Development of Stone Tool Use by Wild Chimpanzees (Pan troglodytes)
Noriko Inoue-Nakamura, Tetsuro Matsuzawa
Figure 1. Nut cracking by chimpanzees in the outdoor laboratory.
At the age of 3.5 years, wild chimpanzees at Bossou, Guinea, begin to use hammer and anvil stones to crack oil-palm nuts to get the kernels. To clarify the developmental processes, the authors did a field experiment in which stones and oil-palm nuts were provided. Infant chimpanzees' stone-nut manipulation was observed and video recorded. Data were collected from 3 infants younger than 4 years old from 1992 to 1995. The authors analyzed 692 episodes of infants' stone-nut manipulation and 150 episodes of infants' observation of nut cracking performed by adults. Infants observed other chimpanzees' nut cracking and got the kernels from them. The stone-nut manipulation developed from a single action on a single object to multiple actions on multiple objects. Although infant chimpanzees at the age of 2.5 years already acquired basic actions necessary for nut cracking, they did not combine the actions in an appropriate sequence to perform actual nut cracking.
The acquisition of tool-use behavior is a sort of problemsolving task and inevitably reflects cognitive development. Connolly and Dalgleish (1989) examined the developmental change of a spoon-using skill by human children younger than 2 years old. They suggested that the acquisition of tool-using skills indicated the cognitive capacity of the children and the emergence of strategies to solve a pm-ticular problem. Thus, it is important to investigate tool-using skills for understanding the cognitive development of human children.
In an evolutionary context, it is also important to study tool-use behaviors in chimpanzees (Pan troglodytes), the species genetically closest to humans. So far there have been a number of observations about the various kinds of tool use by chimpanzees in the wild (e.g., Goodall, 1970, 1986; McGrew, 1992; Nishida & Hiraiwa, 1982). However, as far as we know, there have been few studies in the wild on the acquisition processes of tool-using skills (Boesch, 1991a; Matsuzawa, 1994). Although there were a couple of studies in groups of captive chimpanzees (Hannah & McGrew, 1987; Sumita, Kitahara-Frisch, & Norikoshi, 1985), no previous studies in the wild have reported the quantitative analysis of behavioral change from an ontogenetic perspective through direct observation and video recording.
Among the tool-use behaviors in the wild, nut-cracking behavior is of special interest because it is the most "complicated" tool use chimpanzees perform. It basically consists of the following actions: (a) picking up a nut, (b) putting it on an anvil stone, (c) holding a hammer stone, (d) hitting the nut on the anvil stone with the hammer stone, and (e) picking up the kernel in the cracked hard shell and eating it. This behavior is characterized by the collaboration of bimanual and asymmetric manipulation. The chimpanzees at Bossou, Guinea, use a pair of stones as a hammer and an anvil to crack open oil-palm nuts to get the kernels (Sugiyama & Koman, 1979b). This tool-use behavior has been reported only among chimpanzees of a limited area of West Africa (Boesch, Marchesi, Fruth, & Joulian, 1994; Kortlandt, 1986; McGrew, 1992; Sugiyama, 1993). No chimpanzees in East Africa have been observed to crack nuts with stones although stones and nuts are available (Goodall, 1986; Nishida, 1990). Nut cracking with stones is a strong example of the diversity of material culture among chimpanzees.
In the field, however, it is difficult to directly observe nut cracking of chimpanzees because the bush beneath the palm trees is thick, and also because the chimpanzees are too timid to be observed when they are in the secondary forest near the village where palm trees are. Therefore, field experiments have been carded out at Bossou since 1987 to study chimpanzees' tool-use behavior in detail. An outdoor laboratory was established in the central part of chimpanzees' free-ranging area where stones and oil-palm nuts were provided by the experimenters (Fushimi, Sakura, Matsuzawa, Ono, & Sugiyama, 1991; Matsuzawa, 1991, 1994; Sakura & Matsuzawa, 1991; Sugiyama, Fushimi, Sakura, & Matsuzawa, 1993).
The present study aims to investigate the developmental process of the tool-use skill from the point of ontogenetic view. The field experiments have revealed that chimpanzees at Bossou younger than 3.5 years old cannot perform nut cracking (Inoue, Tonooka, & Matsuzawa, 1996; Matsuzawa, 1994). The purpose of the present study was to answer the following questions by means of behavioral observations and an analysis of video-recorded data in the outdoor laboratory: How do infant chimpanzees younger than 3.5 years old acquire nut-cracking skill? What kind of social influences are involved in the acquisition process?
The subjects of this study were wild chimpanzees (Pan troglodytes) at Bossou, on the southeastern edge of the Republic of Guinea, West Africa. There was a group of 18-19 chimpanzees at Bossou during the present study of 1992-1995.
The chimpanzees at Bossou have been investigated without supplemental feeding since 1976 by Sugiyama and his colleagues (Sugiyama, 1984, 1988, 1994a; Sugiyama & Koman, 1979a, 1987). Each member of this group has been identified. Chimpanzees at Bossou show various tool-use behaviors, such as nut cracking, ant catching (Sugiyama, Koman, & Sow, 1988), leaf olding for drinking water (Tonooka, Inoue, & Matsuzawa, 1995), and pestle pounding for extracting sap from oil-palm trees (Sugiyama, 1994b; Yamakoshi & Sugiyama, 1995).
This study focused on 3 infant chimpanzees younger than 4 years old in January 1995: Yolo, Fotayu, and Vuavua. The subjects were observed for 4 years. The longitudinal data of the 3 infant chimpanzees were divided into four age groups: 0.5 (0-1 year old), 1.5 (1-2 years old), 2.5 (2-3 years old), and 3.5 (3-4 years old). Table 1 shows the information about each age" group. Chimpanzees younger than 4 years old are usually classified as "infants." Weaning occurs around age 2.5 to 3 years. The interbirth interval is about 4 to 6 years. Length of life is estimated as 40 to 50 years (Goodall, 1986).
Observational periods. The present study reports the field experiments and the behavioral observation done during four dry seasons: (a) January 1992 (18 days), (b) December 1992 to January 1993 (20 days), (c) January 1994 (5 days) and February to March 1994 (30 days), and (d) January 1995 (21 days). Table 1 also provides general information about the records of observation: the days we observed, the total time of observation, and the number of parties that visited the outdoor laboratory.
Setting. In general, about 50 stones and approximately 2-5 kg of oil-palm nuts (Elaeis guineensis) were provided at an outdoor laboratory sized 10 m wide by 5 m deep. The observer hid behind a screen made of grass about 4 m long and 2 m high. The distance between the observer and the cracking site was about 20 m. The observer stayed there continuously from 7 a.m. till 6 p.m. All behaviors were directly observed and videotaped (Sony, CCDTR3) for further analysis. Figure 1 shows a representative scene of nut cracking by chimpanzees in the outdoor laboratory.
Definition of behavioral episodes and observational episodes. Focal animal sampling was performed on all of the video-recorded data of the 4 years (4,674 min total). The target behaviors were (a) manipulation of stones and/or nuts and (b) observation of the other chimpanzees that performed nut cracking.
We defined a behavioral episode as infants' manipulating stones and/or nuts by themselves. As a definition, the episode started when the chimpanzee contacted stones and/or nuts and ended when the subject moved without contacting stones and/or nuts. In other words, each behavioral episode continued as long as each subject did not change the sitting spot and also when the subject moved while contacting the same stones and/or nuts (e.g., transporting stones and/or nuts or stepping on stones).
As to their observing the other chimpanzees that performed nut cracking, we recorded it as an observational episode. It did not matter whether there was physical contact between the performer and the infants or not. For the definition of the observational episode, it was required that the subject's face was oriented to the performer for 3 s or more from less than 1 m away.
Definition of fundamental actions. We postulated that each behavioral episode could consist of single fundamental action or various combination of fundamental actions. Fundamental actions on stones and those on nuts were separately defined and listed in Table 2. The number and sequence of fundamental actions involved in a behavioral episode could vary. The duration of a behavioral episode was also different among the episodes.
A coding system of behavioral episodes. First, the behavioral episodes were divided into two main categories: manipulation of stones and/or nuts while not in physical contact with the other chimpanzees (designated Category I) and manipulation of stones and/or nuts while in physical contact with the other chimpanzees (designated Category II). Then, the types of behavioral episode were coded with a system designed to capture a range of behavioral episodes involving from a single action on a single object, and multiple actions on multiple objects, to nut cracking in each category.
Table 3 illustrates the coding system of behavioral episodes we adopted. The coding system was hierarchically organized according to the number of manipulated objects, the relation of multiple objects manipulated, and the variety of actions involved in a behavioral episode. All behavioral episodes were divided into two main classes according to the number of objects manipulated at a moment: a single object (Class A) and multiple objects (Class B). The behavioral episodes belonging to Class A were then subdivided into two classes according to the total number of objects manipulated in a behavioral episode: a single object (Class A-I) and multiple objects (Class A-2). The behavioral episodes belonging to Class A-1 were further divided into two classes according to the variety of actions: a single action (Class A-l-i) and multiple actions (Class A-l-ii). The behavioral episodes belonging to Class B were subdivided into two classes according to the relation of objects manipulated simultaneously: no relation at all (Class B-l) or any kind of relation among objects (Class B-2). Finally, the behavioral episodes in which a subject successfully hit a nut on a stone with another stone were classified into Class C. Each class of A-l-i, A-l-ii, A-2, B-l, B-2, and C is described as "a single action on a single object," "successive actions on a single object," "successive actions on multiple objects," "simultaneous actions on multiple objects while no relation among objects," "simultaneous actions on multiple objects while any relation among objects," and "nut cracking," respectively.
Each behavioral episode was coded according to this hierarchically organized, mutually exclusive and exhaustive system. The highest code was recorded for each behavioral episode. Recorded data. We measured the start time and the end time of an episode to get the duration of the episode. For each behavioral episode, we also recorded the manipulated objects, the fundamental actions involved, and the code of behavioral episodes. For observational episodes, we recorded the duration of an episode and the performer of nut cracking that the infants observed.
Table 1 Composition of Age Group and Total Record of Observation
a Male subject, bin the observational days, the first number represents the days in which the chimpanzees were actually observed, and the second number represents the days in which the observers waited for the chimpanzees in the outdoor laboratory, c One party means a group of chimpanzees that come to the outdoor laboratory. As a definition, a party lasted from the point when the first chimpanzee came in to the point when the last one left. Therefore, the size of one party could vary from 1 chimpanzee to all of the 18-19 chimpanzees. The number in parentheses represents the mean party size.
Table 2 List of Fundamental Actions on Objects
Table 3 Hierarchical Classification of Behavioral Episodes
Table 4 shows the records of observation in each age group: the total time the 3 infants were actually observed, the mean duration of stay in the outdoor laboratory by an infant, and the number of stays for each infant. According to the definitions, we identified 692 behavioral episodes and 150 observational episodes in total.
Total Duration of Behavioral and Observational Episodes
Figure 2 shows proportions of the amount of time of behavioral episodes and observational episodes to the total time that the infants stayed at the outdoor laboratory for each age group. At the age of 0.5 years, the infants spent little of their time (7%) in behavioral and observational episodes. At the age of 1.5 years, the infants spent 13% of their total stay time in either behavioral episodes or obser vational episodes. At the age of 2.5 and 3.5 years, they spent 37% and 44% of their total stay time, respectively. Duration increased significantly across age groups in Kruskal-Wallis test, H(3, N = 842) = 48.3, p < .001.
Duration was examined separately for Category I and Category II of behavioral episodes and observational episodes. Category I involved the behavioral episodes in which the subjects manipulated objects while not in physical contact with the other chimpanzees, and Category II while in physical contact with the other chimpanzees.
The relative amount of time in Category I (Figure 2, filled columns) changed significantly as a function of age, H(3, N = 410) = 65.5, p < .001. Duration increased significantly between the age of 0.5 and 1.5 years and between 1.5 and 2.5 years in Mann-Whitney U test, U(N = 29, 38) = 232, p < .001; U(N = 38, 15) = 102, p < .001, respectively. Duration also tended to increase between the ages of 2.5 and 3.5 years, although the difference was slightly less than significant, U(N = 15, 32) = 159, p = .065. The relative time of Category II (Figure 2, hatched columns) also changed across age groups, H(3, N = 282) = 10.64, .01 <p <.02. Duration of the 0.5-year age group was significantly shorter than the other groups, U(N = 29, 38) = 336, p = .007. However, those of the 1.5-, 2.5-, and 3.5-age groups did not differ from each other significantly (1.5-2.5 years, U -- 349, p > .05; 2.5-3.5 years, U = 292.5, p > .05).
By contrast, the relative amount of time of observational episodes (open columns) did not change across age groups, H(3, N = 150) = 4.4, p > .05. At the age of 3.5 years, 2 chimpanzees, Fotayu and Vuavua, succeeded in performing nut cracking. However, they still observed adults' per formances (Fotayu, eight instances; Vuavua, seven instances).
These data indicate that the older infants spent more time in behavioral episodes in which the subjects manipulated objects while not in physical contact with the other chimpanzees and that they still observed adults' performances even after succeeding in performing nut cracking.
Manipulated Objects Involved in Each Behavioral Episode
We analyzed developmental processes of objects manipulation by classifying all behavioral episodes into three different types according to the objects manipulated in each behavioral episode: (a) manipulation of only stones, (b) manipulation of only nuts, and (c) manipulation of both stones and nuts, in a behavioral episode. Relative frequency of behavioral episodes involved in each type was compared among four age groups in each category (see Figure 3).
In both cases of Category I (Figure 3a) and Category II (Figure 3b), the manipulation of the objects significantly changed as a function of age in chi-square test, X2(4, N = 410) = l13,p < .001 and X2(6, N = 282) = 42.6,p < .001, respectively. In both cases of Category I and II, the manipulation of both stones and nuts in a behavioral episode increased significantly, X2(2, N = 410) = 90.1, p < .001 and X2(3, N = 282) = 31.7, p < .001, respectively, while it was not observed at the age of 0.5 years. In both cases of Category I and II, the manipulation of only stones decreased significantly, X2(2, N = 410) = 15.9, p < .001 and 9(2(3, N = 282) = 19.5, p < .001, respectively. In the case of Category I, the manipulation of only nuts decreased significantly, XZ(2, N = 410) = 72.6, p < .001. However, it did not change in the case of Category II, X2(3, N = 282) = 6.6, p > .05. In the case of Category II, the proportional frequency of scrounging the kernel cracked open by an adult to the total frequency of the manipulation of only nuts were 0% (0/24), 50% (37/74), 76% (28/37), and 86% (25/29) for the 0.5-, 1.5-, 2.5-, and 3.5-year age groups, respectively. At the age of 3.5 years, 2 chimpanzees that succeeded in performing nut cracking still scrounged adults' kernels (60%, 15/25 in frequency). These results indicate that the manipulation of both stones and nuts in an episode increased as the age progressed and that the infants stole kernels from their mothers continuously even after succeeding in performing nut cracking.
Fundamental Actions Involved in Each Behavioral Episode
A variety of fundamental actions on stones and nuts were performed in behavioral episodes. The number of behavioral episodes in which each fundamental action occurred for each age group is presented in Table 5.
At the age of 0.5 years, the variety of the fundamental actions were only two on stones and seven on nuts. The variety of actions on both stones and nuts dramatically increased at the age of 1.5 years. Five basic actions are needed to perform nut cracking: (Action A) taking (or picking up) a nut, (Action B) putting a nut, (Action C) holding a stone, (Action D) hitting a nut, and (Action E) eating a nut (in bold in Table 5). All of them already occurred at this age. The fundamental actions, such as Brush, Support, and Replace, occurred at the age of 2.5 and 3.5 years. These actions were accompanied in some cases with nut cracking. The four fundamental actions--laying on a stone, pointing a stone, raking nuts, and rolling a nut-- disappeared at the age of 3.5 years. None of these actions are essential for performing nut cracking.
The frequency of fundamental actions also changed across age groups. Especially, the proportional frequency of the five basic actions for nut cracking--(Action C) Hold on stones, and (Action A) Take (or Pick), (Action E) Eat, (Action B) Put, and (Action D) Hit on nuts--increased significantly as a function of age: )(2(3, N = 2,084) = 73.3, p < .001; X2(3, N = 2,084) = 11.2, p < .02; X2(3, N = 2,084) = 10.1, p < .02; X2(3, N = 2,084) = 71.2, p < .001; and X2(3, N = 2,084) = 77.9, p < .001, respectively.
Frequency of Behavioral Episodes in Each Hierarchical Class
Figures 4a and 4b present relative frequency of the behavioral episodes in each hierarchical class compared among four age groups in Category I and Category II, respectively.
In Category I, the infants at the age of 0.5 years did not manipulate any object while not in physical contact with the other chimpanzees (n = 0)because they were always attached to their mothers. At the age of 1.5 years, behavioral episodes in Class A-1 were dominant (77%, 82/107). The infants at the age of 1.5 years also showed behavioral episodes in Class A-2 (10%, 11/107) and Class B (13%, 14/107). At the age of 2.5 years, behavioral episodes in Class A-l-i decreased (7%, 6/83) whereas those in Class A-2 (19%, 16/83) and Class B (31%, 26/83) increased. At the age of 3.5 years, 2 chimpanzees, Fotayu and Vuavua, started nut cracking (Class C; 39%, 86/220). At this stage, 75% of behavioral episodes observed (166/220) belonged to Class B or C. The relative frequency of behavioral episodes except the 0.5-age group changed as a function of age, )(2(10, N = 410) = 185.9, p < .001.
In Category II, at the age of 0.5 years, Class A-1 was dominant (97%, 32/33). At the age of 1.5 years, behavioral episodes in Class A-1 were still dominant (73%, 95/131), but those in Class A-2 increased (19%, 25/131) and Class B started to appear (8%, 11/131). At the age of 2.5 years, Class B increased (29%, 17/58) whereas Class A-l-i decreased (17%, 10/58). At the age of 3.5 years, as in Category I, nut cracking started to be observed. Of behavioral episodes, 52% (31/60) belonged to Class B and Class C. The relative frequency of behavioral episodes changed significantly as a function of age, )(2(15, N = 282) = 97.8, p < .001.
In both Category I and Category II, relative frequency of behavioral episodes in hierarchically higher classes increased whereas that of behavioral episodes in lower classes decreased as a function of age.
Duration of Behavioral Episodes in Each Hierarchical Class
Durations of all behavioral episodes in each hierarchical class were compared in four age groups in Category I and Category II. The values of duration are presented in Table 6. Figure 5 shows statistical results among durations in each hierarchical class, using Mann-Whitney U test. At the age of 0.5 years, in Category II, durations of behavioral episodes in Class A-l-i were significantly shorter than those in Class A-l-ii. At the age of 1.5 years, in both Category I and Category II, durations of behavioral episodes in Class A-l-i were significantly shorter than those in Classes A-1-ii, A-2, and B. Class A-l-ii were significantly shorter than Class A-2. At the age of 2.5 years, in both Category I and Category II, durations of behavioral episodes in Class A-l-i were significantly shorter than those in Classes A-l-ii, A-2, B-I, and B-2. In Category I, durations of behavioral episodes in Class B-1 were shorter than those in Class B-2. In Category II, durations of behavioral episodes in Class A-l-ii were shorter than those in Class B-1. At the age of 3.5 years, in Category I, durations of behavioral episodes in Classes A-l-i and A-l-ii were significantly shorter than those in Classes A-2, B-l, and B-2, which were shorter than Class C. In Category II, durations of behavioral episodes in Classes A-l-i and A-l-ii were significantly shorter than those in Classes B-1, B-2, and C. In both Category I and Category II in each age group, duration of each behavioral episode lengthened as a function of hierarchy defmed by the coding system of the present study.
Interaction With the Other Chimpanzees
In Category II of behavioral episodes, the other chimpanzees with whom the infants were in physical contact were mostly their own mothers: at the age of 0.5 years, 33 out of 33 behavioral episodes (100%); at the age of 1.5 years, 128 out of 131 (98%); at the age of 2.5 years, 51 out of 58 (88%); and at the age of 3.5 years, 54 out of 60 (90%). There were 16 instances in total in which the infants were in physical contact with other chimpanzees than their own mothers: 2 instances in Yolo, 13 instances in Fotayu, and 1 instance in Vuavua, respectively. Except their own mothers, the other adult females (75%) and the sibling (25%) were the chimpanzees that were in physical contact with the infants manipulating stones and/or nuts. We also investigated the proportion of subjects that the infants observed in observational episodes (i.e., whether the infants observed their own mothers or not). The proportion of observation of their mothers were 100%, 71%, 42%, and 14% for the 0.5-, 1.5-, 2.5-, and 3.5-year age groups, respectively. The proportion decreased significantly as a function of age, X2(3, N = 150) = 74.8, p < .001. These data indicate that older infants observed more often the other chimpanzees than their own mothers. Other than their own mothers, the infants observed the other adult females (64%), juveniles (21%), an alpha male (9%), and the siblings (7%).
Five Basic Actions for Nut Cracking
Table 7 shows the frequency of behavioral episodes in which at least one of the five basic actions necessary for nut cracking were involved. The behavioral episodes of actual nut cracking were omitted from this analyses. At the age of 0.5 years, 15 out of 33 behavioral episodes (45%) involved a basic action, but the remaining 18 behavioral episodes did not involve any basic action. The basic action observed at this age was limited to Take or Pick. At the ages of 1.5 and 2.5 years, three basic actions at maximum were involved in a behavioral episode. At the age of 3.5 years, four or even all of the five basic actions were involved in a behavioral episode. Out of six behavioral episodes in which all of the five basic actions were involved, two behavioral episodes were performed by Fotayu, which was already able to do nut cracking, and the others were performed by Yolo, which was not yet able.
Figure 6 shows the conditional probabilities of occurrence of each basic action in each age group except those in the 0.5-year age group. At the age of 1.5 years, the sequence from Action A (Take or Pick) to Action E (Eat) was most dominant, and it decreased as a function of age. In contrast, the probability of the sequence from Action A (Take or Pick) to Action B (Put) increased as a function of age. At the age of 2.5 and 3.5 years, the probabilities of the sequence from Action A (Take or Pick) to Action B (Put), from B to D (Hit), and from D to A were getting higher. The probability of the sequence from Action C (Hold) to Action D (Hit) was almost zero for all age groups. It indicates that the infants seldom hold a stone to hit. It has to be noted that a lot of "reverse" and "short-cut" sequences were observed for all age groups.
Table 8 provides a more detailed analysis of the types of Hit action excluding the behavioral episodes in which actual nut cracking was involved. Among the various possible combinations, the infants did not show the action of "hitting a nut on the ground by a stone."
Another analysis of Hit actions revealed that Hit actions did not appear in the class of a single action on a single object (Class A-I-i). In contrast, more than 97% of simultaneous actions on multiple objects (Class B-1 and B-2) involved Hit actions. It was also revealed that Hit actions frequently appeared when the infants manipulated both stones and nuts in a behavioral episode (81%).
Table 4 Records of Observation in Each Age Group
Figure 2. Percentage amount of time of behavioral and observational episodes in each age group.
Figure 3. Developmental processes of the objects manipulated in each behavioral episode in (a)
Category I and (b) Category II. The number above each column represents the total number of
Table 5 Variation and Frequency of Fundamental Actions
Figure 4. Developmental processes of the behavioral episodes for each hierarchical class in (a)
Category I and (b) Category II (see Table 3 for class description). The number above each column
represents the total number of behavioral episodes.
Table 6 Duration of Behavioral Episodes (in Seconds) in Each Hierarchical Class in Four Age
Figure 5. Statistical results among durations in each hierarchical class (see Table 3 for class
description) using Mann-Whimey U test. The results are schematically drawn in each age group and
in Category I and Category II. The single and double lines represent statistically significant
differences at the level ofp < .05 and p < .01, respectively. Each arrow is drawn from the shorter
to the longer duration.
Table 7 Frequencies of Behavioral Episodes Involving Basic Actions
Figure 6. Conditional probabilities of occurrence of each basic
action. The thickness of arrows represents conditional probabilities.
Only the five basic actions were taken into account in the
present analysis. Because the other fundamental actions were
omittr..d, A ---> B, for example, does not always represent "taking a
nut and then putting it on a stone." There is the probability of
existence of some fundamental actions between the two basic
actions, "taking a nut" and "putting a nut on a stone."
Table 8 Frequencies of Each Type of Hit Action on
Stones and Nuts
The present study has focused on the acquisition of nutcracking skill by wild infant chimpanzees younger than 4 years old in the outdoor laboratory. The quantitative analysis of stone-nut manipulation shown in the results has illuminated the developmental changes leading to the actual nut cracking. The following two major points are discussed: the ontogeny of chimpanzees' stone-nut manipulation and the social influence on the acquisition of nut cracking.
Ontogeny of Stone-Nut Manipulation
As shown in the process of acquisition of nut cracking, the infants manipulated a single object in a single action in the early stage of development. As a function of age, a single action on a single object (only stone or only nut) developed into multiple actions on multiple objects (some stones, some nuts, or both stones and nuts). The older infants had a tendency to manipulate both stones and nuts successively or simultaneously and to physically contact nuts with stones in a behavioral episode. As the age progressed, the infants showed the more complex and the more hierarchically organized pattern of action-object relationship.
To accomplish the nut cracking, the infant chimpanzee must put together the five basic actions: Take (Pick), Put, Hold, Hit, and Eat. At the age of 1.5 years, all of these basic actions had already been performed, and the frequency of these basic actions increased as a function of age. However, infants did not combine with each basic action in an appropriate way. It took 3.5 years for infant chimpanzees to succeed in combining these actions in an adequate sequence and performing actual nut cracking.
For instance, the infant chimpanzees of the age of 2.5 years often put a nut on a stone but hit the nut with the back of a hand, and then picked up a piece of kernel stuck inside of a broken shell from the ground and ate it. In another case, the infants once put a nut on a stone and then turned to the mothers to scrounge a kernel in the nut that was cracked open by their mothers.
Similar observation was also reported in human infants (Connolly & Dalgleish, 1989). Human infants younger than 2 years old sometimes put their spoon with one hand into and out of the dish repeatedly while taking food from the dish with the other hand.
Both infant chimpanzees of the age group of 2.5 years old and human infants younger than 2 years old have already acquired each basic action. However, neither of them can combine together each action in an appropriate manner. The lack of adequate composition of each action is a common characteristic of the chimpanzees and humans just before acquiring the tool-use skills.
On the other hand, we found an interesting difference between the two species in an action involved in the stonenut manipulation. Infant chimpanzees in the present study that were not able to perform nut cracking never hit a nut with a hammer stone. On the contrary, human infants showed "hitting a nut on the ground with a stone held by a hand" in the developmental process to acquire the nutcracking skill (Matsuzawa, 1994). The infant chimpanzees had a tendency to easily relate a nut with an anvil in an early stage like 1.5 years old; that is, they could put a nut on a stone. However, it may be difficult for chimpanzees to hold a stone and hit a nut with the stone. A juvenile female, Yunro (7 years old in 1992), was not able to use stones as a tool to crack open a nut. In her case, she put a nut on a stone and hit the nut with her left knuckle or her right foot, whereas she never held a hammer stone. This behavior was exactly the same as those shown in the 2.5-year-old chimpanzees just before starting actual nut cracking.
Social Influence and Practice
Infant chimpanzees observed nut-cracking behaviors performed by their own mothers or the other chimpanzees while manipulating stones and nuts on their own in the process of acquiring the nut-cracking skill. As the age progressed, infant chimpanzees spent more time in observing the other chimpanzees other than their own mothers throughout a period of learning skills of nut-cracking behavior. Moreover, they even scrounged kernels from chimpanzees other than their own mothers. It suggests that mothers are not the only resource to provide the model of the tool-use skill. The other chimpanzees in the same community seems to play very important roles in acquiring nutcracking behaviors. For example, at Bossou there are two adult females, Nina and Pama, who cannot perform nut cracking. Their offspring, Na (6.5 years old in 1992) and Pili (5 years old in 1992), showed no retardation in acquiring nut-cracking skill (Matsuzawa, 1994). They were likely to acquire the skill under the influence of members in the community and through practicing on their own.
Social learning is likely to play an important role in the acquisition of tool-use behaviors for those chimpanzees living in a group in which such behaviors are relatively common (Beck, 1980; Thorpe, 1956). According to Thorpe (1956), there are three social learning processes. The first is social facilitation. The demonstrator may simply act as a releasing stimulus to the observer. The second is stimulus/ local enhancement. The demonstrator may influence learning by making a particular stimulus or location more salient and thus more likely to draw the observer's attention. Stimulus/ local enhancement increases the likelihood that the observer will learn the behavior on its own by trial and error. The third is true imitation. The demonstrator may model the behavior, which the observer then copies in its entirety. To qualify as true imitation, the observer must produce a topographical duplication of a novel behavior without trial and error.
Tomasello, Davis-Dasilva, Camak, and Bard (1987) conducted an experimental study of social learning of a tool-use behavior by captive juvenile chimpanzees. They suggested that chimpanzees learned the skill of tool-use not through true imitation learning but through emulation learning. Emulation is a process similar tO stimulus enhancement but is a bit more sophisticated. Chimpanzees did not copy the demonstrator's actual methods of tool use by true imitation. However, they did not simply pay attention to the tool (stimulus enhancement) but also learned something about the general functioning of the task and the results obtained by the demonstrator (emulation).
What type of social learning process is involved in the acquisition of nut-cracking skill? True imitation cannot explain the results of the present study. The infants showed various combinations of stones and nuts. They also showed a variety of fundamental actions. Not all of them were adequate actions for actual nut cracking. They gradually increased the relative frequency of adequate sequence of the basic actions through each stage of development. They did not copy the motor patterns or the way to relate nuts with stones, which were shown in the tool use by mothers and the other members of the community. As the present results suggest, they learned the general functional relations of stones and nuts and also learned the goals obtained by the demonstrator. This learning process might be called emulation.
Active teaching seems to be a clear distinction in the acquisition process between humans and chimpanzees. Active teaching is rare in chimpanzees in the wild although it is popular in humans. So far, there have been no instances of active teaching or guidance except two reported episodes in which mothers influenced their infants' attempt to crack nuts (Boesch, 1991b). Human mothers usually give their infants verbal instructions about how to grasp a spoon. They even mold the hands of infants to show the correct way of grasping. In human societies, they also give their infants social reinforcement, such as social praise. On the contrary, in the case of chimpanzees of the present study, they did not show any kind of active teaching to the infants. They also did not give them any social reinforcement. The chimpanzee mothers' attitude toward their infants was characterized by the lack of any feedback to infant chimpanzees' attempts at nut-cracking behaviors.
In chimpanzee societies, adults are extremely tolerant of infants, which are common in some primate species including humans (e.g., Fragaszy & Visalberghi, 1989; Visalberghi & Fragaszy, 1990). In the case of juveniles, however, the tolerance is weaker. For example, juvenile chimpanzees were often observed not to be able to take the stones and/or nuts beside dominant adults. They were often chased out when they were trying to get stones and/or nuts at the side of the adults. In summary, the members in the community provided only the infants with the opportunities to freely access stones and nuts. These opportunities could facilitate the individual experience of stone-nut manipulation and result in the apparent social transmission of the tool-use behavior among the wild chimpanzees.
The present research was financed by Grant 01041058 from the International Research Program of the Ministry of Education, Science, Sports, and Culture, Japan. Preparation of the article was supported by the Fellowship of the Japan Society for the Promotion of Science for Japanese Junior Scientists (No. 0801). The field study was carried out with the collaboration of the Direction Nationale de la Recherche Scientifique et Technique, R6publique de Guinee, and the villagers at Bossou.
We wish to thank Yukimaru Sugiyama, who has conducted field research at Bossou since 1976 and who gave us the opportunity to do this research. Special thanks are due to the following colleagues during the research periods: Rikako Tonooka, Gen Yamakoshi, Jeremy Koman, Guano Goumy, and Tino Camara. A part of the video-recorded data (February-March 1994) was collected by R. Tonooka. We also express our thanks to members of the Japanese Embassy in the R6publique de Guin6e and Côte d'Ivoire and the Japan International Cooperation Agency of Côte d'Ivoire. The preparation of the manuscript was done during the period when Noriko Inoue-Nakamura stayed at the Department of Ecology and Evolutionary Biology, Princeton University. We wish to express our thanks for the helpful comments on an earlier draft made by Alison Jolly. Noriko Inoue-Nakamura is grateful to Hiroshi Imada for his generous guidance throughout the present study and also thanks Katsuki Nakamura for invaluable suggestions on data analyses and constant encouragement.
- Beck, B. B. (1980). Animal tool behavior. New York: Garland.
- Boesch, C. (1991a). Handedness in wild chimpanzees. International Journal of Primatology, 12, 541-558.
- Boesch, C. (1991b). Teaching among wild chimpanzees. Animal Behaviour, 41, 530-532.
- Boesch, C., Marchesi, P., Fruth, B., & Joulian, F. (1994). Is nut cracking in wild chimpanzees a cultural behavior? Journal of Human Evolution, 26, 325-338.
- Cormolly, K., & Dalgleish, M. (1989). The emergence of a toolusing skill in infancy. Developmental Psychology, 25, 894-912.
- Fragaszy, D. M., & Visalberghi, E. (1989). Social influences on the acquisition of tool-using behaviors in tufted capuchin monkeys ( Cebus apella). Journal of Comparative Psychology, 103, 159-170.
- Fushimi, T., Sakura, O., Matsuzawa, T., Ono, H., & Sugiyama, Y.(1991). Nut-cracking behavior of wild chimpanzees (PantrogIodytes) in Bossou, Guinea (West Africa). In A. Ehara, T. Kimura, O. Takenaka, & M. Iwamoto (Eds.), Primatology today (pp. 695-696). Amsterdam: Elsevier.
- Goodall, J. (1970). Tool using in primate and other vertebrates. In D. S. Lehrman, R. A. Hinde, & E. Shaw (Eds.), Advances in the study of behavior (pp. 195-249). New York: Academic Press.
- Goodall, J. (1986). The chimpanzees of Gombe: Patterns of behavior. Cambridge, MA: Harvard University Press.
- Hannah, A., & McGrew, W. (1987). Chimpanzees using stones to crack open oil palm nuts in Liberia. Primates, 28, 31-46.
- Inoue, N., Tonooka, R., & Matsuzawa, T. (1996). Developmental processes of nut-cracking skill among infant chimpanzees in the wild. The Japanese Journal of Developmental Psychology, 7, 148-158. (in Japanese with English summary)
- Kortlandt, A. (1986). The use of stone tools by wild-living chimpanzees and earliest hominids. Journal of Human Evolution, 15, 77-132.
- McGrew, W. C. (1992). Chimpanzee material culture. Cambridge, MA: Harvard University Press.
- Matsuzawa, T. (1991). Nesting cups and metatools in chimpanzees. Behavioral and Brain Sciences, 14, 570-571.
- Matsuzawa, T. (1994). Field experiments on use of stone tools by chimpanzees in the wild. In R. Wrangham, W. McGrew, F. de Waal, & P. Heltne (Eds.), Chimpanzee cultures (pp. 351-370). Cambridge, MA: Harvard University Press.
- Nishida, T. (1990). The chimpanzees of the Mahale mountains: Sexual and life history strategies. Tokyo: Tokyo University Press.
- Nishida, T., & Hiraiwa, M. (1982). Natural history of a tool-using behavior by wild chimpanzees in feeding upon wood-boring ants. Journal of Human Evolution, 11, 73-99.
- Sakura, O., & Matsuzawa, T. (1991). Flexibility of wild chimpanzee nut-cracking behavior using stone hammers and anvils: An experimental analysis. Ethology, 87, 237-248.
- Sugiyama, Y. (1984). Population dynamics of wild chimpanzees at Bossou, Guinea, between 1976 and 1983. Primates, 25, 391-400.
- Sugiyama, Y. (1988). Grooming interactions among adult chimpanzees at Bossou, Guinea, with special reference to social structure. International Journal of Primatology, 9, 393-407.
- Sugiyama, Y. (1993). Local variation of tools and tool use among wild chimpanzee populations. In A. Berthelet & J. Chavaillon (Eds.), The use of tools by human and non-human primates (pp. 175-187). Oxford, England: Clarendon Press.
- Sugiyama, Y. (1994a). Age-specific birth rate and life time reproductive success of chimpanzees at Bossou, Guinea. American Journal of Primatology, 32, 311-318.
- Sugiyama, Y. (1994b). Tool use by wild chimpanzees. Nature, 367, 327.
- Sugiyama, Y., Fushimi, T., Sakura, O., & Matsuzawa, T. (1993). Hand preference and tool use in wild chimpanzees. Primates, 34, 151-159.
- Sugiyama, Y., & Koman, J. (1979a). Social structure and dynam ics of wild chimpanzees at Bossou, Guinea. Primates, 20, 323-339.
- Sugiyama, Y., & Koman, J. (1979b). Tool-using and making behavior in wild chimpanzees at Bosson, Guinea. Primates, 20, 513-524.
- Sugiyama, Y., & Koman, J., (1987). A preliminary list of chimpanzees' alimentafion at Bossou, Guinea. Primates, 28, 133-147.
- Sugiyama, Y., Koman, J., & Sow, M. B. (1988). Ant-catching wands of wild chimpanzees at Bossou, Guinea. Folia Primatologica, 51, 56-60.
- Sumita, K., Kitahara-Frisch, J., & Norikoshi, K. (1985). The acquisition of stone tool use in captive chimpanzees. Primates, 26, 168-181.
- Thorpe, W. H. (1956). Learning and instinct in animals. London: Methuen.
- Tomasello, M., Davis-Dasilva, M., Camak, L., & Bard, K. (1987). Observational learning of tool-use by young chimpanzees. Human Evolution, 2, 175-183.
- Tonooka, R., Inoue, N., & Mastuzawa, T. (1995). Leaf-folding behavior for drinking water by wild chimpanzees at Bossou, Guinea: A field experiment and leaf selectivity. Primate Research, 10, 307-313. (in Japanese with English summary)
- Visalberghi, E., & Fragaszy, D. (1990). Do monkeys ape? In S. Parker & K. Gibson (Eds.), "Language" and intelligence in primates: Comparative developmental perspectives (pp. 247- 273). Cambridge, England: Cambridge University Press.
- Yamakoshi, G., & Sugiyama, Y. (1995). Pestle-pounding behavior of wild chimpanzees at Bossou, Guinea: A newly observed tool-using behavior. Primates, 36, 489-500.