Primates and Philosophers_How Morality Evolved

by Frans de Waal

Once, finding myself in a similar situation with Georgia (i.e., aware that she had gone to the spigot and was sneaking up on me), I looked her straight in the eyes and pointed my finger at her warning, in Dutch, “I have seen you!” She immediately stepped away and let part of the water drop, swallowing the rest. I certainly do not wish to claim that she understands Dutch, but she must have sensed that I knew what she was up to, and that I was not going to be an easy target.

  Figure 8 Georgia, our naughtiest chimpanzee, fascinated by her own reflection in the camera lens. Photograph by the author.

  The curious situation in which scientists who work with these fascinating animals find themselves is that they cannot help but interpret many of their actions in human terms, which then automatically provokes the wrath of philosophers and other scientists, many of whom work with domestic rats, or pigeons, or with no animals at all. Unable to speak from firsthand experience, these critics must feel confident indeed when they discard accounts by primatologists as anthropomorphie, and explain how anthropomorphism is to be avoided.

  Although no reports of spontaneous ambush tactics in rats have come to my attention, these animals could conceivably be trained with patient reinforcement to retain water in their mouth and stand amongst other rats. And if rats can learn to do so, what is the big deal? The message of the critics of anthropomorphism is something along the lines of “Georgia has no plan; Georgia does not know that she is tricking people; Georgia just learns things faster than a rat.” Thus, instead of seeking the origin of Georgia’s actions within herself, and attributing intentions to her, they propose to seek the origin in the environment and the way it shapes behavior. Rather than being the designer of her own disagreeable greeting ceremony, this ape fell victim to the irresistible rewards of human surprise and annoyance. Georgia is innocent!

  But why let her off the hook that easily? Why would any human being who acts this way be scolded, arrested, or held accountable, whereas any animal, even of a species that resembles us so closely, is considered a mere passive instrument of stimulus-response contingencies? Inasmuch as the absence of intentionality is as difficult to prove as its presence, and inasmuch as no one has ever proven that animals differ fundamentally from people in this regard, it is hard to see the scientific basis for such contrasting assumptions. Surely, the origin of this dualism is to be found partly outside of science.

  The dilemma faced by behavioral science today can be summarized as a choice between cognitive and evolutionary parsimony (de Waal 1991, 1999). Cognitive parsimony is the traditional canon of American Behaviorism. It tells us not to invoke higher mental capacities if we can explain a phenomenon with ones lower on the scale. This favors a simple explanation, such as conditioned behavior, over a more complex one, such as intentional deception. This sounds fair enough (but see Sober 1990). Evolutionary parsimony, on the other hand, considers shared phylogeny. It posits that if closely related species act the same, the underlying mental processes are probably the same, too. The alternative would be to assume the evolution of divergent processes that produce similar behavior, which seems a wildly uneconomic assumption for organisms with only a few million years of separate evolution. If we normally do not propose different causes for the same behavior in, say, dogs and wolves, why should we do so for humans and chimpanzees?

  In short, the cherished principle of parsimony has taken on two faces. At the same time that we are supposed to favor low-level over high-level cognitive explanations, we also should not create a double standard according to which shared human and ape behavior is explained differently. If accounts of human behavior commonly invoke complex cognitive abilities—and they most certainly do (Michel 1991)—we must carefully consider whether these abilities are perhaps also present in apes. We do not need to jump to conclusions, but the possibility should at least be allowed on the table.

  Even if the need for this intellectual breathing room is most urgently felt in relation to our primate relatives, it is neither limited to this taxonomic group nor to instances of complex cognition. Students of animal behavior are faced with a choice between classifying animals as automatons or granting them volition and information-processing capacities. Whereas one school warns against assuming things we cannot prove, another school warns against leaving out what may be there: even insects and fish come across to the human observer as internally driven, seeking, wanting systems with awareness of their surroundings. Descriptions that place animals closer to us than to machines adopt a language that we customarily use for human action. Inevitably, these descriptions sound anthropomorphic.

  Obviously, if anthropomorphism is defined as the misattribution of human qualities to animals, no one wishes to be associated with it. But much of the time, a broader definition is employed, namely the description of animal behavior in human, hence intentionalistic, terms. Even though no anthropomorphism proponent would propose to apply such language uncritically, even the staunchest opponents of anthropomorphism do not deny its value as a heuristic tool. It is this use of anthropomorphism as a means to get at the truth, rather than as an end in itself, that distinguishes its use in science from that by the layperson. The ultimate goal of the anthropomorphizing scientist is emphatically not the most satisfactory projection of human feelings onto the animal, but testable ideas and replicable observations.

  This requires great familiarity with the natural history and special traits of the species under investigation, and an effort to suppress the questionable assumption that animals feel and think like us. Someone who cannot imagine that ants taste good cannot successfully anthropomorphize the anteater. So, in order to have any heuristic value at all, our language must respect the peculiarities of a species while framing them in a way that strikes a chord in the human experience. Again, this is easier to achieve with animals close to us than with animals, such as dolphins or bats, that move through a different medium or perceive the world through different sensory systems. Appreciation of the diversity of Umwelten (von Uexkłl 1909) in the animal kingdom remains one of the major challenges of the student of animal behavior.

  The debate about the use and abuse of anthropomorphism, which for years was confined to a small academic circle, has recently been thrust into the spotlight by two books: Kennedy’s (1992) The New Anthropomorphism, and Marshall Thomas’s (1993) The Hidden Life of Dogs. Kennedy reiterates the dangers and pitfalls of assuming higher cognitive capacities than can be proven, thus defending cognitive parsimony. Marshall Thomas, on the other hand, does not make any bones about the anthropomorphic bias of her informal study of canine behavior. In her best-seller, the anthropologist lets virgin bitches “save” their virginity for future “husbands” (i.e., ignore sexual attentions prior to meeting a favorite male, p. 56), watches wolves set out for the hunt without “pitying themselves” (p. 39), and looks into her dogs’ eyes during a vicious gang attack seeing “no anger, no fear, no threat, no show of aggression, just clarity and overwhelming determination” (p. 68).

  There is quite a difference between the use of anthropomorphism for communicatory purposes or in order to generate hypotheses, and the sort of anthropomorphism that does little else than project human emotions and intentions onto animals without justification, explication, or investigation (Mitchell et al. 1997). The uncritical anthropomorphism of Marshall Thomas is precisely what has given the practice a bad name, and has led critics to oppose it in all of its forms and disguises. Rather than let them throw out the baby with the bathwater, however, the only question that needs to be answered is whether a certain dose of anthropomorphism, used in a critical fashion, helps or hurts the study of animal behavior. Is it something that, as Hebb (1946) already noted, allows us to make sense of animal behavior, and, as Cheney and Seyfarth (1990: 303) declared, “works” in that it increases the predictability of behavior? Or is it something that, as Kennedy (1992) and others argue, needs to be brought under control, almost like a disease, because it makes animals into humans?

  While it is true that animals
are not humans, it is equally true that humans are animals. Resistance to this simple yet undeniable truth is what underlies the resistance to anthropomorphism. I have characterized this resistance as anthropodenial, the a priori rejection of shared characteristics between humans and animals. Anthropodenial denotes willful blindness to the human-like characteristics of animals, or the animal-like characteristics of ourselves (de Waal 1999). It reflects a pre-Darwinian antipathy to the profound similarities between human and animal behavior (e.g., maternal care, sexual behavior, power seeking) noticed by anyone with an open mind.

  The idea that these similarities require unitary explanations is anything but new. One of the first to advocate cross-specific explanatory uniformity was David Hume (1985 [1739]: 226), who formulated the following touchstone in A Treatise of Human Nature:

  Tis from the resemblance of the external actions of animals to those we ourselves perform, that we judge their internal likewise to resemble ours; and the same principle of reasoning, carry’d one step farther, will make us conclude that since our internal actions resemble each other, the causes, from which they are deriv’d, must also be resembling. When any hypothesis, therefore, is advanc’d to explain a mental operation, which is common to men and beasts, we must apply the same hypothesis to both.

  It is important to add that, in contrast to American behaviorists, who two centuries after Hume would accommodate animals and humans within a single framework by seriously downgrading human mental complexity and relegating consciousness to the domain of superstition (e.g., Watson 1930), Hume (1985 [1739]: 226) held animals in high esteem, writing that “no truth appears to me more evident than that beasts are endow’d with thought and reason as well as men.”

  Strictly speaking, one cannot boast a unified theory of all behavior, human and animal, while at the same time decrying anthropomorphism. After all, anthropomorphism assumes similar experiences in humans and animals, which is exactly what one would expect in case of shared underlying processes. The behaviorists’ opposition to anthropomorphism probably came about because no sane person would take seriously their claim that internal mental operations in our species are a figment of the imagination. The masses refused to accept that their own behavior could be explained without considering thoughts, feelings, and intentions. Don’t we have mental lives, don’t we look into the future, aren’t we rational beings? Eventually, the behaviorists relented, exempting the bipedal ape from their theory of everything.

  This is where the problem for other animals began. Once cognitive complexity was admitted in humans, the rest of the animal kingdom became the sole light-bearer of Behaviorism. Animals were expected to follow the law of effect to the absolute letter, and anyone who thought differently was just being anthropomorphic. Attribution of human-like experiences to animals was declared a cardinal sin. From a unified science, Behaviorism had deteriorated into a dichotomous one with two separate languages: one for human behavior, another one for animal behavior.

  So, the answer to the question “Isn’t anthropomorphism dangerous?” is that, yes, it is dangerous to those who wish to uphold a wall between humans and other animals. It places all animals, including humans, on the same explanatory plane. It is hardly dangerous, though, to those working from an evolutionarily perspective so long as they treat anthropomorphic explanations as hypotheses (Burghardt 1985). Anthropomorphism is a possibility among many, but one to be taken seriously given that it applies intuitions about ourselves to creatures very much like us. It is the application of human self-knowledge to animal behavior. What could be wrong with that? We apply human intuition in mathematics and chemistry, so why should we suppress it in the study of animal behavior? Stronger yet: does anyone truly believe that anthropomorphism is avoidable (Cenami Spada 1997)?

  In the end we must ask: What kind of risk are we willing to take, the risk of underestimating animal mental life or the risk of overestimating it? There is a symmetry between anthropomorphism and anthropodenial, and since each has its strengths and weaknesses, there is no simple answer. But from an evolutionary perspective, Georgia’s mischief is most parsimoniously explained in the same way we explain our own behavior—as the result of a complex, and familiar, inner life.

  Appendix B

  Do Apes Have a Theory of Mind?

  Intersubjectivity research on primates began with Menzel (1974), who released young chimpanzees into a large outdoor enclosure with only one of them aware of the location where food (or a toy snake) was hidden, while his or her companions possessed no such knowledge. The companions seemed perfectly capable, however, of “guessing” what to expect based on the behavior of the “knower.” Menzel’s classic experiment, combined with Humphrey’s (1978) notion of animals as “natural psychologists” and Premack and Woodruff’s (1978) “theory of mind” (ToM), inspired the guesser-versus-knower paradigm still popular today in intersubjectivity research on both apes and children.

  ToM refers to the ability to recognize the mental states of others. If you and I meet at a party, and I believe that you believe that we have never met before (even though we did meet), I have a theory about what is going on in your head. Given that some scientists claim this ability to be uniquely human, it is ironic that the whole ToM concept originated with primate research. It has had serious ups and downs since. Some concluded from failed demonstrations that apes must lack ToM (e.g., Tomasello 1999; Povinelli 2000). Negative results are impossible to interpret, though. As the saying goes: Absence of evidence is no evidence of absence. An experiment may not work for reasons that have nothing to do with the capacity in question. When comparing apes and children, for example, one problem is that the experimenter is invariably human, so that only the apes face a species barrier (de Waal 1996).

  To captive apes, we must come across as all-mighty and all-knowing. We approach chimpanzees in our care while having heard the latest news about them from others (e.g., we have been called on the telephone about a new injury or birth). The chimpanzees must notice that we often know things before we have seen them. This makes humans inherently unsuitable as participants in experiments on the role of seeing in knowing, a centerpiece of ToM research.

  All that most experiments have done thus far is test the ape’s theory of the human mind. We would do better to focus on the ape’s theory of the ape mind. When the human experimenter is cut out of the picture, chimpanzees seem to realize that if another has seen hidden food, this individual knows (Hare et al. 2001). This finding, along with growing evidence for visual perspective-taking in apes (Shillito et al. 2005; Bräuer et al.; 2005; Hare et al., in press; Hirata 2006), has thrown the question of animal ToM wide open again. In an unexpected twist (because the debate revolves around humans and apes), a capuchin monkey at the University of Kyoto recently passed a number of seeing-knowing tests with flying colors (Kuroshima et al. 2003). A few such positive outcomes suffice to question all previous negative ones.

  The only way to get to the bottom of ape intelligence is to design experiments that engage them intellectually and emotionally. Rescuing an infant from an attack, outmaneuvering a rival, avoiding conflict with the dominant male, sneaking away with a mate, are the kinds of problems apes are good at solving. There are many reports suggesting ToM in ape social life, and even though these are usually single events— sometimes disparagingly labeled “anecdotes”—I consider them extremely meaningful. After all, one step by one man on the moon has been sufficient for us to claim that going there is within our ability. If an experienced, reliable observer reports a remarkable incident, science had better pay close attention. At the very least, such reports can be of heuristic value (de Waal 1991). With regard to apes taking another’s perspective, we do not just have a few stories, but a great many of them. In the lead text I have related the stories of Kuni and the bird and Jakie and his aunt. Let me offer two more examples from de Waal (1989a).

  The two-meter-deep moat in front of the old bonobo enclosure at the San Diego Zoo had been drained for cleaning. After ha
ving scrubbed the moat and released the apes, the keepers went to turn on the valve to refill it with water when all of a sudden the old male, Kakowet, came to their window, screaming and frantically waving his arms so as to catch their attention. After so many years, he was familiar with the cleaning routine. As it turned out, several young bonobos had entered the dry moat but were unable to get out. The keepers provided a ladder. All bonobos got out except for the smallest one, who was pulled up by Kakowet himself.

  This story matches another moat observation made at the same enclosure a decade later. By this time, the zoo had wisely decided against water in the moat as apes cannot swim. A chain hung permanently down into it, and the bonobos visited the moat whenever they wanted. If the alpha male, Vernon, disappeared into the moat, however, a younger male, Kalind, sometimes quickly pulled up the chain. He would then look down at Vernon with an open-mouthed play face while slapping the side of the moat. This expression is the equivalent of human laughter: Kalind was making fun of the boss. On several occasions, the only other adult, Loretta, rushed to the scene to rescue her mate by dropping the chain back down, and standing guard until he had gotten out.

  Both observations tell us something about perspective-taking. Kakowet seemed to realize that filling the moat while the juveniles were still in it wouldn’t be a good idea even though this would obviously not have affected him. Both Kalind and Loretta seemed to know what purpose the chain served for someone at the bottom of the moat and to act accordingly, the one by teasing, the other by assisting the dependent party.

  I am personally convinced that apes take one another’s perspective, and that the evolutionary origin of this ability is not to be sought in social competition, even if it is readily applied in this domain (Hare and Tomasello 2004), but in the need for cooperation. At the core of perspective-taking is emotional linkage between individuals—widespread in social mammals—upon which evolution (or development) builds ever more complex manifestations, including appraisal of another’s knowledge and intentions (de Waal 2003).