Monkeys and Humans Form Categories in Strikingly Different Ways, New Research Shows

Monkeys' learning is perceptually based; humans rely on rules

Release Date: September 21, 2004 This content is archived.


BUFFALO, N.Y. -- The ability to form categories is a crucial cognitive ability shared by humans and animals. It plays an important role in the way in which humans and animals behave toward objects in their worlds.

Some monkeys, for example, have even developed call signs to name the categories of snakes, eagles and big cats that threaten them in their environment.

New research reported in this month's Journal of Experimental Psychology: General, sheds new light on the difference between the way humans and animals learn categories, and whether different species use the same processes and brain systems to do so.

"We were very interested in learning whether human consciousness or human language let humans categorize in different or more powerful ways than animals do," says lead researcher J. David Smith, associate professor of psychology at the University at Buffalo.

In experiments conducted at Georgia State University's Language Research Center, Smith and colleagues J. Paul Minda, assistant professor of psychology at University of Western Ontario, and David A. Washburn, director of Georgia State's Language Research Center, gave humans and rhesus monkeys six qualitatively different category tasks that required different learning approaches. The four monkeys completed more than 300,000 experimental trials overall, using a joystick to make their category decisions and earning food rewards whenever they were correct.

The results of this survey of category tasks are striking, Smith says.

The researchers showed that for monkeys, task difficulty depended on how strongly objects in a category resembled one another. "Given strong perceptual resemblances, monkeys learn well. Given weak perceptual resemblances, monkeys learn poorly," Smith explains.

In sharp contrast, for humans, task difficulty depended on the character of the rules that humans can hold in mind during the task, the researchers found. "Given succinct, economical rules, humans learn well. Given complex, elaborate rules, humans learn poorly," Smith says. "In these tasks, monkey category learning was more perceptually based. Human category learning was more rule based."

The researchers illustrate this striking species difference this way: Suppose that black triangles and white squares belong together in Category A, whereas white triangles and black squares belong together in Category B. If you focus on just black, or white, or triangle, or square in trying to learn, you become confused because the cues singly do not separate the categories. There are white and black things, and squares and triangles, in both categories. Monkeys grow confused, the researchers say, because they tend to focus on single perceptual cues or because they cannot easily learn to include exact perceptual opposites (black triangles, white squares) in the same category. They learn this kind of task poorly.

Instead, one needs to learn this kind of task by considering the relation between the shape and color cues. You could use a simple rule for this task: If triangle, Black (A), White (B); If square, White (A), Black (B). Humans do so, and they learn this kind of task easily.

Humans are generally comfortable using rules to let them include perceptual opposites within the same category, the researchers say.

"For example, we know that it is okay to wear black shoes and black socks, or white shoes and white socks. These perceptual opposites are the same for being a defensible fashion statement," Smith says.

"But we know that it is less okay to wear black shoes and white socks, or white shoes and black socks. These perceptual opposites are the same for being a nerd fashion statement. Perceptual opposites, but simple rules, and the human categorization system thrives. Not so that of monkeys," he adds.

In the case of monkeys, the research of Smith and co-researchers may be giving scientists a glimpse at the stem categorization system, based in perceptual resemblance, from which the system of humans emerged during cognitive evolution. This raises the interesting question of whether chimpanzees might possess a categorization system that is transitional between that of monkeys and humans, the researchers say.

In the case of humans, the research of Smith and colleagues may be showing one of the profound advantages conferred on human cognition by the evolution of explicit thought and language.

Media Contact Information

Patricia Donovan has retired from University Communications. To contact UB's media relations staff, call 716-645-6969 or visit our list of current university media contacts. Sorry for the inconvenience.