CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE

preference for stones is reasonable Tool Use in Wild : from the standpoint of energy use. New Light From Dark Forests Nut cracking is restricted to some populations in some Hedwige Boesch-Achermann and Christophe Boesch parts of the West African rain forests, although the same tree species and tools exist at other chimpanzee study sites in Central African forests. In the early 1960s, direct obser- aspects of the nut-cracking behavior This fact has led several authors to vations of wild chimpanzees began observed in our study community in propose that nut-cracking behavior in the wooded savanna of East Af- the tropical rain forest.2 is cultural in chimpanzees because rica. This work revealed the surpris- ecological differences seem not to ing abilities of chimpanzees in using be responsible for its distribution. and making tools as well as in hunt- In contrast to other ing for meat.1 However, although NUT-CRACKING BEHAVIOR w e l l - documented types of tool use, the majority of chimpanzees live in a such as termite fishing and ant forest environment, it was not until dipping, which are acquired at about 1979, when we started our project Five species of nuts with different age 4, nut-cracking techniques are in the , that chimpanzees shapes and degrees of hardness are fully acquired only at adulthood (13 living in the tropical rain forest were eaten by chimpanzees in Tai: Coula to 15 years in chimpanzees). Why is observed on a long-term basis. edulis nuts are very abundant and this behavior so demanding? Habituation to human observers is the softest of all; nuts of the Panda Consider the difficulty of the task: a long process, for chimpanzees are oleosa are the hardest; Parinari First, it is the only tool use by very shy. Earlier studies relied on ar- excelsa and Detarium senegalense chimpanzees requiring three tificial provisioning to speed up this are both very large trees with different objects to be assembled process. We did not because a feed- irregular fruit production; and before starting the task: an anvil, a ing behavior was the main topic of Sacoglottis gabonensis nuts are hammer, and the nuts. As the three our study. The Tai chimpanzees are cracked only rarely at our site. This are never found together in the forest, today still the only community fully report deals with Coula edulis and transporting hammers and nuts to habituated without relying on artifi- Panda oleosa. These nuts are rich in the anvil is the rule. Second, the cial provisioning. protein, sugar, fat, and amino acids. chimpanzee does not see any result Tool use is rare in animals, and The anvils used by the chimpan- from hitting a nut until it opens. the chimpanzees stand out as the zees to secure the nuts include Third, this is the only known tool use most proficient tool users besides emerging roots, the bases of large in chimpanzees in which two types humans. Toolmaking is even more trees, suitable branches in trees, and of tools are often used to gain access rarely observed, and chimpanzees rocks. Most hammers used to pound to the same food (i.e., a stone ham- are the only animal species observed the nuts are fallen branches of vari- mer and a stick; see Fig. 1). to make tools regularly in the wild. ous shapes, sizes, and degrees of hardness. Tool selection is related to Knowledge of tool use in chimpan- Transport of Tools and Nuts zees leads to better understanding of the hardness of the nut, and chim- what is unique about humans' tool panzees select harder and heavier use. Here, we review some of the tools for the harder nuts. Stones are Some 12 nuts are collected at a rare in the forest and highly sought time in the tree or on the ground and after because a desirable nut spe- carried to an anvil for cracking. Per Christophe Boesch is presently As- cies, Panda oleosa, is so hard that it day, chimpanzees crack on average sistant Professor at the Institut of is opened only with a stone. It is about 270 nuts in 2 hr; this activity Zoology, , Swit- harder than any food processed by involves more than 20 transports of zerland. He and his wife, Hedwige human hunter-gatherer societies. nuts. Similarly, tools are transported Boesch-Achermann, have studied Physical experiments revealed that between anvils during cracking ses- the chimpanzees of Tai National using a stone rather than wooden sions when chimpanzees move be- Park in Cote d'lvoire, West Africa, hammer has an energy advantage, tween different nut trees. They trans- since 1979. Address correspon- reducing pounding energy by 30% port their hammers more frequently dence to Christophe Boesch, Insti- for Coula nuts. This gain increases to and over longer distances for harder tut of Zoology, Rheinsprung 9, 43% when cracking the harder nuts than for softer ones because of 4051 Basel, . 2 Panda nuts, so the chimpanzees' the relative rarity of the stones needed to open the harder nuts.

18 Copyright © 1993 American Psychological Society CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE

ented in all directions in the forest, the chimpanzees must mentally re- orient these distances before com- paring them. The target tree and the positions of the stones change regu- larly, and these calculations have to be redone each time.3 Thus, in planning their hammer transports, Tai chimpanzees use a mental representation of space that allows them to conserve distances between objects (hammers and trees) at least over several days, to compare several of these distances, to permute objects (hammers) in this map, and to permute the point of reference (nut-bearing trees) to which distances will be estimated.

■ ' ■■* ■ "' In Piagetian terms, the simultaneous presence of these operations com- Fig. 1. Chimpanzees cracking nuts. An adult female is cracking nuts using a 7.6-kg pares with the euclidean mental stone hammer. Her 5-year-old youngster is watching the adult male getting kernel map observed for spatial representa- remains out of shells with the help of a small stick he prepared before use. tion abilities in 9-year-old children.3 The selective pressure to survive in the tropical rain forest seems to favor For nuts of Coula edulis, clubs are species are not very common, chim- the development of spatial abilities regularly transported for short dis- panzees have to transport the stones that are not required in more seden- tances on the ground, 20 m or less in over long distances (average dis- tary conditions (e.g., chimpanzees 85% of the cases; stones are moved tance = 120 m), a special challenge in captivity and even some human up to 50 m. In addition, hammers for one can see 25 m at most in the populations).4,'5 are always transported when nuts forest. These facts lead to some ob- are cracked in the tree before they vious questions: How do chimpan- fall on the ground, which implies an zees select their hammers when they Sex Differences in Frequency and anticipation process (i.e., thinking of cannot see more than one at a time? Performance of Cracking Behaviors the hammer before climbing the And what cognitive capacities are tree). Cracking in the tree requires required? We distinguish here three crack- particular dexterity. The chimpan- To arrive at some answers, we ing techniques: Coula cracking on zee must be able to transport the marked all the fruiting Panda trees in the ground, Coula cracking directly nuts and the tool and to handle them a given region of 30 ha, measured in the tree, and cracking the very so that nothing falls to the ground. the distances between them, marked hard Panda nuts. The latter two tech- This feat is especially impressive and weighed all the available niques are particularly demanding when a mother, supporting her baby stones, and kept a precise record of feats. They require transporting on her belly, holding spare nuts in their locations. Analysis of 76 trans- tools, and the physical peculiarities one foot and in her mouth, supports ports showed that chimpanzees of the Panda nut demand special the nut on the anvil with one hand cracking at a tree with no available technical skill. The Panda nut has while hitting it with the other. Given hammer would choose, within a three seeds independently embed- that most transports of Coula nuts given class of weight, the nearest ded in a hard wooden shell. Very are to a place visible from the origi- stone to that tree (average transport powerful hits must be made at the nal location, the decision to make a distance = 120 m, weight class = beginning to open the nut and get at transport can be explained by a sim- 1-12 kg). To make this selection, the first seed. Then the nut has to be ple association process. they have to compare the distances precisely repositioned two or three For Panda nuts, the situation is to the target tree for several possible times while hitting gently to extract more complicated because stones, stones (five on average within 300 the remaining seeds without smash- necessary to crack the hard nuts, are m). In addition, to be able to men- ing them. rare in the forest. As the trees of this tally compare distances that are ori-

VOLUME 2, NUMBER 1, FEBRUARY 1993

Both techniques are used more the nuts they crack to their offspring. get access to a highly important food frequently by females than males, During this time, the infants start source, for this independence en- and the females' performance (num- handling the tools and the nuts and ables the mothers to invest fully in ber of hits used to open the nut and attempt to crack nuts, usually with new offspring. number of nuts eaten per minute) is an incomplete setting (e.g., by Recent critical reviews of animal superior.6 The reason for this sex dif- pounding a nut lying on the ground learning processes have denied that ference is social in that males favor rather than on an anvil, hitting on an animals have the ability to imitate, social contact whenever there is a empty anvil with a hammer, or hold- but the teaching instances we ob- conflict between cracking more nuts ing a nut in the hand and pounding served would have no functional or remaining with other group mem- it on the anvil or the ground). role if the chimpanzees did not have bers. These two particular tech- Later, when the offspring are an imitative capacity. Many people niques are relatively solitary activi- around 4 to 5 years of age (normally still consider pedagogy one of the ties, one taking place high up in the before a new baby is born), the uniquely human attributes; our ob- trees, usually out of sight of other mothers start to share less but while servations of chimpanzees indicate chimpanzees, the other being re- collecting nuts leave behind, on an- otherwise. stricted to one or two animals be- vils, intact nuts and their own good cause of the lack of stones.7 We may tools. No chimpanzee without speculate that humans may already young offspring ever does so. Usu- CONCLUSION have had such a sex difference at a ally the youngsters use this opportu- very early stage. The first tools, nity and start using the tools, learn to non-lithic ones, were possibly used recognize an optimal tool, and Our observations of nut-cracking for gathering and food-processing sooner or later succeed in opening a behavior in Tai chimpanzees have activities, such as cracking nuts. The nut. This is the beginning of the true revealed surprisingly high cognitive first users and makers of these apprenticeship—when the offspring capacities that were not suspected tools might well have been females. start assembling the three necessary from previous studies. Studies of This hypothesis is in contrast with objects for this activity and use them cognitive abilities done with captive the common theory that tools were correctly. chimpanzees who had received lan- invented for hunting purposes.8 It is during this period that we ob- guage training led some authors to served the very rare true cases of ac- state that the language training al- tive teaching in a wild animal.9 We lowed the animals to acquire abili- Ontogeny, Food Sharing, saw two mothers correct errors in ties that would be somehow dor- and Teaching their infants' nut-cracking technique mant and never expressed in the by demonstrating the right method. wild.5'11 Such a suggestion sounds The nut-cracking behavior of the In both cases, only the infants were strange to a biologist, for what Tai chimpanzees may be the most cracking nuts, and the mothers, who mechanism could favor the evolu- sophisticated tool behavior observed were resting nearby, saw the prob- tion of unused abilities in an animal so far in wild chimpanzees, says lem and intervened just for the time species? Contrary to this hypothesis, Jane Goodall. How do youngsters necessary to correct, in one case, there is evidence that chimpanzees acquire this demanding behavior, the hammer grip and, in the other, in the wild have more developed and in what aspects do mothers in- the position of the nut. In the first cognitive abilities than chimpanzees fluence their apprenticeship? case, the infant obviously learned in captivity. First, considerable evi- Sharing nuts is the dominant ele- from her mother's demonstration, dence from psychological studies ment in this apprenticeship, occur- and she then succeeded in opening shows that an individual who devel- ring for 6 years between a mother nuts four times better without repeat- ops in socially or physically de- and her infant. (In contrast, Gombe ing the mistake corrected by her prived conditions is impaired in its chimpanzee mothers never share mother. We think it is not surprising cognitive development (e.g., as that teaching happened in the shown in Mason's experiments with tool-acquired food with their off- 12 spring.) The sharing pattern is differ- nut-cracking context. Active surrogate mothers ). Second, entiated according to the age of the teaching is rare, even in across human populations, degree spontaneous interactions between of school education correlates posi- youngsters and seems directed to 10 13 motivate them to crack. Up to 3 mother-infant pairs in humans, tively with cognitive capacities. and occurs only when truly years, mothers let them take nuts di- Thus, it seems more reasonable to necessary. In the case of nut rectly from the anvil, from the moth- suggest that chimpanzees living in a cracking, mothers have a direct in- ers' hands, and even from their deprived environment (i.e., captivity) terest in accelerating their young- do not have the opportunity to mouths. Mothers leave up to 40% of sters' performance and helping them CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE develop fully, but, through the sup- Johanson and M. Edey, Lucy: The Beginnings of Hu- Acknowledgments—We thank the Minis- mankind (Simon and Schuster, New York, 1981). plementary enrichment provided by tries of Scientific Research and Agronomy 9. C. Boesch, Teaching in wild chimpanzees, daily language training, some of of Cote d'lvoire for permitting this study in Animal Behaviour, 41, 530-532 (1991). 14 the Tai National Park and the Swiss Na- 10. A. Whiten and P. Milner, The educational their faculties are stimulated. But experiences of Nigerian infants, in Niigerian Chil tional Foundation for Scientific Research life in the wild puts strong survival dren: Development Perspectives, H.V. Curran, Ed. for financing it. (Routledge and Kegan Paul, London, 1984). pressure on most learning processes 11. M. Tomasello, A. Kruger, and H. Ratner, and would constitute an even Cultural learning, Behavioral and Brain Sciences (in press). greater stimulant to cognitive devel- 12. W.A. Mason, Dogs as monkey compan- opment than language training in Notes ions, in New Perspectives on Our Lives With Com- panion Animals, A.H. Katcher and A.M. Beck, Eds. captivity. (University of Pennsylvania Press, Philadelphia, Not surprisingly, therefore, evi- 1. j. Coodall, The Chimpanzees of Gombe 1983). (Belknap Press, Cambridge, MA, 1986). 13. P. Dasen and A. Heron, Cross-cultural tests dence of the highest capacities in 2. C. Boesch and H. Boesch, Optimization of of Piaget's theory, in Handbook of Cross-Cultural nut-cracking with natural hammers by wild chim Psychology-Developmental Psychology, H. Triandis representational abilities ever ob- panzees, Behaviour, 83, 256-286 (1983). and A. Heron, Eds. (Benjamin/Cummings, Menlo served in chimpanzees comes from 3. C. Boesch and H. Boesch, Mental map in Park, CA, 1981). the wild—the rich and highly de- wild chimpanzees: An analysis of hammer trans 14. M. Tomasello, S. Savage-Rumbaugh, and ports for nut cracking, Primates, 25, 160-170 A. Kruger, Imitation of object related actions by manding rain forest habitat of our (1984). chimpanzees and human infants, paper presented at 4. P. Dasen, Concrete operational develop the conference "Understanding Chimpanzees," study community. It is also from this ment in three cultures, Journal of Cross-Cultural Chicago (November 1991); S. Savage-Rumbaugh, forest-dwelling population that Psychology, 6, 156-172 (1975). R.A. Sevcik, D. Rumbaugh, and E. Rubert, The ca 5. D. Premack and A. Premack, The Mind of an pacity of animals to acquire language: Do species novel evidence has emerged on top- Ape (Norton, New York, 1983). differences have anything to say to us? Philosophical ics such as tool use, food sharing, 6. C. Boesch and H. Boesch, Sex difference in Transactions of the Royal Society of London, 308, the use of natural hammers by wild chimpanzees: A 177-185 (1985). teaching, cooperation in hunting, preliminary report, journal of Human Evolution, 10, 15. C. Boesch and H. Boesch, Tool use and and adaptation to predation pressure 585-593 (1981). making in wild chimpanzees, fo/ia Primatologica, 7. C. Boesch and H. Boesch, Possible causes of 54, 86-99 (1990); C. Boesch and H. Boesch, Hunt- by leopards. This evidence suggests, sex difference in the use of natural hammers by wild ing behavior of wild chimpanzees in the Tai Na- in contrast to prevailing theories, chimpanzees, journal of Human Evolution, 13, tional Park, American journal of Physical Anthropol- 415-440 (1984). ogy, 78, 547-573 (1989); C. Boesch, The effect of that the roots of human evolution 8. Y. Coppens, Le Singe, I'Afrique et I'Homme leopard predation on grouping patterns in forest might be found in a forest environ- (Fayard, Paris, 1983); R. Leakey, The Making chimpanzees, Behaviour, 117, 3-4, 220-242 ment. of (1991). Mankind (Book Club Associates, London, 1980); D.