Behavioural Processes 79 (2008) 19–27

Contents lists available at ScienceDirect

Behavioural Processes journal homepage: www.elsevier.com/locate/behavproc

Chimpanzees fail to plan in an exchange task but succeed in a tool-using procedure V. Dufour a,b,∗ , E.H.M. Sterck b,c a

Ethology Research, Animal Science Department, Biomedical Primate Research Centre, Lange Kleiweg 139, 2288 GJ Rijswijk, The Netherlands Behavioural Biology, Utrecht University, Kruijt Gebouw Room O 211, Padualaan 8, 3584 CH Utrecht, The Netherlands c School of Psychology, University of St. Andrews, St. Mary’s College, South Street, St. Andrews, Fife, Scotland KY16 9JP, United Kingdom b

a r t i c l e

i n f o

Article history: Received 28 November 2007 Received in revised form 29 March 2008 Accepted 11 April 2008 Keywords: Planning Tool-use Exchange task Cooperation Chimpanzees

a b s t r a c t Planning has long been considered a uniquely human capacity. Lately, however, it has been shown that apes and a corvid species act now to derive a material future benefit. Since primates are highly social animals and their sociality is considered a strong selective force that resulted in complex cognitive capacities, planning is also expected in social situations. Unfortunately, prompting from social partners cannot be excluded in a social setting. Therefore, we controlled for this factor by testing the capacity to plan in chimpanzees using an exchange paradigm, that involves both a material and a social component, and a tool-use paradigm, similar to the one used on two other ape species. All chimpanzees failed to plan in the exchange task, but three individuals showed planning behavior in the tool-use task. Our methods controlled for the fact that chimpanzees were not prompted by the visibility of the reward at the moment of planning and also could not repeat a previously acquired routine. The best interpretation for our results is that chimpanzees can plan. However, planning was limited to the situation where the action to attain the future benefit only depended on a chimpanzee’s own behavior. © 2008 Elsevier B.V. All rights reserved.

Whether humans are the sole species that evolved episodic memory, the capacity to mentally travel in time, is currently subject of a heated debate (Schwartz and Evans, 2001; Clayton et al., 2003a,b; Roberts, 2002; Suddendorf and Busby, 2003a,b). Many researchers (Roberts, 2002; Suddendorf and Busby, 2003a) claim that animals have no recollections of personal events and consequently are not able to plan for future needs. Instead, it has been suggested that animals are “stuck in time” (Roberts, 2002). Animals generally favor the immediate reward (Richards et al., 1997; Tobin et al., 1996), when exposed to one small immediate and one larger but delayed reward. Silverberg et al. (1998) noted that chimpanzees (Pan troglodytes) were unable to stock fruits in advance, thus exhibiting ‘temporal myopia’ – the inability to anticipate a future need –. Yet in a non-primate species, Clayton and Dickinson (1998) showed that scrub jays (Aphelocoma californica), a food hoarding bird species, possess episodic-like memory of a past event, since they integrate information on ‘what-where-when’ to recover cached food (Clayton et al., 2001). Moreover, scrub jays react to the potential of future pilfering by reaching the food (Dally et al., 2006) and future planning was suggested as an explanation for food caching behavior in the evening before a predictable lack

∗ Corresponding author. Tel.: +44 7954 989813; fax: +44 1334 463042. E-mail address: [email protected] (V. Dufour). 0376-6357/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.beproc.2008.04.003

of “breakfast” the next morning (Raby et al., 2007). Also primates show future-oriented behavior. Monkeys going to distant food sites choose the appropriate route minutes in advance (Cebus apella, Janson, 1998). Captive chimpanzees spontaneously save tokens, indicating anticipation of a delayed opportunity to exchange them (Sousa and Matsuzawa, 2001). Recent work (Dufour et al., 2007; Beran and Evans, 2006) shows that chimpanzees can postpone an expected reward for a time scale measured in minutes, not seconds. In both common chimpanzees and bonobos, individuals trained to use symbols were able to announce intended action and travel destination plans (Menzel et al., 2002). However, only one study has yet experimentally shown that apes performed future-oriented behavior that was not driven by current need, a necessary condition for planning (Mulcahy and Call, 2006). In this study, orangutans (Pongo pygmaeus) and bonobos (Pan paniscus) collected, saved and transported tools to use them later. Individuals of both species could plan for a predictable event 1 h, and even one night, ahead. It has, however, been argued that prompting, the food was visible at the time of tool collection, or learning from the first task to solve the second task may also explain these results (Suddendorf and Corballis, 2007). Advanced cognitive capacities, such as future planning, may have evolved in primates to cope with a complex environment. This complexity may concern their social environment (Byrne and Whiten, 1988), extractive foraging requiring tool use (Reader and

20

V. Dufour, E.H.M. Sterck / Behavioural Processes 79 (2008) 19–27

Laland, 2001) or a combination of the two (Reader and Laland, 2002). Anecdotal evidence on chimpanzees suggests that they are able to plan for future material needs. Wild chimpanzees travel over hundreds of meters in dense vegetation to pick up and transport tools necessary to crack nuts (Boesch and Boesch, 1984). Chimpanzees may also plan for future social needs. They seem to maintain a high dominance rank through tactical social maneuvering (de Waal, 1982). Moreover, chimpanzees that groom in the morning have a higher chance to obtain a share of food from their grooming partner in the afternoon (de Waal, 1997). However, interpreting these observations as evidence of planning remains speculative since other, lower-level, explanations are possible. For example, chimpanzees’ tool transport may result from the current motivational state of the individual (Suddendorf and Corballis, 1997). Moreover, in a cooperative context chimpanzees fail to show other regarding behavior (Jensen et al., 2006; Silk et al., 2005) or to understand cues givens by others who communicate cooperative intentions (Hare and Tomasello, 2004). As suggested by Byne (2007), cooperation observed in chimpanzees may not result from a calculation of future benefits (planning), but from their inclination to favor those with whom cooperation has been successful before. Herrmann and Tomasello (2006) have shown that chimpanzees express more cognitive skills when in a competitive context rather than in a cooperative one: chimpanzees deduce from a communicative gesture from a competitor toward a bucket that the food must be hidden there. A similar gesture made by a cooperative partner with the intent to communicate the location of the food is not understood. Cognitive skills in a cooperative context may be constrained by the high level of competition in this species (Hare and Tomasello, 2004; Hare et al., 2007). If this is the case, planning for the future involving cooperation may be a more challenging task than planning for a future activity where success only depends on one’s own behavior. We aimed to test the capacity to plan in chimpanzees. Most tests of planning in animals concern a future material benefit and do not involve uncertainty related to other factors such as a cooperative partner. In the present study we tested for chimpanzees’ planning capacities involving material benefits based on a cooperative interaction (Part 1), and planning capacities involving material benefits only (Part 2).

1. Study 1: planning to selectively collect items for later exchange To be qualified as planning, a behavior must be performed in the present without being driven by current need (Tulving, 1983) or by current cues to future benefits (Suddendorf, 2006). Tulving (2005) proposed the ‘spoon-test’ to test for planning in animals. In this test, a girl was disappointed because she could not get desert at a party since she did not bring her own spoon. The crucial test for planning capacities is to check whether this memory of disappointment will lead her to bring a spoon to the next party. In this study, we devised an analogue of the spoon test (Tulving, 2005), to assess chimpanzees’ capacity to plan for a future benefit involving a cooperative interaction. Unfortunately, purely social interactions cannot be used to test future directed behavior, since the anticipated future event will always involve other individuals and direct reactions to the others cannot be ruled out. Therefore, we tested chimpanzees’ capacities in an exchange paradigm, where the cooperative individual was represented by a human partner. While the human partner could be considered as a social tool, failure may be considered from the subjects point of view as a failure in the partner behavior, and not his own. From this point of view it involves a social component that the subject cannot control. In this paradigm,

a food reward could be obtained from a human partner upon the return of a specific object at a specific time of the day. To succeed in this task, chimpanzees first had to collect earlier in the day a specific category of object among three available for a limited time interval, and return with it in the test compartment where the exchange activity would take place with the human partner. 1.1. Methods 1.1.1. Subjects Two groups of chimpanzees from the Biomedical Primate Research Centre’s colony participated to this study. They were housed in a similar enclosure with the same facilities and daily routine. Group 1 was composed of two males and three females aged from 19 to 23 years. Group 2 was composed of two females and four males aged from 12 to 15 years. Each group was housed in its communal room (a 40 m2 room with elevated sitting devices) to which individual feeding compartments were permanently connected. These compartments were used as individual testing compartments. One individual in group 1 never learned how to use objects and was therefore not included in the study. The study was designed to provide enrichment without disturbing the daily routine of the groups. Participants were given the choice to come or not and where always rewarded with a small treat at the end of the session, whatever their level of participation to the task. Both groups had never been observed for ethological purposes before. They had never been trained for or involved in cognitive tasks. The chimpanzees were not reported to have been trained to exchange with humans before. 1.1.2. Training Prior to the tasks, the chimpanzees were all invited to enter their individual compartment, a familiar procedure. For the exchange task, chimpanzees were trained to return an object to the human partner to obtain a food reward. The object to trade was a colorful plastic straw that was easy to insert through the front of the compartment and was unknown to all subjects. The human partner (V.D.) proceeded as follows: she begged the chimpanzee with the palm up presented in front of her/him, holding a peanut in the other hand, and asked verbally for the return of the token (‘terug’ the Dutch equivalent for ‘back’) 12 times in a row. After a successful exchange, the chimpanzee was given 15 s to eat the peanut. Chimpanzees learned to give the straw back with a mean of 4.7 sessions of 12 trials each (range 1–17 sessions). Once all individuals mastered the exchange task, this task was given by the partner as a daily activity for 20 consecutive daily sessions at 13:10 every afternoon. This insured predictability of the activity and also insured that the human partner was associated with the exchange task. The human partner had at the start of this study been known to the chimpanzees for half a year and she was only involved in neutral or positive interactions with the animals. The chimpanzees had known the keepers from at least 5 years. To determine whether the chimpanzees planned for the exchange task at hand or were interested in just any object, other objects were present as distracters at the time of testing. These objects, like the straws used in the exchange activity, were also associated with a desired food, a specific function, and a specific time of the day. The chimpanzees were provided a natural wooden branch that was to be used in a dipping task. The chimpanzees were trained to dip it in a box full of honey that was held in front of them by a keeper. Since the dipping task was part of their weekly enrichment, only 3 sessions of 3 min per day and per individual were required to insure that all the individuals performed the task. The third object consisted in a sharp wooden stick that was used in a picking task. The chimpanzees were trained to retrieve, by picking into them, orange pieces from a plastic box held in front of the

V. Dufour, E.H.M. Sterck / Behavioural Processes 79 (2008) 19–27

21

Fig. 1. Chronology of events in the planning to exchange (study 1, top) and planning to use a tool (study 5, bottom) tasks. On the pre-test day, chimpanzees should not collect the correct tool at 12:00, since they do not know that tools will not be available at the testing period, after noon. During the testing period on this day they encounter their first failure. The memory of this failure is critical for the following testing days. If chimpanzees remember missing the correct object, they should adjust their behavior, collect the required tool at 12:00 and bring them along at 13:10. A non-attributed compartment is a compartment in which none of the subject is invited at the time of the task. Categories of objects present are straw, branch and stick in study 1, hook, straw, branch and stick in study 5.

testing compartment by a keeper. Fifteen training sessions were conducted until all individuals efficiently performed the task for a 3 min period. Once this criterion was reached, the dipping and the picking task were alternatively conducted by the keepers for 10 sessions each and at the same hour every morning, while exchange tasks were conducted in the afternoon. This insured that all tasks and objects had all been recently practiced by subjects and involved the presence of a human being, either the cooperative partner in the exchange task, or a keeper who merely presented the food in the dipping and picking task. The human partner never assisted in these last two tasks. 1.1.3. Testing From the first day the testing procedure was run, the chimpanzees were given at noon a collection period of 10 min where three objects of each kind per individual were available in one of the individual compartments (in total 12 object for each category in group 1 and 15 object of each category in group 2). During

this collecting period, all members of the group were free to come and collect one or more objects in the compartment. This was done in the absence of the human partner, to avoid any prompting effect. After 10 min, a keeper closed the compartment and removed the remaining objects when any were left. Collected objects were left with chimpanzees in the communal room where all individuals could handle them if the initial collectors had dropped them. The keeper reported afterwards the rough numbers of object collected. One hour after the collecting period, the keeper invited each individual to enter its feeding compartment (different from the collecting compartment that was therefore not attributed to any individual), still in absence of the human partner. The chimpanzees could bring an object when these were present in the communal room. The keeper locked the compartment door behind each chimpanzee. Only once all chimpanzees occupied their individual compartment, did he partner enter the room for the exchange session. The partner held the peanuts in her hand in full view of the subject and begged, asking “terug” with palm open, for 3 min per

22

V. Dufour, E.H.M. Sterck / Behavioural Processes 79 (2008) 19–27

Table 1 Study 1: the planning to exchange task: bringing objects inside the individual compartment during pre-test and the testing period Phase 1

Phase 2

Success

Object brought in

Success

Object brought in

Group 1 Phil Ton Femma Peggy

2 1 1 0

PT (1Br), T3 (1Sw), T5 (1Sw, 1Br) T2 (1Br), T4 (1Br), T5 (1Sw, 1Br) T5 (1Sw) 0

1 1 1 0

PT (2Br), T4 (1Sw, 1Sk, 1Br) T1 (1Sw, 1Br), T7 (1Br) T1 (1Sk), T4 (2Sw) 0

Group 2 Thomas Kenny Zorro Juus Iris Willy

0 0 0 0 0 0

T2 (2Br), T3 (1Sk) T5 (1Br) 0 0 0 0

0 0 2 1 NT NT

T7 (1Br), T8 (1Br) 0 PT (Sk), T3 (1Sw, 1Sk), T4 (2Sw) T1 (2Sw, 1Sk) NT NT

Success: a success is qualified as a test session where the individual brought at least one straw. Objects brought in: test session in which objects were brought in (PT: pre-test session, T: test session; number of the test session); number of objects; and quality of objects (Br: branch, Sw: straw, Sk: stick) per individual and per phase. NT = no testing. Testing in phase 1 was interrupted after the fifth session, due to technical failure of the housing facilities system. After retraining, subjects were tested again in phase 2 that was conducted similar to phase 1.

individual. A begging gesture was performed once every 30 s. If at least one straw was brought and returned to the partner upon request, a series of 12 exchanges was performed with this straw. Although exactly the same procedure was used, we labeled the first day of the procedure the pre-test session and all other days test sessions. In the pre-test session, chimpanzees could not anticipate that they would not have the usual access to straws at the time of the exchange session, since they had not been exposed to the procedure before. They were not expected to collect or bring any object into the test compartment. The following days were the crucial testing days. If chimpanzees remembered their previous failure and anticipated the coming exchange session, they were expected to collect the straws and carry some into the communal room, taking them later to their individual compartment for the exchange activity (for a summary of the procedure see Fig. 1). For technical reasons, this study was run in two identical phases. In a first phase (Phase 1), chimpanzees were given a pre-test session followed by 5 test sessions. This testing was interrupted because of technical failures in the housing facilities system. Ten individuals participated consistently in this phase. A second phase (Phase 2) was run 2 months later. All subjects were retrained on the 3 tasks, receiving 3 consecutive daily sessions of 12 trials each of the exchange task after noon to refresh the predictability of this activity; and receiving the 2 other tasks from the keepers for 3 morning sessions. A pre-test session was then run similarly to phase 1, followed by 10 consecutive test sessions. In this phase Iris and Willy (group 2) did not come consistently into the test compartment and were not tested further. Analyses were conducted using nonparametric statistics for small samples. Due to the limited number of testing days in phase 1, the analyses are conducted on phase 2 (and on the compiled dataset of both phases to illustrate results’ consistency between phases, see end of Section 1.2). 1.2. Results Individual collection behavior in phase 2, where a series of 10 testing days was completed, could not be assessed accurately since chimpanzees were tested in a group context. In group 1 all individuals took part in object collection and all objects were taken out of the compartment and transported into the main room during all sessions in phase 2. In group 2, from the pre-test day to day 8 of phase 2, all objects had been collected and brought back in the common room by all individuals. At days 9 and 10 however, none of the objects had been collected and the subjects did not enter

the compartment where the objects were available. The keeper reported that apart for these 2 last testing days, all individuals generally came for collection time, and showed no apparent selectivity, taking handfuls of objects back with them in the main room. Concerning success in returning the objects in phase 2, the results show that six individuals out of eight brought objects into the test compartment before the exchange task 1 h after the collecting period (see Table 1, for individual results at each phase). They all brought the correct object at least once. However, only one subject, Zorro, took the correct object to the testing chamber and solved the task more than once, on testing days 3 and 4, but did not succeed further. The subjects brought in total nine straws, seven branches and five sticks. Straws were brought significantly more often than sticks, but straws were not significantly preferred to branches (Wilcoxon matched-pairs signed-rank test: straws vs. branches, tied-Z = 0.54, P = 0.59; straws vs. sticks, tied-Z = 2, P = 0.046; branch vs. stick, tied-Z = 0.65, P = 0.52). When combining the results of both phases for the 8 individuals that participated in every test session, we observed no significant preference for straws over other objects (Wilcoxon matched-pairs signed-rank test: straws vs. branches, tied-Z = 0.26, P = 0.8; straws vs. sticks, tied-Z = 1.8, P = 0.07; branch vs. stick, tied-Z = 1.02, P = 0.31). 1.3. Conclusion Subjects brought objects with them when invited into their individual compartment, yet despite a preference in phase 2 for straws over sticks, they failed to accurately select straws over branches. Subjects rarely brought the correct object with them, and one success did not lead them to subsequently solving the task again, since only 1 subject solved the task twice in phase 2. This made us question whether they anticipated the exchange task. Moreover, subjects showed no clear selectivity for the object to exchange. Therefore, our results cannot be considered as evidence of planning by any of the individuals. One possible interpretation of our results is that chimpanzees brought inside objects because they liked to carry them along. This implies that bringing something in is not related to any planning of the task. Indeed, during the pre-test sessions of phase 1 and phase 2, when they could not predict that the human partner would come without straws, two individuals brought in one or more objects (Table 1). Another interpretation of this result is that chimpanzees carried objects with them in the test session with no consideration for their function, but following their preference for those objects or for the

V. Dufour, E.H.M. Sterck / Behavioural Processes 79 (2008) 19–27

23

Table 2 Study 2: Object and food preferences in relation to quality and quantity of objects brought in per individual at study 1, phase 2 Object brought at both phase: (including PT)

Object preferences (80% or over for 1 item)

Food preferences (80% or over for 1 item)

ST/BR

BR/SK

ST/SK

P/H

H/O

P/O

O H

BR

ST ST ST

P H

P

BR BR BR

ST

BR

SK

Group 1 Phil Ton Femma Peggy

1 1 2 0

3 2 0 0

1 0 1 0

Group 2 Thomas Kenny Zorro Juus Willy

0 0 3 2 0

2 0 0 0 0

0 0 2 1 0

P H P

ST ST

ST BR

H O

P P P P P

A preference is indicated when an individual preferred one item over the other in 80% or more of the presentations of objects or foods (ST: straw, BR: branch, SK: stick, P: peanut, H: honey, O: orange).

associated food reward. To test for this hypothesis, individual preferences for objects and associated rewards were checked in study 2. 2. Study 2: checking for object and food preferences Preference for one food, one object or one task may have driven the chimpanzees’ response in study 1. Chimpanzees may have brought one object because they preferred this object over others and considered it more suitable for exchange. They may also have carried it for its association with a reward they like best, or for its association with the task they prefer. For example, they may have brought a branch because they preferred the dipping honey task over the exchange activity. While testing preferences for tasks appeared too challenging, we assessed in study 2 chimpanzees’ preferences for the objects and the food reward associated with each task. 2.1. Methods All subjects except 1 female (Iris) participated in the study. For the object preference task, chimpanzees were invited to enter their individual compartment and were given a choice between two objects in a series of 12 trials. Five series of 12 trials were run for each individual and for each possible object combination: straw versus branch; straw versus stick; and branch versus straw. The left–right position of each object was semi-randomized across the trials so that they were presented on each side equally often. Objects were presented equally often by a keeper and by the human partner, to prevent that the sight of the human partner or the keeper prompted the chimpanzees to preferentially pick up one object. Pointing fingers or trying to reach through the mesh toward one item or the other was considered as a choice from the chimpanzees. Items were given to the chimpanzee and a new trial was run after 30 s had elapsed. Every three trials choices were rewarded by a small piece of apple, whatever the choice. Food preference tasks were run 2 weeks after the end of this testing. Chimpanzees were individually asked to choose between two different food items presented on a plastic spoon held in each hand of the main investigator. Orientation of mouth or fingers toward a food was considered as a choice from the chimpanzees that was then given the item. Five series of 12 trials were performed for each individual and for each combination: peanut versus orange, peanut versus honey, and orange versus honey. As in Brosnan and de Waal (2004), the individual criterion to indicate preference was set at 80% choice for one of the two objects, or for one of the two offered foods.

2.2. Results In the object choice tasks, four individuals clearly preferred straws over sticks and branches over sticks (Table 2). Three individuals expressed a clear preference either for branches over straws or for straws over branches. In food choice tasks, 7 out of 9 chimpanzees preferred peanuts to oranges. Whereas a strong preference for peanuts over honey was observed in only two individuals, two individuals showed the reverse pattern. The significant preference to bring in straws rather than sticks in study 1, phase 2, may be related to this low interest in sticks and oranges, since these were the less preferred objects and food. At the individual level, however, a preference for objects or food did not strictly match with the objects brought in phase 2 of study 1. Phil, for example, brought more branches in study 1, but did not show a preference for this object compared with other objects. In addition, she appears to dislike honey, the reward associated with branches. Femma brought more straws and sticks than branches, but appeared to prefer branches over other objects. Moreover, she had no clear preference for peanuts or orange over honey. Thomas brought more branches, despite not showing a preference of for a specific object, and his preferred reward was peanuts that were not associated with branches. Three males may show some correspondence between what they brought and what they chose. Ton brought more branches in study 1 and showed a strong preference for branches over sticks, but not over straws. Similarly, he preferred honey over any other food reward. Zorro brought straws and sticks, and showed a clear preference for straws over other objects, but not for sticks. Juus brought straws and 1 stick, which was consistent with a preference for straws over branches as well as a preference for peanut over orange but not over honey. 2.3. Conclusion Three individuals may have brought the objects inside in study 1 because they preferred this object or the food related to this object over other objects or food and were therefore more attracted to these objects. However, preferences could not explain the object choice of the other three individuals. Alternatively, it can be argued that chimpanzees did plan to solve the task, but did not understand that they had to be selective about which particular object to return with. They may have considered all objects equally valid for exchange, despite the fact that they were never trained to give back objects other than straws. Another possibility is that, while they still planned for exchange, the chimpanzees were confused by the choice between the different objects, because they did not

24

V. Dufour, E.H.M. Sterck / Behavioural Processes 79 (2008) 19–27

strictly relate the straw to the exchange task and consequently lacked selectivity. These hypotheses were tested in study 3 and 4. 3. Study 3: testing for the association straw-exchange The lack of selectivity observed in study 1 may result from a weak association between the straw and its function as a token to exchange. The function of the straws used in the exchange task has a rather abstract and symbolic component compared to the sticks and the branches, whose function has some mechanical properties associated with their use. In addition, the function of the straw, as a token, was implemented artificially by training. In the present study, we established whether chimpanzees still connected the straws to the exchange task. 3.1. Methods As in normal exchange sessions, individuals were locked in by the keeper at 13:10. Chimpanzees were given 12 exemplars of each category of object, i.e., straws, branches and sticks. Only then did the partner enter the room facing each individual with 12 peanuts and begging for something back. Only 1 session was given to minimize the possibility of certain types of associative learning. Each straw given was rewarded. Other objects were put aside. The session ended when all the straws had been exchanged. Non-parametric statistics were used to compare the proportion of objects of each categories being given once all straws had been given back. 3.2. Results Four subjects from group 2 selected the straws preferentially over the other objects when requested to give ‘something’ back (Chi-square test: Thomas: d.f. = 2, P < 0.02; Willy: d.f. = 2, P < 0.001; Juus: d.f. = 2, P < 0.01; and Zorro: d.f. = 2, P < 0.05). The other subjects did not show selectivity for the straw (Chi-square test: Phil: d.f. = 2, P > 0.5; Ton: d.f. = 2, P > 0.3; Femma: d.f. = 2, P > 0.3; Peggy: d.f. = 2, P > 0.05; Kenny: d.f. = 2, P > 0.2).

in study 1. However, they were not required to be selective at the time of collection, at 12:00, since only one category of object, the straws, was available. 4.1. Methods Only seven individuals of groups 1 and 2 participated consistently. Two individuals did not complete the testing because of illness (flu, Ton: sessions 1–5; Peggy session 7 group 1) and two individuals never came after the pre-test (Willy and Iris, group 2). Training consisted in re-exposing all individuals with the three different objects and tasks. Both groups were given 12 sessions of the picking and the dipping task in the morning with a keeper and 12 exchange sessions with the human partner, always after noon to make the activity predictable. Testing consisted of a pre-test day followed by 7 testing sessions in which only straws (3 per individual) were available during the collection period. The time of the collection period and the testing period were similar to study 1, and occurred in the group context. As before, the human partner entered the room only once the chimpanzees had entered their individual compartment. 4.2. Results In both groups, the keeper reported a consistent interest for the straws during the collection period. They were all taken back to the main room, and all subjects participated in the collection. During the seven test sessions two individuals brought a straw and solved the task on one occasion: Juus (day 2) and Phil (day 5). Phil, however, came in possession of her straw by accidentally kicking it into the individual compartment while entering before the exchange session. None of the two subjects repeated their success in the following trials. From session 4, individuals slowly started to give up on the task and stopped attempting to give something (hair, remains of salad leaves) when requested “terug” by the human partner: Zorro stopped from session 4, Femma and Ton from session 6, and Phil and Thomas from session 7. Testing stopped after this session.

3.3. Conclusion

4.3. Conclusion

This study shows that four individuals remembered the connection between the straw and the exchange task, among them Juus and Zorro who succeeded at least once in study 1, phase 2. For these two individuals, failure to repeat further success is therefore not related to a loss of the association between the straw and the exchange task. Some individuals, however, picked up objects indistinctively when requested to give something back. This may explain why individuals lacked selectivity when bringing objects in study 1. These chimpanzees may have anticipated that they had to exchange something while in the individual compartment, and since they eventually carried objects with them inside the testing room they sporadically succeeded in study 1. However, having to pick up one of three different kinds of objects added a level of difficulty and may have confused chimpanzees if they did not understand that they had to be selective. One possible consequence is that non-selective chimpanzees stopped planning, since their efforts did not yield a consistent result. This possibility was explored in study 4.

Our results indicate that even with a simplified version of the exchange task where different objects could not be confused, chimpanzees failed to show future-oriented behavior. Failure may partly be related to a progressive loss of motivation for the task. Indeed, in the end only seven out of ten individuals participated consistently. Not receiving the presented reward day after day may have led the chimpanzees to slowly give up on the task. Still, occasional success did not lead to more understanding or success. The two individuals who succeed once could for example have regained some motivation from this success. The lack of a capacity to plan an exchange may be another explanation for our results. Chimpanzees may differ in their planning capacity from bonobos and orangutans (Mulcahy and Call, 2006). Also, planning based on the calculation of what to give, where, when and to whom, may rely on more complex cognitive capacities than planning to use what tool, where and when. This was investigated in part two. 5. Study 5: Chimpanzee planning in a tool-using task

4. Study 4: planning to exchange with no selectivity In study 4, we aimed at testing planning capacities while avoiding the potentially confusing factor when requiring a choice between different objects. Chimpanzees had to plan to bring an object to exchange it later following the same general method as

To get a better insight into chimpanzee future oriented capacities, we tested them with a tool using task. In the previous study that involves cooperative features, chimpanzees failed to solve the planning task. Whether this failure is related to the social dimension sustaining additional complexity “to whom”, or to the specific

V. Dufour, E.H.M. Sterck / Behavioural Processes 79 (2008) 19–27

lack of skills or motivation in cooperative task, as suggested by research from Hare and Tomasello (2004), needs to be investigated further. Field observations suggest that chimpanzees may show some calculation when interacting with others, but observation of future-oriented behavior mainly concerns tool transport and use in this species. In this context, future success does not depend on social interferences, but on the chimpanzee’s own behavior. In the current study we aimed at testing the chimpanzees with a procedure as close as possible to the one used in study 1, except that the task would only provide material benefit from tool use. To do so, we attempted to replicate the Mulcahy and Call (2006) study that showed planning capacities for tool use in orangutans and bonobos. In the first test of Mulcahy and Call, apes had to collect and later bring a tool needed to access food in a test apparatus. Some individuals succeeded in doing so. However, the apparatus was in view of the subject both during the initial collection period and when the individuals were invited back into the test chamber, potentially prompting them in taking the right tool. In their second test, Mulcahy and Call (2006) controlled for prompting and the apparatus remained out of sight both during the collection period and before the testing period. Two individuals of each species succeeded in this new task. However, critics (Suddendorf and Corballis, 2007) argue that even if the task was different and the apparatus not visible, apes could solve it using the same response as the one learned in the first study. Earlier learning and success in the previous paradigm may have facilitated their success in the new one, despite it being a new task. We aimed at replicating their second test, where apes first had to collect a hook, then transport it back 1 h later in the test room to retrieve juice from a bottle placed outside their compartment. The advantage of this procedure over the exchange task is that animals face the task without a human partner, and the solution of the task is not biased by social factors that chimpanzees cannot control. Although our individuals were not na¨ıve in our procedure, since it was very similar to the one used in study 1 and 3, they did not show any learning or understanding of the prior experimental situation, suggesting that any impact of prior exposure to the experimental planning procedure in collecting and bringing a object may have been rather limited, if present at all. 5.1. Methods 5.1.1. Training Prior to the study, the chimpanzees were first trained every day after noon in their individual compartment to use a hook to access a hanging bottle that contained 200 ml of diluted strawberry juice, to be refilled twice by the human partner (V.D.). The human partner had minor role compared with the exchange task, since she would install the bottle, provide the hook and then step back. She would step forward and refill the bottle only once an individual had managed to empty it. Otherwise, she was staying meters away, not looking at the animals. Ten chimpanzees out of ten successfully learned to use the hook in 3 sessions or less. Following this learning phase and for 12 consecutive days, they were given daily sessions of the hook task, where they could retrieve a bottle filled with 200 ml of strawberry juice three times; the apparatus was placed outside the compartment and out of reach without the hook. These sessions always took place at 13:10 to make the hook activity predictable. In the morning, individuals were re-trained by the keepers to use the three other tools, i.e., stick, branch and straw, for 5 sessions each, making all objects familiar before the start of the testing. 5.1.2. Testing Prior to the collecting period at 12:00, the human partner entered the room and quietly sat in a corner to record individual

25

collecting behavior, a procedure that could not be followed in the exchange task since the human partner was strongly connected to this activity. Only then did the keeper enter the room and made the four categories of objects (hooks, straws, branches, sticks: 3 of each kind per individual) available for 10 min. All chimpanzees were free to come and collect the objects during the collecting period that was run in the group context, similarly to study 1 and 4. Testing occurred 1 h later at 13:10; subjects were invited to enter their compartment by the keeper in absence of the human partner or the experimental setting. Only once all chimpanzees had entered the compartment and the door compartment had been closed behind them did the partner (V.D.) come with bottles, hang them out of reach in front of the individual compartments and fill them with strawberry juice before stepping a few meters back as in a normal training session. If chimpanzees had brought the hook and successfully reached the bottle of juice, the partner refilled the bottles for up to two times. Similar to study 1 and 4, the first day was the pre-test day, where individuals could not anticipate that they would need to save and carry the appropriate tool. After the pre-test day, a total of 17 sessions were given to the chimpanzees to match the number of trials given in Mulcahy and Call (2006). 5.2. Results During the collection period in group 1, a mean of 13.6% of the objects was collected (hook: 7.8%, stick: 6.4%, branch: 19.6% and straw: 20.6%). In group 2, a mean of 55% of objects were collected (hook: 58.8%, stick: 56.5%, branch: 52.9%, straw: 51.8%). Thomas from group 2 collected most of the objects and used them in a display back in the main room in 11 out of 17 sessions. None or few of the objects remained in the collection compartment for other individuals to collect after Thomas’ display and most of them were available in the main room at the start of the testing period. No selectivity was observed at the collection period, but Thomas’ collecting behavior induced a strong bias in any attempt to interpret other individuals’ collecting behavior. Interestingly, Thomas, Juus and Willy, repeatedly collected hooks and branches outside the compartment from the first trial to the last one. In comparison in group 1, hooks were collected on 3 trials only (between trials 1 and 9) by Femma, Peggy and Ton. Other object collection persisted until the last trial, but hook collection ceased after trial 9 in that group. During the testing period not all individuals participated consistently. In group 1, Ton refused to come after the second test session. Three individuals consistently came, Femma, Phil and Peggy, but none of these three females ever brought any object with them to the individual compartment and they never solved the task. In group 2, the three older males consistently participated to the task. The three remaining individuals came at different rates (7, 11 and 14 sessions for Iris, Willy and Kenny, respectively). In this group, three individuals solved the task more than once (Table 3). Juus and Thomas brought the hook five times (out of 17) and Willy seven times (out of 11). The hooks were brought in significantly more often than other objects (Chi square test: P < 0.001, hook: n = 17; straw: n = 1; stick: n = 2, branch: n = 3). 5.3. Conclusion Three chimpanzees brought in the appropriate object and therefore anticipated the tool-using task. Since this anticipation was based on the memory on the previous days, and probably not on current motivation as the human partner and the apparatus were not present in the room when individuals were invited to enter the individual compartment, we suggest that these chimpanzees successfully planned for the task. In addition, water was available ad libitum in the main compartment, suggesting that chimpanzees did

26

V. Dufour, E.H.M. Sterck / Behavioural Processes 79 (2008) 19–27

Table 3 Planning for a tool use task: bringing objects inside the individual compartment during the pre-test and the testing period Success

Object brought in

Group 1 Phil Ton Fetnma Peggy

0 0 0 0

0 NT (T2-T17) 0 0

Group 2 Thomas

5

Kenny Zorro Juus

0 0 5

Iris Willy

0 7

T2 (1Hk), T3 (1br), T7 (1Hk), T8 (1Hk), T13 (1Br), T14 (1Sk), T15 (1Hk, 1 Br, 1Sw), T16 (1Hk) NT (T14-T17) 0 T6 (2Hk), T7 (1Hk), T8 (1Hk), T13 (10t), T14 (1Hk), T15 (1Br), T16 (1Hk) NT (T2, T5, T11–17) T6 (1Hk), T7 (1Hk), T8 (1Hk), T10 (10t), T13 (1Hk), T15 (1Hk), T16 (1Hk), T17 (1Hk), NT (T1-3, T5, T9)

Success: a success is qualified as a test session where the individual brought at least one hook. Objects brought in: test session in which objects were brought in (PT: pre-test session, T: test session; number of the test session); number of objects; and quality of objects (Hk: hook; Br: branch, Sw: straw, Sk: stick; Ot: other) per individual and per phase. NT = no testing. Additional comments: The other object brought by Juus (T13) was a sock that he in found in the common room, he attempted without success to reach for the bottle. The other element brought by Willy (T10) was a long and curved piece of plastic taken out of a plastic bucket just before entering the room. She successfully reached for the bottle when using it and the bottle was refilled like during a successful event with a hook (this event is not included in the seven successful trials).

not collect or return with the hook out of a current need for water. The level of success (5–7 repeated successes on 17 or 11 sessions) is comparable to the level of success recorded by Mulcahy and Call (2006) with orangutans and bonobos (Fig. 2). 6. General discussion Following Tulving’s definition, planning in animals can only be demonstrated when several requirements are fulfilled. First, the planning behavior is based on the memory of a previous experience either unique or unreinforced. Second, the current behavior is relevant for the future need only and is not driven by current one, i.e., neither thirst, hunger nor cues given in the experimental context should prompt the animals into producing the appropriate future-oriented behavior. From this perspective, the results of chimpanzees in the tool use task appear to meet the requirements of Tulving’s definition. The chimpanzees anticipated that they would be locked in without access to a tool, anticipated the opportunity to collect juice, remembered previous failures and consequently brought in the hook. It could be argued that the sight of the hook at the collection period may by association have driven the chimpanzees into collecting them out of current desire for juice.

Fig. 2. Illustration of three different apes species’ percentage of success in the planning to use a tool task. Results for bonobos and orangutans are extracted from Mulcahy and Call (2006) published work, experiment 3. Despite some differences in methodology, the responses of these two apes’ species are comparable with the chimpanzees’ percentage of success in our study.

This possibility cannot be ruled out. However, since the hook had always been used in the compartment, collecting it outside should not be the conditioned response to that motivation, unless the chimpanzees anticipated the future need and acted upon it in the present. In addition, subjects had never been trained to transport any of the objects prior to the testing period. Objects were present in the main room for 1 h and had probably lost their attractivity at the time of test. Therefore, transporting them into the individual compartment appeared to be based on a combination of anticipation of what is to come, and previous failure in fulfilling the anticipated need. The best interpretation of our results is therefore that chimpanzees, like bonobos and orangutans, showed the capacity to plan in a tool using task. In addition, we suggest that our test reinforce the results of Mulcahy and Call (2006), since additional controls were performed. In our study, neither the apparatus nor the partner was present in the room when individuals were invited into their individual compartment. Therefore, only the recollection of the previous failure could induce the animals into bringing the hook for the testing period. It could be argued that individuals had learned associative rules from the repeated exposure to the paradigm. However, despite not being na¨ıve in the procedure, subjects remained na¨ıve to how to solve the task, since none of them solved it more than once before. Our findings show that chimpanzees could solve the hook task when no prior associative rules had been learned or reinforced prior to the task, which supports Mulcahy and Call (2006) conclusions. In addition, the four objects that the chimpanzees choose from were equally familiar, all recently used and rather similar in shape or size. All were related to a different task, and had a function of gaining some reward. This guaranteed that when chimpanzees choose the correct object, they had to plan what object to use, for which task and when. Chimpanzees, however, failed to show planning in the exchange task. Chimpanzees are described as exhibiting more “sophisticated cognitive skills” and motivation in a competitive context rather than in cooperative one (Hare and Tomasello, 2004). If the failure to plan in an exchange task reflects a context-driven limitation in chimpanzees’ cognitive skills, this suggests that the capacities of chimpanzees to plan and calculate future benefits of social interactions may have been overestimated. Alternatively, the exchange task may involve some additional level of complexity compared to the tool use task, due to its social-like dimension. Indeed, chimpanzees had to remember what, where, how, and to whom they should give their object. Two accompanying attributes of this social dimension may also have interfered with the understanding of the task. First, the memory concerning the testing period in the exchange and in the tool use task was different. In the first task, chimpanzees may have remembered that they did not succeed because they did not have a straw, but they may also have remembered the ‘unwilling’ character of the human partner since she did not give the peanuts that she held in her hand. The memory about the unwilling partner may have interfered with the memory about how to solve the task (next time bring a straw), thus inhibiting the comprehension of the task. In the hook task, the chimpanzee’s main memory was that they did not succeed for lack of the hook, but the partner, whose role only was to refill the bottle, was not directly involved in failure or success to obtain the treat. Second, the ‘unwilling’ attitude of the partner in the exchange task may have lead to an unexpected after-effect. Chimpanzees may have considered that the partner was of poor quality and therefore not reliable. Melis et al. (2006) have shown that chimpanzees are aware of the quality of their partner and accordingly select the best among two in a cooperative task. The partner’s attributes influence a chimpanzee’s expectancy of success. In an exchange context such as ours, reliability of partners is required. The chimpanzees may have considered this lacking. Still, occasional success in the task, that could have

V. Dufour, E.H.M. Sterck / Behavioural Processes 79 (2008) 19–27

restored “trust” in the partner, did not improve a chimpanzee’s success in the following trials. Whether chimpanzees cannot plan in an exchange task or whether only a long-term trustful partner can illicit planning in this context remains to be established. While the impact of social factors on the capacity to plan for an exchange needs to be investigated further, other non-social factors may also account for the failure to plan in this task. The causal link between the exchange task and the straw was probably more symbolic, and thereby less direct, than the relationship between the hook and the bottle of juice. Indeed, the reason why a hook was needed was related to a physical aspect of the task: “to bring the bottle closer”. Apes have been shown to infer the correct solution to a new task based on its physical characteristics (Call, 2004). The use of the straw as a token had been arbitrarily learned and may be more abstract and/or concept-related, in other words, it required reasoning about abstract representations. Studies using tokens having salient “food-related” properties should be conducted to explore further this aspect of our results. Altogether, great apes are capable of future-oriented behavior on a tool use task (Mulcahy and Call, 2006; this study), but fail to plan in an exchange task. This can be understood to suggest that the capacity to plan is specific to material benefits and was not generalized to a more social-like setting. However, in another study we also employed the exchange task where the reward remained visible to determine how long chimpanzees can wait. Our results show that they can inhibit their impulses for up to 8 min (Dufour et al., 2007), similar to duration of inhibition in children (Mischel et al., 1989). Therefore, we consider it premature to firmly conclude that chimpanzees cannot plan in cooperative tasks with a social dimension. The debate on time travelling in non-humans is still open (Suddendorf and Corballis, 2007) and more evidence is needed to understand the characteristics of future-oriented behavior in nonhumans, their flexibility and their limits (Sterck and Dufour, 2007). Yet experimental work on scrub jays (Dally et al., 2006; Raby et al., 2007) and on apes (Mulcahy and Call, 2006; Dufour et al., 2007; Beran and Evans, 2006; and this study) provide promising data that stress the need for further research to explore animal capacities to understand past and future. Acknowledgements The authors are grateful to Dr. Matthew Bruce for language advices on the manuscript, and to the animal keepers for continuous help and support in the course of the study. References Beran, M.J., Evans, T.A., 2006. Maintenance of delay of gratification by four chimpanzees (Pan troglodytes): the effects of delayed reward visibility, experimenter presence, and extended delay intervals. Behav. Proc. 73, 315–324. Boesch, C., Boesch, H., 1984. Mental map in wild chimpanzees: an analysis of hammer transports for nut cracking. Primates 25, 160–170. Brosnan, S.F., de Waal, F.B.M., 2004. A concept of value during experimental exchange in brown capuchin monkeys, Cebus apella. Fol. Primat. 75, 317–330. Byne, R.W., 2007. Ape society: trading favours. Curr. Biol. 17, R775–R776. Byrne, R.W., Whiten, A., 1988. Machiavellian Intelligence: Social Expertise and the Evolution of Intellect in Monkeys, Apes and Humans. Oxford University Press, Oxford. Call, J., 2004. Inferences about the location of food in the great apes. J. Comp. Psychol. 118, 232–241.

27

Clayton, N.S., Dickinson, A., 1998. Episodic-like memory during cache recovery by scrub jays. Nature 395, 272–278. Clayton, N.S., Yu, K.S., Dickinson, A., 2001. Scrub Jays (Aphelocoma coerulescens) form integrated memories of the multiple features of caching episodes. J. Exp. Psychol.: Anim. Behav. Proc. 27, 17–29. Clayton, N.S., Bussey, T.J., Dickinson, A., 2003a. Can animals recall the past and plan for the future? Nat. Rev. Neurosci. 4, 685–691. Clayton, N.S., Yu, K.S., Dickinson, A., 2003b. Interacting cache memories: Evidence for flexible memory use by Scrub-jays. J. Exp. Psychol.: Anim. Behav. Proc. 29, 14–22. Dally, J.M., Emery, N.J., Clayton, N.S., 2006. Food-caching western scrub-jays keep track of who was watching when. Science 312, 1662–1665. de Waal, F.B.M., 1982. Chimpanzee Politics. Jonathan Cape, London. de Waal, F.B.M., 1997. The chimpanzee’s service economy: food for grooming. Evol. Hum. Behav. 18, 375–386. ´ M., Sterck, E.H.M., Thierry, B., 2007. Chimpanzee anticipation of Dufour, V., Pele, food return: coping with waiting time in an exchange task. J. Comp. Psychol. 121, 145–155. Hare, B., Tomasello, M., 2004. Chimpanzees are more skilful in competitive than in cooperative cognitive tasks. Anim. Behav. 68, 571–581. Hare, B., Melis, A., Woods, V., Hastings, S., Wrangham, R., 2007. Tolerance allows bonobos to outperform chimpanzees in a cooperative task. Cur. Biol. 17, 619– 623. Herrmann, E., Tomasello, M., 2006. Apes’ and children’s understanding of cooperative and competitive motives in a communicative situation. Dev. Sci. 9, 518– 529. Janson, C.H., 1998. Experimental evidence for spatial memory in foraging wild capuchin monkeys, Cebus apella. Anim. Behav. 55, 1229–1243. Jensen, K., Hare, B., Call, J., Tomasello, M., 2006. What’s in it for me? Self-regard precludes altruism and spite in chimpanzees. Proc. R. Soc. B 273, 1013–1021. Melis, A.P., Hare, B., Tomasello, M., 2006. Chimpanzees recruit the best collaborators. Science 311, 1297–1300. Menzel, C.R., Savage-Rumbaugh, S.E., Menzel, E.W., 2002. Bonobo (Pan paniscus) spatial memory and communication in a 20-hectare forest. Int. J. Primat. 3, 601–619. Mischel, W., Shoda, Y., Rodriguez, M.L., 1989. Delay of gratification in children. Science 244, 933–938. Mulcahy, N.J., Call, J., 2006. Apes save tools for future use. Science 312, 1038–1040. Raby, C.R., Alexis, D.M., Dickinson, A., Clayton, N.S., 2007. Planning for the future by western scrub-jays. Nature 445, 919–921. Reader, S.M., Laland, K.N., 2001. Primate innovation, age sex and social rank. Int. J. Primat. 22, 787–805. Reader, S.M., Laland, K.N., 2002. Social intelligence, innovation and enhanced brain size in primates. Proc. Natl. Acad. Sci. 99, 4436–4441. Richards, J.B., Mitchell, S.H., de Wit, H., Seiden, L.S., 1997. Determination of discount functions in rats with an adjusting-amount procedure. J. Exp. Anal. Behav. 67, 353–366. Roberts, W.A., 2002. Are animals stuck in time? Psychol. Bull. 128, 473–489. Schwartz, B.L., Evans, S., 2001. Episodic memory in primates. Am. J. Primat. 55, 71–81. Silk, J.B., Brosnan, S.F., Vonk, J., Henrich, J., Povinelli, D.J., Richardson, A.F., Lambeth, S.P., Mascaro, J., Schapiro, S.J., 2005. Chimpanzees are indifferent to the welfare of other group members. Nature 435, 1357–1359. Silverberg, A., Widholm, J.J., Bresler, D., Fujita, K., Anderson, J.R., 1998. Natural choice in nonhuman primates. J. Exp. Psychol.: Anim. Behav. Proc. 24, 215–228. Sousa, C., Matsuzawa, T., 2001. The use of tokens as rewards and tools by chimpanzees (Pan troglodytes). Anim. Cogn. 4, 213–221. Sterck, E.H.M., Dufour, V., 2007. First test, then judge future-oriented behaviour in animals. Behav. Br. Sci. 30, 333–334. Suddendorf, T., 2006. Enhanced: foresight and evolution of the human mind. Science 312, 1006. Suddendorf, T., Busby, J., 2003a. Mental time travel in animals? Trend Cogn. Sci. 7, 391–396. Suddendorf, T., Busby, J., 2003b. Like it or not? The mental time travel debate. Trend Cogn. Sci. 7, 437–438. Suddendorf, T., Corballis, M.C., 1997. Mental time travel and the evolution of the human mind. Gen. Soc. Gen. Psychol. Monogr. 123, 133–167. Suddendorf, T., Corballis, M.C., 2007. The evolution of foresight: What is mental time travel and is it unique to humans? Behav. Br. Sci. 30, 3 (published online). Tobin, H., Logue, A.W., Chelonis, J.J., Ackerman, K.T., May, J.G., 1996. Self-control in the monkey Macaca fascicularis. Anim. Learn. Behav. 24, 168–174. Tulving, E., 1983. Elements of Episodic Memory. Clarendon Press, Oxford. Tulving, E., 2005. Episodic memory and autonoesis: uniquely human? In: Terrace, H.S., Metcalfe, J. (Eds.), The Missing Link in Cognition: Origins of Self-Reflective Consciousness. Oxford University Press, Oxford, pp. 3–56.