimps

highly desired, as the hidden treat. A member of the research group hiding with a tranquilizer dart was the

threat. All of the animals in the study had previously been sedated by a dart or had seen a tranquilizer dart

used, and saw it as a threat.

In half of the test conditions, both animals in the pair were able to watch as either the grapes were hidden in

the cage, or a researcher with the tranquilizer dart hid as a predator.

For the rest of the experiments, one animal was placed in an adjoining cage with a clear view of the food or

threat while the other animal was kept off in a nearby room.

Boysen wanted to know if one animal would "tell" the other about the reward or threat. If they did, it would

mean one animal would have to decide how well informed the other was about a given situation.

When she tested the animals with the hidden grapes, absolutely nothing happened. No information was

exchanged between the two chimps.

"You wouldn't expect it to work with the food since no chimpanzee is going to willingly inform another

about the presence of food that they themselves don't have access to," she said.

But when the grapes were replaced by the predator, the results changed dramatically. When Kermit was

released into the cage area where the researcher was hidden out of sight behind a wall, Darrell became

extremely agitated.

He turned to Kermit displaying fear grimaces and alarm vocalizations - two common warning gestures for

chimpanzees. Darrell's hair also stood on end all over his body, which reflected his state of arousal. Kermit

mirrored the same fear responses, turned and left the cage area before getting close to the hidden predator.

"Based on what we believe about the emergence of these skills in humans," Boysen said, "this suggests that

Darrell, in a sense, put himself in Kermit's place. I think Darrell was aware that Kermit couldn't have

known that the predator was hidden in there."

During the research all six animals ran through the same experiments and the results were always the same

-- the chimps "told" their peers of the hidden predator.

"This suggests that one chimp does recognize the different knowledge state in the other chimp," Boysen

said. "Darrell didn't know what the predator was going to do so he had to make a prediction, a rapid

assessment of the situation. He made the decision that there would be trouble if he didn't let Kermit know

about the predator."

Primatologists have long known that chimpanzees in the wild will give off warning calls when they see a

potential threat. Boysen said that these were generalized warnings -- not specific ones. These experiments

removed most of the variables that