Read an Excerpt
Intelligence of Apes and Other Rational Beings
By Duane M. Rumbaugh David A. Washburn YALE UNIVERSITY PRESS
Copyright © 2003 Duane M. Rumbaugh
All right reserved. ISBN: 978-0-300-09983-6
Chapter One
Introduction
Thorndike tried to dispose of the difficulty (of thought) 55 years ago by denying its existence. He buried thought in 1898, but the ghost insists on walking. Often we decline to say that animals think; but comparative psychology has been unable to avoid concluding that animals have expectancies, insights, hypotheses, conceptual activities, a variable attention and so forth. These are but aspects of thought; and if we cannot deal with the comparatively simple behavior of animals without taking account of thought, how adequate can a thought-less human psychology be? Hebb 1953, 99
The need for a better understanding of animal intelligence has become clear in recent years as, particularly in the laboratory, animals have done truly remarkable things, things that would not have been anticipated even ten years ago. Here we will make the case that to view animals as irrational, unsmart creatures is no longer tenable. In measure, we must now view them as beings with intelligence and rationality appropriate to their species. Otherwise, how canwe account for their remarkable achievements in advanced levels of communication (including, in some cases, language), in counting, and in computer-based tasks? The intelligence of animals is just now coming to be understood, and the search for an appropriate understanding of it has been joined only recently. The following example serves to illustrate the point. How is this example to be understood and appreciated as a natural phenomenon in a natural world?
Panzee, a chimpanzee (Pan troglodytes), age eleven, had been coreared with Panbanisha, a bonobo (P. paniscus). From six weeks of age, these two apes were constant companions, and researchers worked with them intensively. The apes were together day and night and were cared for much as one would care for a human child (fig. 1).
The purpose of this study of corearing was to examine the apes' spontaneous, untutored mastery of the word-lexigrams (defined and discussed in Chapter 9) and to see whether Panzee would come to comprehend human speech. We assumed that Panbanisha would, as her half-brother, Kanzi, had come to do so spontaneously during the course of his early rearing at the Language Research Center of Georgia State University. Both apes did in fact learn to comprehend their word-lexigrams and to comprehend novel utterances of human speech (Brakke and Savage-Rumbaugh 1995).
Our question here is, "What did the apes do with their language skills that otherwise they could not have done?" We take an example of Panzee's accomplishment to illustrate generally what is meant when we assert that apes display intelligence by using their acquired skills and knowledge innovatively to unique advantages in solving novel problems.
A researcher, Charles Menzel at our Language Research Center, placed one of Panzee's favorite foods, a kiwifruit, several feet away from the back of Panzee's fenced exercise yard in an area of dense brush and trees. She appeared to be fascinated with his so doing but did nothing about it until later, when she recruited another researcher, Shelly Williams, to help her. Williams was inside the building and knew nothing about the food having been placed among the twigs and leaves on the forest floor. Panzee first had to attract the attention of Williams, who was nearby, and did so by presenting her hindquarters. When Williams acknowledged this greeting verbally, Panzee went promptly to the keyboard mounted on a wall of her cage and pointed to the word-lexigram for kiwifruit. She held her finger on the lexigram until Williams acknowledged what it stood for-kiwifruit. With that assurance, Panzee moved up into her tunnel that led to a large outdoor cage. Then, covering her eyes with a hand-her standard way of uttering "hide"-she gestured with outstretched arm to the tunnel door. Williams surmised that Panzee was telling her that a kiwifruit was hidden somewhere. Williams said, "You want me to go outdoors?"-whereupon Panzee ran through the tunnel outdoors. At first, Williams didn't know what she was to do. Panzee then seemingly set about giving her specific instructions. Panzee vocalized and gestured with an extended arm toward the general area where the kiwifruit had been placed. Williams understood that she was to go the area behind the cage. Panzee then went to the spot closest to the kiwifruit and pointed with her finger to the fruit's specific location. As Williams approached the spot to which Panzee pointed, Panzee began to make low-frequency vocalizations of excitement-all the while focusing her eyes specifically upon the kiwifruit. Williams could not see the fruit at first because it was well nested in among sticks and leaves. Panzee became increasing excited whenever Williams approached the location of the kiwifruit. After a brief search, Williams found the kiwifruit, where-upon Panzee headed back indoors after making a beckoning hand movement to Williams to return indoors. Thus Panzee got her prized kiwifruit through the coordinated use of her several skills. She had solved a novel challenge, one never encountered before.
This procedure has been replicated dozens of times, with a host of foods and variety of personnel to help Panzee. Panzee has no problem remembering what food was hidden and where it is hidden. Although typically she will recruit someone within twenty-four hours to help her, she has on occasion waited several days and even weeks to ask for assistance to get an item specified by name. Panzee is only rarely incorrect in her use of lexigrams in identifying the hidden object and in pointing to the several locations where the food has been hidden (Menzel 1999). Only rarely has she recruited someone to assist her when, in fact, no item had been hidden.
Without mastery of her word-lexigrams, Panzee would not be able to identify the food hidden and the materials that hide it. Her early rearing afforded her experience in pointing and gesturing in communicative settings. All that said, the impressive behavior is that years later she put various skills together, all with speech comprehension-not by conditioning, shaping, or by selective reinforcement. Rather, the skills were put together by her highly complex brain and its covert cognitive processes. No trial and error was involved. Panzee was competent even on her first challenge with hidden food-the episode recounted above in which she recruited Shelly Williams. She was able to structure the solution to a problem that never before had been faced by her or modeled for her by others.
Now how did Panzee do this? What enabled her to do it? Without a reinforcement history that would quite probably lead her to the task solution in some specific manner, what mediated the solution?
Pointing at Intelligent Behavior
As we shall discuss in some detail, language skills and enriched early rearing were critical requisites to Panzee's creativity in this case. We will investigate the ways that silent learning and covert operations afforded by a complex brain and enriched rearing generate emergent behaviors, skills, and invention. Panzee provides us with a fine example with which to start this book that focuses upon questions of animal intelligence.
We cannot account for intelligence in terms of basic reinforcement of responses to stimuli. In a relatively recent book, Cognitive Aspects of Stimulus Control, Honig and Fetterman (1992) tacitly reject that the central challenge for psychology was to "define the stimulus." Rather, they argue that even the analysis of stimulus control, as advocated by traditional behaviorism, can benefit from the use of cognitive constructs. They recognize, as do we, that even relatively simple animals can learn seemingly complex concepts with great facility if the materials are relevant to their environments. When tested in the artificial surroundings of a laboratory with methods of yesteryear, animals can appear to be the plodding trial-and-error learners that historically they have been declared by Descartes and others even to this day. Accordingly, we will argue here that we must look beyond the basic conditioning procedures of traditional behaviorists to understand both animal and human intelligence. At the same time, we recognize that the long and rich history of research into the parameters of both respondent (Pavlovian) and operant (Skinnerian) conditioning have provided us with a vast repository of valuable findings that are not to be rejected. Those findings need only be embraced with a new framework that allows us to look beyond the procedures of traditional behaviorism if the full richness of animal intelligence is to be defined.
Historical Perspective: (Nonhuman)Animals as "Beast Machines"
For much of history, animals have been viewed essentially as wild beasts or as chattel. Indeed, the Cartesian view (Descartes 1637) that animals are senseless beast machines probably is still more commonly believed (at least tacitly) than is Darwin's (1859, 1871) postulate that humans share psychological, as well as biological, dimensions with animals. Thus animals remain important sources of food, sport, work, and pleasure. Though admired for their beauty, strength, and agility, animals generally have been declared dumb-that is, without intelligence and language.
In fact, there has historically been great pressure to discourage writing and the collection of data that might document psychological dimensions shared with animals, even with our nearest primate relatives, the great apes. G. J. Romanes's initial efforts to document animal intelligence through use of anecdotes, in an 1882 book called Animal Intelligence, proved to be nothing short of a disaster. His report, which was intended to be followed by empirical research, was so ridiculed that the whole concept of animal intelligence was discredited. The American psychologist Edward L. Thorndike, who also wrote a book called Animal Intelligence (1911), meticulously and colorfully documented the mindless trial-and-error behavior of animals in problem boxes. He concluded that animal problem solving and learning basically was highly mechanistic, sans intelligence, and should be conceptualized as the gradual establishment of bonds between stimuli and randomly successful responses.
This perspective became the basis of the school of behaviorism founded by John B. Watson (1913) and supported by the later behaviorists, notably B. F. Skinner (1953), Clark L. Hull (1943), Edwin R. Guthrie (1935), and Kenneth W. Spence (1956, 1960). All of these scientists rejected rationality as a parameter of behavior-either animal or human. Their conclusions, based on studies of rats, pigeons, and chickens, were advanced as valid for understanding all aspects of human behavior! All behavior, except perhaps for language, was to be explicated by studies of rats making choices in mazes or pigeons pecking targets in a box. Language was allowed as a possible human distinction (although Skinner 1957 generated a complex and largely unsuccessful treatment of "language behaviors" based on the same associative-learning mechanisms that are evident in rats and pigeons). Language, it was assumed, allowed for a new psychology based on thought, and because earlier scientists of the twentieth century were convinced that no non-human had language, they were convinced that no nonhuman could think.
Strange!
It was held that it was unsafe to posit thinking in nonhuman animals. Why? Because nonhuman animals could not tell us of their thoughts and experiences, as we do with one another. Interestingly, the verbal report of humans was accepted as a valid reflection of consciousness, thought, intelligence, values, and so on. Yet a moment's reflection makes it clear that even as two humans talk with each another, one has no proof whatsoever that the other's speech accurately reflects his thought. One human assumes that because her interlocutor is a human, he shares the private experiences of her own self. Although this assumption is probably generally true, there is no way of proving the inference. Thus in a real sense we trustingly "anthropomorphize" among ourselves all the time. We assume that it is proper to attribute to other humans aspects of our own private lives. But to do so with non-human animals had been declared by behavioral scientists as unscientific, unwarranted, silly, even dangerous. Yet now, in light of comparative behavioral research of the past fifty years, it is no longer tenable not to allow for advanced operations of cognition and even language-at least in our nearest living relatives, the great apes (Savage-Rumbaugh and Lewin 1994; Savage-Rumbaugh et al. 1993; Rumbaugh 1990).
During the heyday of behaviorism, opposing views, although rare, were also influential. Robert M. Yerkes unabashedly discussed primate intelligence in his monograph of 1916, The Mental Life of Monkeys and Apes, reflecting a comparative perspective that included learning, intelligence, and the mental lives of apes, monkeys, mice, rats, frogs, crabs, turtles, dogs, cats, pigs, and even earthworms. During World War I, Wolfgang Köhler (1925), a cofounder of the school of Gestalt psychology, studied chimpanzees' problem solving and concluded that it could be insightful. That is, apes were able to solve novel problems by reasoning and creativity-not simply by trial and error as held by Thorndike. In the decades that followed, the avowed behaviorist Edward C. Tolman (1948) created a cognitive framework that he called "purposive behaviorism" to account for maze learning by rats. Tolman ascribed learning to the perception of relationships, called "cognitive maps," between such stimuli as those that depict relationships between routes to the goal box of a complex maze.
The key issue that divided Tolman and Köhler from Thorndike and his followers was how to account for the observed performances of animals. Thorndike considered the animal to be like a machine that generated a series of random responses until one of the responses succeeded in bringing about a "satisfying state of affairs." The response was hence "stamped in" or learned, and the probability that the response would recur in that situation in the future was increased. Clearly this view credits animals with little or no intelligence! By contrast, Tolman and Köhler viewed organisms as thinking beings that, given the opportunity, would evaluate situations and choose responses that were likely to satisfy a purpose or achieve a goal. Such organisms could learn passively as well as actively, through observing the performances and successes or failures of others.
Harry F. Harlow (1949; see also Chapter 17) provided a bridge between these contrasting perspectives with his research on learning set, or "learning to learn." Harlow studied changes in learning proficiency by monkeys over a long series of problems. Initially their learning was essentially like that of Thorndike's trial-and-error improvement-slow and arduous. With experience, however, the monkeys became increasingly skilled and learned new problems with apparent insight-quickly and nearly errorlessly. Thus Harlow defined the course whereby certain protracted experiences provide for the emergence of insightful, one-trial learning, even of novel problems of a kind that initially required gradual trial-and-error learning. Moreover, Harlow's learning-set procedure became the basis of numerous subsequent attempts to assess species differences in learning and transfer ability and, by extension, intelligence.
(Continues...)
Excerpted from Intelligence of Apes and Other Rational Beings by Duane M. Rumbaugh David A. Washburn Copyright © 2003 by Duane M. Rumbaugh. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
