Do animals think?

To follow is an excerpt from the CQ Researcher report on "Animal Intelligence" by Marcia Clemmitt, October 22, 2010.

Just a few decades ago, the jury was still out on animal intelligence. But agreement now is virtually universal among scientists that animals of all kinds perform remarkable feats of mind — including actual reasoning. However, while some argue that several species perform very high-level cognitive activities including “metacognition” — loosely defined as “thinking about thinking” — others contend that studies of such complex thought are prone to experimental designs that tempt researchers to overinterpret.

“Abstract concepts are extremely widespread in the animal kingdom, all the way down to bees,” says Peter Carruthers, a professor of philosophy at the University of Maryland, College Park. Studies in which bees navigate a maze demonstrate that the insects grasp the concepts “same” and “different” because they can learn and follow a navigational plan that requires them to turn right, for example, when they spot a design that's the same as one they previously saw but turn left when the picture is different, he explains.
Just How Smart Are They?

Furthermore, “Bees have a cognitive map and can find their way home even if they've never flown that way before,” Carruthers says.

Some animals, such as monkeys, show a fairly sophisticated ability to form “representations” of things in their minds, rather than being able to reason only about real-life objects that they can see in front of them at the present time, says Herbert S. Terrace, a professor of psychology at Columbia University in New York City.

For example, a monkey can memorize an arbitrary sequence of photographs, and then, when later shown only two pictures from the sequence, arrange them in the order of the original sequence, even though the picture sequence isn't in view, Terrace says. To do this, the animals must consult mental “representations” of what they've seen — evidence of an ability to “think without language,” Terrace says.

Recently, wild crows in a New Zealand laboratory experiment showed an especially remarkable cognitive feat — “insight” — the ability to devise a correct solution to a novel problem without doing any trial-and-error manipulation in the real world, says Carruthers. Confronted with a situation that required the birds to use one stick to retrieve a second longer stick, which they could then use to retrieve a food reward, one bird “looked at the setup for about a minute, and then accomplished it on the first try,” he says. [Footnote 12]

“What would a human do to arrive at the answer? Go through possibilities” mentally, waiting for an “insight.” It's hard to escape the conclusion that the crow did something similar, Carruthers says.

Recently many studies have examined whether some animals show forms of “higher” thinking traditionally considered the sole province of humans, such as a “theory of mind” — awareness that other animals or humans have thoughts going on inside them, just as one does oneself — and metacognition. Unlike humans, animals can't tell us what's going on in their minds, so to examine metacognition researchers set up experiments that give animals a way to demonstrate through their behavior that they recognize that they're in a certain mental state — such as being uncertain about which of two test answers is correct.

In a typical experiment, an animal is offered a test with two possible answers — such as that one musical tone is higher or lower than another — and has correct choices reinforced with a substantial food reward. Once the animal knows what constitutes a correct answer, the task is made harder — the tones get closer together, for example. At this point, the animal gets a third response option — usually the choice to opt out of choosing either of the other answers — for which there is a guaranteed, but relatively small reward.

Opting for this less rewarding “uncertainty” response demonstrates that the animal recognizes its own mental state — i.e., that it's uncertain which of the two other choices will yield the big reward that's reserved for getting the right answer, says J. David Smith, a professor of psychology at the State University of New York at Buffalo. When macaque monkeys and humans take the same test, they choose the “I'm uncertain” response at the same rate. That's evidence, he says, that, like the human test subjects, the monkeys recognize their own mental feeling of uncertainty — a mark of “metacognition.”

In a similar experiment by Columbia's Terrace, a monkey was presented with two “confidence icons” just after the animal had given its answer on a perception-based test similar to the musical tones test. “One icon signified high confidence; the other, low confidence” in the answer the monkey has just given, Terrace explains. [Footnote 13]

“Choosing the high-confidence icon was a ‘risky’ bet.” If the monkey chose that icon after it had given the right answer on the test, it won three tokens, but if the monkey chose the “high confidence” response after a wrong answer, it lost three tokens, “and they really don't like that,” Terrace says. Choosing the “low-confidence” icon always got a reward of one token.

Monkeys chose the high-confidence icon more often after they had given correct answers and the low-confidence icon more often after they had given incorrect answers, the exact same response one gets from humans — who can not only feel a mental state like “uncertainty” at the moment we experience it but can also remember the feeling. The experiment shows that a monkey “can monitor its accuracy on perceptual tasks and transfer that ability to monitoring its memory” — i.e., it can consult an after-the-fact “mental representation” of the feeling of uncertainty it previously experienced. This activity indicates some level of metacognition — an ability to think about mental states — wrote Terrace and his fellow researchers. [Footnote 14]

But other scholars say that so far no experiments show that animals can use abstract concepts to reason about their own minds, or about anything else except concrete objects, a clear limitation to their thinking.

“To some extent, this is a verbal confusion,” says Carruthers. Experiments do show that some animal species — just like humans — are aware of their feelings of uncertainty, but that awareness alone doesn't meet the standard definition of “metacognition,” as it's used in human psychology. True metacognition requires actual “thinking about thinking” — reasoning based on one's awareness of the state — and that hasn't been proven, Carruthers argues.

Evidence has shown that many animals form abstract concepts based on sensory perceptions, but no evidence actually shows that they can form concepts about things that they cannot see or touch, like mental states, says Daniel J. Povinelli, a professor of biology at the University of Louisiana, at Lafayette.

“The question is whether the experiments as designed have the power to” produce these more far-reaching conclusions — such as that “monkeys and parrots are interpreting their own mental states” or “that crows think about the principles of physics.” While it's possible the animals do these things, “there is simply no evidence that they do,” Povinelli says.

For example, after observing many individual instances of fellow chimpanzees pursing their lips and bristling their fur just before hitting or charging them, chimps certainly form a catch-all concept — like “threat display; better look out!” — to reference such occasions. [Footnote 15] There is no evidence, however, that a chimp goes beyond this concrete representation to form a concept about some state existing within the pursing, bristling chimp that motivates the behavior — as a human would do by positing “anger” or “aggressiveness,” for example, says Povinelli.

From experience with lifting things, both a chimp and a human can develop the concept of “heavy” and sort objects by whether they're “heavy” or “light,” for example, he says. But humans routinely take abstraction much farther, generalizing beyond concrete objects to things we can't see or touch, for example, by applying the concept of “heavy” even to nonphysical things like sadness, as when we have a “heavy heart,” and by immediately realizing that an object from the “heavy” pile is the one to choose, if the goal is to knock another object over, he says.

“We haven't seen any evidence” that chimps or other animals can handle these levels of abstraction, Povinelli says.

The Issues:

* Do animals think?
* Do animals use language?
* Are animal and human minds more similar than once thought?

For more information see the CQ Researcher report on "Animal Intelligence" [subscription required] or purchase the PDF

[12] For background, see Alex H. Taylor, Douglas Elliffe, Gavin R. Hunt and Russell D. Gray, “Complex Cognition and Behavioral Innovation in New Caledonian Crows,” Proceedings of the Royal Society B: Biological Sciences, April 21, 2010.
[13] Nate Kornell, Lis K. Son and Herbert S. Terrace, “Transfer of Metacognitive Skills and Hint Seeking in Monkeys,” Psychological Science, January 2007, p. 64, .
[14] Ibid.
[15] For background, see Daniel J. Povinelli and Jennifer Vonk, “Chimpanzee Minds: Suspiciously Human?” TRENDS in Cognitive Sciences, April 2003, p. 157,.


Anonymous said...

Interesting. While this study cannot reveal whether animals experience the type of metacognition most humans do, it is clear that many animals are intelligent and can reason. It is clear, also, that they are aware of their surroundings.

Yes, animals think. Animals feel. We should treat them far better than we do.