Kanzi: The Ape at the Brink of the Human Mind

by Sue Savage-Rumbaugh

  In light of the Sherman and Austin project, it is worth considering whether the probable parallel evolution in human prehistory of the skills of symbolization, tool-use, and interanimal communication was accidental. The answer, it seems, is that they are closely linked, within a complex social nexus. In such an evolutionary setting, selection for bigger brains would have been inevitable.

  For me, the human countenance of Sherman and Austin’s communication argued strongly that they understood what they were doing. I knew, however, that others would not be equally convinced by expressions. To most people, the wealth of information available from the ape’s face is unreadable; to the unaccustomed eye, in fact, all apes look alike. Thus it was important to demonstrate beyond reasonable doubt not only that Sherman and Austin could communicate with each other, but also that they indeed knew that they were communicating. That is to say, we had to demonstrate that Sherman knew what he needed to tell Austin and why, and vice versa.

  How could we prove that Sherman knew that Austin did not know which food had been hidden or which tool was needed? Recently, this issue has been given the somewhat confusing appellation, Theory of Mind. The idea behind this theory is that an important evolutionary breakthrough occurred when our ancestors began to realize that the knowledge that any one person had could differ from the knowledge that others had. Hence, minds were viewed as existing in different states of awareness. When the first individual became cognizant of this, it is said that she developed the theory that others had minds and that other minds differed from her own.

  Did Sherman know that Austin’s mind or knowledge state differed from his own? It seemed so; otherwise, why would he bother to tell Austin which tool he needed or which food was hidden in a container? As experimenters, we had set up the contingency that both chimps needed to ask for the correct food. But it was the chimps who began to attend to each other, to coordinate their communications, and to correct each other’s errors. For me, as I watched them, the question was reversed: How could they not know that they were communicating? Yet I knew that most other people would not have the opportunity to watch them as I did and thus would remain skeptical simply out of lack of knowledge.

  We could not simply ask Sherman if he knew that Austin’s state of knowledge (or mind) could differ from his own. Consequently we decided to turn off the keyboard to see if they would attempt to communicate their privileged information in some other way. Every bit of training they had received to date revolved in one way or another around using the keyboard to communicate. We felt that if they attempted to communicate at all without it, we would have to conclude that they recognized that the reasons underlying the communicative process had to do with differing states of knowledge and the consequent need to share information.

  So we baited a container of food with Peter Pan peanut butter while Austin watched and Sherman waited in the other room. We then took Austin and the container back to where Sherman was waiting in the adjacent room. Knowing that Austin had no sounds or sign at his disposal for communicating, we thought to leave in the room, on the floor, food labels taken off cans, boxes, and bags. The chimpanzees had never used these labels for communicating, nor had we ever purposefully set about to teach them these manufacturers’ brand labels. Yet they always helped us prepare the food and thus saw the containers with these labels on them many times.

  We thought that if they recognized these labels, perhaps they could think to use them as a means of communication if the keyboard were turned off. There were a number of very large ifs to be addressed in this situation. First, would Austin recognize that he needed to tell Sherman what was in the container that he, and he alone, had seen baited, even though the keyboard was turned off? And second, if Austin knew that he needed to communicate to Sherman about the peanut butter, would he recognize that the Peter Pan label, along with many other labels, was lying on the floor of the room? Third, if he recognized the label, would he think to show it to Sherman as a way of telling him the kind of food he had seen hidden? And fourth, if Austin did pick up the Peter Pan brand label and show it to Sherman, would Sherman recognize it and realize that Austin was trying to tell him that it was peanut butter in the container?

  Much to our amazement, Austin and Sherman had no difficulty with this situation at all. When Austin found the keyboard was off, he immediately noticed the food labels scattered on the floor and then picked up the Peter Pan label and looked over at Sherman. Sherman stared at it intently for but a moment, then used his keyboard (which was on) to ask for peanut butter. Immediately after this, we took Sherman out to watch a container of food being baited with Welch’s grape jelly and left Austin behind, thus reversing the situation. Sherman tried a bit harder than Austin to get the keyboard to work upon his return, but then, like Austin, he began to look around, noticed the Welch’s grape jelly label, and picked it up and put it in his mouth, with the printed side down. He then proceeded to show it to Austin, not quite realizing that Austin could not see it because it was upside down. Austin did nothing. Sherman shrugged then put the label down, turning it over as he did so. Austin then saw what it said and immediately went to his keyboard and asked for jelly.

  We continued to alternate trials in this manner through thirty different foods with manufacturers’ labels, including M&M’s, Jello pudding, Doritos corn chips, and so on. Austin and Sherman knew all thirty of these labels and used them essentially without error, from the very first trial forward, to tell each other the food that they had seen hidden in the container. They did this in spite of the fact that no one had ever taught them that it was possible to communicate using the brand names that manufacturers put on their food, and in spite of the fact that no one had ever taught them anything about communicating that did not involve a keyboard. Why would they have utilized a symbol system no one had taught them unless they recognized the need to share their differential knowledge? Even more intriguing was the question of how they had come to learn what each of these brand names represented, with no training at all. It had taken months for them to learn many of the lexigrams, and yet they knew all thirty of these manufacturers’ symbols with no effort on our part at all. I began to wonder if we had not been using both the wrong symbol system and the wrong approach all along.

  It was one thing to show that Sherman and Austin knew that they were communicating and that they even used a novel symbol system to do so. It was yet another to convince people that they indeed understood what it was that these symbols represented. Given my daily interactions with the chimps, it was becoming increasingly difficult for me to understand what it was that others were questioning and why. I recall a dinner in Washington at this stage of the work, in 1978. Most people there were psychologists, interested in animal behavior and human development. They were intrigued by our descriptions of Sherman and Austin, and seemed willing to listen to the suggestion that the chimps’ behavior had gone beyond conditioned responses. “But,” one of them insisted, “how can you know what they know? How can you know Sherman and Austin are using symbols referentially?” It was a good question, and one I needed to address experimentally. I spent the following day thinking how I might do this.

  The issue of representation is fundamental for the field of language acquisition, and we had to demonstrate that lexigrams were indeed representational to Sherman and Austin. I evolved the idea of the categorization test. Until this point in their experience, Sherman and Austin had used specific lexigrams for specific objects; that is, they had learned no generic names. If I were to be able to demonstrate that the chimps understood that symbols represented objects in an abstract manner, I needed to introduce the idea of generic symbols. We would first teach the chimps “tool” and “food” as generic terms, using just three items in each. group for the training regime. Ultimately, I wanted to see if Sherman and Austin, on being shown the symbol for, say, banana, could correctly categorize it by the symbol “food.” The description of one symbol by another would, I thought, be strong evidence of the chimps
’ understanding of the referential nature of the symbols.

  We approached the task in three stages. After first learning the category symbols “food” and “tool,” the chimps would be shown food items and tools that had not been part of the training regime. For a successful test, Sherman and Austin would have to categorize correctly the new items on their first presentation. If a chimp had to be helped even once with one of the new items, that would count as training or imitation, and the test would be a failure. Our plan was then to repeat the process with plastic-covered photographs of food items and tools. Lastly, we would repeat it with symbols for food items and tools. This final test was the most critical of all, for it alone was unequivocally a test of the representational value of Austin and Sherman’s symbols: It required a novel response, one never before given by the chimpanzee, or even the experimenter when working with the chimpanzee.

  Apart from some difficulty Austin had in the second task in seeing clearly the food photographs within the plastic cover, both chimps scored successes on all three tasks. On the first trial of the final task, Sherman scored fifteen of sixteen symbols, and Austin seventeen of seventeen. For comparison, we performed these tests with Lana. Although she was able to sort food and tool items physically into groups, and therefore had some categorization skill, she was unable to do so using the generic symbols “food” and “tool.” This contrast between Lana and Sherman and Austin clearly shows that chimpanzees who have learned symbols have not necessarily learned the same thing. Because of their different experiences, Sherman and Austin had come to view symbols in a very different way from Lana.

  Food label plaques presented to Sherman and Austin. The chimps had not been taught to recognize these logos, but were always interested in the packages as well as the foods. We constructed these “food plaques” by taping the package wrappers to pieces of lexan plastic.

  The categorization study, I believed, showed as convincingly as anything can be demonstrated in behavioral research, that Sherman and Austin had developed a capacity of fundamental importance to language—the ability to use arbitrary symbols representationally. This, I felt, was sufficient justification for continuing ape-language research, despite the very negative atmosphere that had developed in the late 1970s and early 1980s. Such was the impact of Herbert Terrace’s 1979 Science paper that most people inferred that he had declared ape language a dead issue. In fact, he did leave open the possibility that another research group, using different methods, might make important progress. The achievements with Sherman and Austin represent that hoped-for progress. Unlike Nim, Terrace’s chimpanzee, Sherman and Austin did not merely imitate their teachers: They were economic and informative with their symbol production; they did take turns in communication, both with each other and with their teachers; and finally, although much of their symbol product was request oriented, they moved beyond this stage to spontaneous symbol production either to announce or comment on their actions.

  For example, one day while at the store, I noticed some unusually shaped colored plastic glasses. They looked like they would just fit in a young chimp’s grip, so I bought them for Sherman and Austin. The next morning I showed them to the chimps and Sherman in particular seemed immediately interested in these glasses. This was unusual for him, since he generally could not care less what he ate or drank from as long as he approved of the contents. Austin was the one who generally focused on containers, often playing with them by stacking them or by pouring liquids back and forth from one to the other. Austin, however, seem uninterested in the new glasses. To my astonishment, Sherman carried his all around the lab, treating it like a prized and cherished possession the entire morning. At nap time, I put the glasses up. Later that afternoon, I returned with a strawberry drink in a pitcher. Forgetting how much Sherman had liked the new glasses, I grabbed what was handy, which turned out to be one old glass and one new glass. When Sherman saw me, he rushed to the keyboard to say Glass strawberry drink and then hurriedly pointed to the new glass to show me that he wanted his strawberry drink in the new glass, not the old one.

  The important thing about this utterance was that neither he, nor I, nor anyone had ever used the symbol “glass” before. In fact, it was not an assigned symbol, just an extra lexigram on the keyboard that had not been assigned to anything. But Sherman was now assigning it to this favorite new drinking glass and pointing to the glass to let me know exactly what he meant and that he did not want his drink in the old glass. I glossed the lexigram as glass and began to use it for all glasses, but Sherman refused. He would use it only for these special glasses that he liked so much. Once they were broken, which does not take long in a chimp lab, I could never find another and finally had to take the lexigram off the keyboard because I could not get Sherman to use it for anything else. I have never understood why Sherman liked those glasses so much. I purchased many glasses later on, some even looked similar, but none of the others seemed to matter to him.

  Unfortunately, these unplanned demonstrations of cognitive capacity by Sherman and Austin were not considered “appropriate science.” They were relegated to the realm of the “anecdote” by most who elected to follow Terrace in evoking syntax as the sine qua non of mind.

  Herbert Terrace’s primary interest was, of course, in sentences and syntax. He addressed the issue of the meaning of symbols in the following way: “What is important to recognize … is that neither the symbols nor the relationships between the symbols have specific meaning. Although the words and word order may be meaningful to a human listener, they may be meaningless to the animal producing them.”3 To Sherman and Austin, their lexigrams demonstrably have meaning.

  How, then, do we answer our initial question: In what sense can a species other than Homo sapiens develop language? Sherman and Austin clearly have a language capability of a human sort, limited though it is in many ways. This conclusion is heresy to those linguists who would deny any kind of language capability to a species other than Homo sapiens. A second observation from the project is that the language capacity the chimps developed had to be cobbled together piecemeal; it did not emerge as a smooth developmental flow as it does in human children. This conclusion is heresy to linguistic theory in general. It also, I was to discover later, turns out to be wrong.

  4

  An Uncommon Ape

  The bonobo (also called the “pygmy” chimpanzee) is a most unusual species of animal. The last ape to be identified as a species, the bonobo was continually confused with the chimpanzee until relatively recently. In fact, the bonobo is closely related to the chimpanzee. Around five million years ago, an evolutionary split caused a now extinct ape species to divide into two lineages: one that eventually led to Homo sapiens, and a second that led to modern chimpanzees and bonobos.

  The chimpanzee is much more widely known, but it is the bonobo that is far more humanlike than other apes. In their anatomy, social behavior, vocalizations, sexual exploits, infant care, and mental abilities, bonobos possess an eerie human quality. Unfortunately, no one has yet tracked down the origin of the name “bonobo,” and consequently many scientists maintain that they should be called pygmy chimpanzees. But they are not chimpanzees of any size or shape. They are more like persons with small brains and extra-long body hair.

  In 1975 I began a Georgia State University postdoctoral fellowship at the Yerkes Center, where five bonobos were brought on lend-lease from Zaire. I therefore had the chance to observe the behavior of animals that had been wild born and reared. Observation of bonobos in the wild was still in its earliest stages, and so few were in captive exhibits that hardly anyone knew of their existence. Virtually nothing was understood about their behavior. Most zoological curators thought they were little more than geographical variants of the common chimp. Indeed, it was not uncommon for zoos to house them with common chimps or even gorillas.

  My interest in bonobos began when I read the observations of Robert Yerkes, who wrote the book Almost Human in 1925 about a young ape named “Pr
ince Chim.” We now know Prince Chim was a bonobo, but at the time the species had not yet been “discovered.” “Doubtless they are geniuses among the anthropoid apes,” Yerkes wrote, after studying Chim and comparing him with a common chimpanzee, Panzee. “Prince Chim seems to have been an intellectual genius.”1 Yerkes described Chim as sanguine, venturesome, trustful, friendly, and energetic, while Panzee was distrustful, retiring, and lethargic. Chim’s most human aspect, however, was the way he approached problems. “Most surprising and impressive in Chim’s behavior was the continuity of attention, high degree of concentration on task, evident purposefulness of many, if not most, of his acts, and his systematic survey of problematic situations, his rapid elimination of unsuccessful acts or methods, and his occasional pauses for reflection,” Yerkes wrote in a second book, Chimpanzee Intelligence. “I use this term [reflection] without apology, even to the behaviorist, for the simple reason that if Chim were a child instead of a chimpanzee we should apply the term without hesitation and with assurance that it would convey to every intelligent reader what is intended.”2