Kanzi: The Ape at the Brink of the Human Mind

by Sue Savage-Rumbaugh

  The amount of step fractures and battering seen on the stone. Step fractures are unclean breaks in the stone. When a stone is hit at exactly the proper angle, a clean flake falls off, leaving the stone surface as smooth as if someone had run a knife through butter. When a stone is flaked at the wrong angle, it breaks off the rock, leaving many jagged edges at the point of the break. If a stone is merely slammed into another hard surface, with little regard for the angle of the blow, it may break, but it will have a battered appearance, looking precisely as though it had been battered. Well-flaked stone looks as though it had been sculpted or chiseled.13

  Measured against these criteria, the products of the earliest tool-makers score highly. “It seems clear that early tool-making protohumans had a good intuitive sense of the fundamentals of working stone,” observes Nick.14 They knew about angles required on the core, about sharp, glancing blows, and about seeking regions of high mass on the core. They also apparently knew when otherwise suitably shaped cobbles would not flake well.

  For instance, Nick noticed that at Koobi Fora in northern Kenya, heavily weathered cobbles were common. Such cobbles flake unpredictably and inefficiently, and a knowledgeable stone knapper would avoid using them. The outward evidence of the weathering, however, is slight and appears merely as hairline fractures on the surface. When Nick examined the Oldowan assemblages at various archeological sites at Koobi Fora, he found only rare evidence of the use of weathered cobbles; their frequency at the sites was far less than their occurrence on the ancient landscape. “It seems clear that the early hominids had already learned to reject such inferior material,” concludes Nick.15

  By these measures, therefore, the Oldowan tool-makers displayed considerable technological sophistication and perceptual skills. What of Kanzi? His progress in hard-hammer percussion has been considerable, moving from the undirected, timid tapping of rocks together to the forceful hammering directed at the edge of the core. Nick describes the process of learning to make tools as being punctuational, with periods of slow change in between. “You suddenly get an insight into what is required, and then slowly improve on that,” he explains.16 Kanzi clearly had an insight into the importance of hitting the rock close to its edge; and he had important insights when he developed his throwing techniques. Despite this, however, he has not yet developed the stone-knapping skills of the Oldowan tool-makers. In a paper reporting the first eighteen months of the project, we described our assessment of Kanzi as follows:

  So far Kanzi has exhibited a relatively low degree of technological finesse in each of [the four criteria] compared to that seen in the Early Stone Age record. The amount of force he uses in hard-hammer percussion is normally less than ideal for fracturing these rocks. His flake angles when using hard-hammer percussion tend to be steep (approaching 90°), while Oldowan flakes were generally detached from more acute-edged cores (flake angles typically 75°-80°). As yet, Kanzi’s cores retain a very high proportion of their original cortex and are steep-edged and rather battered. The flakes he produces tend to be relatively small (generally less than 4 cm long) and often stepped or hinged, and his cores generally exhibit marginal (non-invasive) flake scars.17

  There is, therefore, a clear difference in the stone-knapping skills of Kanzi and the Oldowan tool-makers, which seems to imply that these early humans had indeed ceased to be apes. It isn’t yet certain, however, whether Kanzi’s poorer performance is the result of a cognitive or an anatomical limitation. Or simply lack of practice. Certainly most of us working with Kanzi are unable to make stone tools ourselves. Without a good teacher and constant practice it is a very difficult skill to acquire. Were an anthropologist to excavate our site a million years from now, I doubt that he or she could distinguish between the human stone artifacts and those produced by Kanzi—except, of course, those that were made by Nick. Making stone tools is not easy and does not seem to be a skill that normal human beings acquire readily with little instruction, as we are asking Kanzi to do.

  Nick hopes to learn whether or not Kanzi, with minimal demonstration, can acquire a skill that took, at best, many generations for our ancestors to perfect. If Kanzi does not succeed in matching the skills of Oldowan tool-makers in the span of one research career, it would still be foolish to rule out the potential of the ape mind to do so, given a few generations of exposure to need to use such tools.

  The structure of bonobos’ arms, wrists, and hands is different from that of humans, and this structure constrains the ability to deliver a sharp blow by snapping the wrist, a movement that Nick considers important in effective tool-making. I suspect that if Kanzi is limited in the quality of flaking through hard-hammer percussion, it is the result of biomechanical, not cognitive, constraints. His ape hands, with long, curved fingers and short thumbs, prevent him from gripping the stones efficiently enough to allow him to deliver a powerful, glancing blow.

  The greatest surprise of the tool-making project was Kanzi’s development of throwing as a way of obtaining sharp edges. Not only did it reflect a problem-solving process in Kanzi’s mind, but it also produced material that addresses an important archeological problem: What did early humans do before they made Oldowan tools? Application of the criteria mentioned earlier to identify genuine artifacts as compared with naturally fractured stone would reject Kanzi’s flakes and cores as tools. And yet they are artifacts, and they can be used as effective cutting tools.

  Some of Kanzi’s cores look rather similar to Oldowan core tools, acknowledges Nick, but most do not, because of the angles and preponderance of small flake scars. “If I were surveying a Stone Age site and found some of these things, I’d definitely check them out, but I would almost certainly conclude they were naturally flaked,” he says. But if pre-Oldowan tool-makers used Kanzi’s approach, and smashed rocks by throwing them, how could an archeologist know? “Most natural processes that break rocks also tend to smooth them,” explains Nick. “Rocks may fracture as they crash into each other while rolling along a stream bed, but the rolling process quickly dulls the edges.”

  Nick Toth’s tool is on the left; the one made by Kanzi is on the right. Kanzi has removed only part of the cortex, or soft material on the outside of the flint. The chopperlike shape has not been produced by design, but results from simple attempts to remove enough of the cortex to get a sharp edge. Nick’s tool, by contrast, is made with a specific shape in mind and many blows have been deliberately struck to produce a tool with the shape seen here. Kanzi recognizes the value of a tool such as that made by Nick, and when given a choice between a tool he has made and a handaxe made by Nick, Kanzi chooses Nick’s tools over his own, without hesitation. Occasionally, when I have left a new handaxe of Nick’s in Kanzi’s enclosure, I have observed him rubbing his fingers across it while looking at it very carefully. Of course I cannot know for certain, but the impression he leaves is that he is admiring the workmanship in Nick’s tool.

  Finding a sharp-edged cobble with angles of close to 90° might therefore be indicative of primitive tool-making. “You have to examine the context of the rock, to eliminate natural processes that might have produced sharp edges. But after seeing these incipient flaking skills with Kanzi, we certainly have to consider it as a possible model for the earliest stone tool-making. He has taught us what we should be looking for, to find tool-makers earlier than what we usually call ‘the earliest tool-makers.’”18

  When I agreed to participate in the Wenner-Gren conference I had no idea a collaboration such as the one with Nick and Kathy would arise. Both sides—in psychology and archeology—benefited tremendously. And Kanzi became the first nonhuman to learn humanlike stone tool-making in a natural setting. Nick joked that Kanzi should be awarded an honorary doctorate, pointing out that he would need a small cap and a gown with long arms. He wasn’t joking, however, when in the spring of 1991, Kanzi was awarded the inaugural CRAFT Annual Award for Outstanding Research Pertaining to Human Technological Origins. Nick and Kathy are co-directors of CRAFT, or the Center
for Research into the Anthropological Foundations of Technology, at Indiana University. “The award is justified, because the work with Kanzi has given us one of our most important insights into paleolithic technology,” says Nick. “It has given us a view of what is possible with apes, and an insight into the cognitive background of what is necessary to go further.”

  9

  The Origin of Language

  According to the evidence of molecular biology, the first hominid species appeared approximately five million years ago, a bipedal ape with long arms and curved fingers that presumably was well at home in the trees. The earliest known fossil evidence of such a creature dates from three to four million years ago, and was found in Ethiopia. These early hominids, with their 400-cubic-centimeter brains, and several later small-brained species, all belonged to the genus Australopithecus. Only when the genus Homo appeared did brain size begin to increase, leaping by 50 percent in Homo habilis, to more than 600 cubic centimeters. The next player on the stage of protoman was Homo erectus, who debuted almost two million years ago. His brain size varied from 850 to 1100 cubic centimeters. Modern levels of brain size—1350 cubic centimeters—came with the evolution of archaic Homo sapiens, probably around two hundred and fifty thousand years ago.

  Much of the increase in brain size from the ape level to the modern human level can be accounted for by an enlargement of the neocortex, the thin coat of nerve cells that forms the outer layer of our brain. This outer covering did not expand with a common equipotential over the entire surface, however; the frontal lobes, associated with planning and foresight, expanded disproportionately. Another part of the brain, located in the lower rear part of the skull and termed the cerebellum, has also expanded disproportionately in man. This area is associated with the automatization of skills such as driving a car, riding a bike, buttoning a shirt, and so on. This human expansion pattern was absent in the australopithecine species, but it appeared in Homo habilis—or handy man—the first stone tool-maker.

  It was once believed that Broca’s area, long thought to be the area of the brain that made language possible, was unique to humans. Located (usually) in the left frontal lobe near the temple, Broca’s area in humans is readily identified as a raised region. When evidence for Broca’s area was discovered in the cranium of a 1.8 million-year-old Homo habilis from northern Kenya, two decades ago, it was taken as an indication of an advanced language faculty. However, we now know that Broca’s area occurs in the brains of other animals, too, and is merely expanded in humans, not unique. Similarly, none of the other brain centers involved in language comprehension and production, including Wernicke’s area (located in the left parietal lobe) and a scattering of some dozen or so nuclei throughout the prefrontal region, represent novel structures. The difference between ape and human brains is essentially quantitative. Those who argue for a uniquely human language-acquisition device, as do proponents of the Chomskian school, do so in the absence of any anatomical evidence for its existence.

  The brain, metabolically speaking, is an extremely expensive organ. It represents 2 percent of body mass but consumes 18 percent of our energy budget. There seems little reason to have such a large brain unless it somehow greatly increases the survival prospects of its bearer. Based on the nature of size increase and reorganization seen in fossil brains, anthropologist Dean Falk, along with many others at the Wenner-Gren conference, believe that it probably was language that propelled the increase in brain size. As Dean expressed it, “If hominids weren’t using and refining language, I would like to know what they were doing with their autocatalytically increasing brains.”1

  Terrence Deacon, a neurologist at Belmont Hospital, Massachusetts, is equally emphatic, but from the perspective of modern brains, not fossil ones: Deacon bases his conclusion on a study of the nature of differences in connectivity in ape and human brains, and on developmental studies of monkey brains. “The brain structures and circuits most altered in the course of human brain evolution reflect some unusual computational demands by natural languages,” he notes.2 These alterations center on the increasing dominance of output from the prefrontal region, which allows voluntary control over vocalizations.

  The modern brain appeared with the first members of a group that is loosely called archaic Homo sapiens, which evolved some quarter of a million years ago. Were these people as linguistically sophisticated as we are today? It’s hard to say, but if the brain-size/linguistic-capacity relationship holds, as has been argued, then the answer should be yes, for these people had brains the size of our own. They differed from us only in that they retained a physical robustness that is absent from modern skeletons.

  The second line of anatomical evidence—that of the vocal apparatus—tells very much the same story as the one we see with brain-size increase. The vocal apparatus consists of the larynx (or vocal organ), the pharynx (or throat), the nasopharynx (or nasal cavity), the tongue, and the lips. In all mammals apart from humans, the larynx is positioned high in the neck, a position that has three consequences. First, the larynx can be “locked into” the nasopharynx—the air space near the “back door” of the nasal cavity. When this occurs, all breathing is done through the nose, as the back of the oral cavity is closed by the overlapping of the soft palate and the epiglottis.

  Second, although the vocal tracts of chimpanzees and other mammals can produce most of the human vowel sounds, it is difficult for them to make some of the sounds readily. Edmund S. Crelin, an anatomist at the Yale University School of Medicine, has done extensive modeling of both the human and the ape vocal tracts, including the construction of manipulable rubber casts, which permit him to determine the sounds that can be produced by the physical structure of the organism. Crelin has studied this problem in detail, comparing the anatomical capacities not only of apes and humans, but also of many hominids. He has done so by reconstructing the vocal tract tissue on the basis of the available skeletal material. In order to investigate the range of sounds that a chimpanzee can make, Crelin built a rubber model of the chimpanzee vocal tract and forced pressurized air through the model as he manually manipulated its shape. These experiments led him to conclude that he could “force a rubber tract of many nonhuman mammals to produce a set of vowel-like sounds, including those of mammals with even longer snouts, such as a horse.”3 Nonetheless, he found that it required extreme constriction of the model ape vocal tract to produce the long e and long u sounds, and that it was also nearly impossible for the chimpanzee to switch rapidly between vowel sounds.

  Finally, the range of noises apes can make does not include the most important element of human speech—the consonant. This is because they have difficulty accomplishing what is called velopharyngeal closure, or the brief blocking off of the nasal passages as air is forced through the mouth. This blockage is needed for the production of consonants; it enables us to generate the brief turbulence and temporary microbursts of air that are the basis of consonants. The action of the vocal cords lays noise over these temporary perturbations and the shape of the vocal tract itself is modulated to amplify or decrease certain frequencies, thereby serving to filter the action of the basic sound produced by the vibration of the vocal folds. These filtered sounds, without turbulence, become vowels; with turbulence, generated by velopharyngeal closure or a sealing off of the nasal cavity by raising the soft palate in the back of the throat very rapidly, we are able to produce interpretable speech.

  Man alone has a vocal tract that permits the production of consonant sounds. These differences between our vocal tract and that of apes, while relatively minor, are significant and may be linked to the refinement of bipedal posture and the associated need to carry the head in a balanced, erect position over the center of the spine. A head with a large heavy jaw would cause its bearer to walk with a forward list and would inhibit rapid running. To achieve balanced upright posture, it was essential that the jaw structure recede and thus that the sloped vocal tract characteristic of apes become bent at a right angle. A
long with the reduction of the jaw and the flattening of the face, the tongue, instead of residing entirely in the mouth, was lowered partially down into the throat to form the back of the oropharynx. The mobility of the tongue permits modulation of the oropharyngeal cavity in a manner that is not possible in the ape, whose tongue resides entirely in the mouth. Similarly, the sharp bend in the supralaryngeal airway means that the distance between the soft palate and the back of the throat is very small. By raising the soft palate, we can block off the nasal passageways, permitting us to form the turbulence necessary to create consonants.

  An obvious question follows from my argument that the evolution of a bipedal mode of locomotion in our ancestors was important in the development of a vocal tract capable of producing consonants: Why did Australopithecus not follow the same evolutionary path as Homo in developing a humanlike larynx? I’ve referred to all species in the human family, including Australopithecus, as bipedal apes. This is true, in the sense of the very close genetic relationship humans have with apes. But it may be a bit misleading with respect to how efficient the different hominid species were in their bipedal locomotion.

  There has been a long-running debate among anthropologists over this question, with some arguing that the australopithecine species walked just as modern humans do, that is, with a fully upright, striding gait. Others disagree, saying that the australopithecines retained many apelike adaptations, including spending a significant amount of their time in the trees and having a more shambling gait while on the ground. I support this latter argument, primarily because the australopithecines had long arms, short legs, and curved bones in their hands and feet, just as apes do. It is true that they were adapted to a degree of bipedal locomotion, but they were not fully bipedal as species of Homo have been, right from their first appearance two and a half million years ago. My argument over the effect of posture on the vocal tract refers to full bipedalism, not the incomplete form that prevailed in Australopithecus,