Low Plaza

Research Shows that Non-human Primates Can Become Expert List Learners Without the Benefit of Language

By Joseph Kennedy

Herbert Terrace

A newly-published study demonstrates that monkeys have significantly higher thinking skills than previously shown.

Herbert S. Terrace, professor of psychology and psychiatry, working with two former graduate students, Lisa Son, now assistant professor of psychology at Barnard College, and Elizabeth Brannon, now assistant professor of psychology at Duke University, taught monkeys to memorize sequences made up of seven arbitrary photographs, and to keep four of those sequences in mind at the same time.

The experiment offers what is arguably the strongest evidence to date of the intellectual abilities of non-human primates, and suggests that monkeys can think about lists logically, notwithstanding their inability to learn language. The team published their findings in the January issue of Psychological Science.

This experiment is an extension of Terrace's earlier research that undermined claims about meaningful and grammatical use of symbols by "language-trained" apes. Just the same, Terrace believed that animals could think without language. With Brannon, he showed that monkeys could learn to order numerical stimuli in an ascending or descending order. The experiment reported in Psychological Science carries that research one step further by showing that monkeys can learn to order unrelated photographs.

 

The monkey's task was to touch, in a particular order, seven photographs that were presented simultaneously on a touch-sensitive video monitor. From trial to trial, the positions of the photographs were varied randomly to prevent the monkeys from learning the required sequence as a series of rote motor responses. Four monkeys were trained to learn four 7-item lists in this manner. The odds of guessing the correct order in which to touch the photographs was less than one in 5,000. The monkeys not only learned, at a high level of accuracy, the correct order in which to respond to the photographs on each list, but they became progressively more efficient at deducing the correct order with each new list.

"This is the equivalent of a child memorizing, for example, the seven days of the week, the first seven letters of the alphabet, the names of the first seven numbers and the first seven months of the year, reciting each sequence in the correct order, and learning each sequence more rapidly than the previous one," said Terrace. "That is a complex problem for humans, even with verbal labels such as, 'first,' 'second,' 'third,' etc."

After the monkeys learned four 7-item lists, they were given another challenging task. They were tested with all possible pairs of the 28 photographs that were used to construct those lists (336 pairs in total) to see if they could chose which of the two photographs occurred earlier on the lists from which they were drawn. (Photographs that occupied the same position on different lists were excluded.) The monkeys responded correctly to 91 percent of the pairs of photographs drawn from different lists on the first trial on which each pair was presented.

Of particular interest were the monkeys' reaction times to the first item of each pair. The larger the gap between the original ordinal positions of the photographs, the shorter the reaction time. This 'distance effect' is also observed when human subjects are asked to order pairs of randomly selected letters from the alphabet, e.g., which comes first, l or f, f or r, c or d, etc. The further apart the letters, the shorter the reaction time. It seems that both monkeys and humans position each member of a pair on a mental line and then compare their positions on that line to decide which came first. The bigger the separation, the easier it is to make that judgment.

"The sequences that these monkeys learned are by far the most difficult lists mastered by a non-human primate, including those trained in experiments on their linguistic and numerical abilities," Terrace said.

"We believe that the upper limit of a monkey's serial expertise is even higher," noted Son. "The ease with which the monkeys learned 7-item lists and the steady decrease in the number of sessions they needed to master new lists suggests that they could learn such lists more rapidly and also master longer lists."

Son played a central role in conducting and analyzing the experimental data, which involved direct interaction and observation of the rhesus macaques during the extensive two-year project, which was supported by grants to Terrace from the National Institute of Mental Health, and a fellowship to Brannon from the National Science Foundation.

The researchers believe their experiment demonstrates that the monkeys use a precursor of what investigators of human cognition refer to as declarative knowledge -- knowledge acquired rapidly and logically. This is in contrast to procedural knowledge, which is inflexible and acquired slowly through repetitive training on a particular problem.

"Although the subjects of this study lack the ability to declare their knowledge verbally, the breadth of their serial expertise suggests that they have all of the other features of human declarative knowledge," Terrace said.

Published: Jan 15, 2003
Last modified: Jan 14, 2003


Search Columbia News    Advanced Search  Help

Phone: 212.854.5573    Office of Public Affairs