Despite various social learning conditions, culminating in a full human demonstration of flake manufacture and use through various techniques (e.g., stone throwing/dropping and freehand percussion), none of the four chimpanzees in our sample made or used a flake in any of the conditions. This outcome is in contrast to the positive results of previous studies who demonstrated (and sometimes trained and moulded) the necessary abilities to enculturated apes14,15,16, and to the findings of a recent study in which unenculturated orangutans showed both of the target skills—but not in combination—during play/exploration, in the absence of previous know-how demonstrations18.
Although we did not know the exact level of enculturation of all subjects, it is likely that one individual (Colin) was not enculturated (or, following the stricter definitions of enculturation, was minimally enculturated by virtue of being in captivity). Therefore, we paid particular attention to this individual. However, the fact that all four chimpanzees failed to show the target behaviours after demonstrations suggests that higher levels of enculturation than those possessed by the subjects in this study are required for chimpanzees to copy human knapping and flake-use demonstrations and/or to enable them to innovate these behaviours. However, another possibility is that the chimpanzees in this study may not have had enough time to develop the behaviour, which perhaps may have emerged instead in long-term tests37. Indeed, wild chimpanzees often take years to acquire some of their skills, such as nut-cracking6,7. However, given the baseline low motivation levels of the chimpanzees (see below), it is unlikely that even longer testing periods would yield different results (see also38, for an overview of how long innovations can be expected to take in apes). Overall, it is therefore possible that some of the chimpanzees were sufficiently enculturated, but may have required longer periods of demonstrations to develop these behaviours. Note that both the enculturated bonobos and the orangutan who previously showed both target behaviours14,15,16 received longer periods of training, and potentially of demonstrations as well (although the exact number of hours of demonstrations and/or training is unclear from the original papers).
As mentioned, three of our subjects have unclear rearing histories. Therefore, it is also conceivable that some (or all) may have been deprived, instead of being enculturated, by living in ‘unnatural conditions’. Deprivation can have the opposite effect of enculturation on cognition, hindering an individual’s ability to innovate novel behaviours39—and potentially also leading to reduced social learning abilities. This, too, could explain the negative findings of this study for most of the subjects.
Future studies could try to better assess the enculturation levels of subjects with unknown backgrounds by using recently proposed methods to experimentally investigate enculturation (e.g., tracking the chimpanzees’ attention and actions compared to those of models performing the behaviour40; although see below for a discussion on human versus conspecific models). Unfortunately, we did not have the resources to carry-out these additional tests for the current study. In general, running the types of studies such as the one described here are very time-intensive and logistically difficult (which might also explain why all previous studies had very small sample sizes). Therefore, often compromises must be made, such as including chimpanzees in the sample that have mixed rearing backgrounds.
It is also possible that there is a species effect for these behaviours, in which great apes other than chimpanzees are capable of flake manufacture and use. Indeed, orangutans seem able to develop two main precursor behavioural types (flake production and their use as tools18). As only four chimpanzees were tested here, it is also possible that our negative findings were due to small sample size. However, previous studies on apes had smaller sample sizes than ours, but were successful (see above). Nevertheless, ideally, future studies should test larger groups of chimpanzees, including (ethical considerations permitting) fully enculturated and unenculturated subjects, controlling for the factors outlined above. If the hypothesis that enculturation is a key factor is correct, then the former should produce target behaviours at much higher rates than the latter.
Another explanation for our negative findings could be carry-over effects from the baseline test19 that we ran with these subjects prior to the current study. Namely, if present, these order effects may have resulted in the chimpanzees losing sufficient motivation for the task to benefit from the demonstrations provided here. To test for this possibility, unenculturated naive chimpanzees should be tested elsewhere again, starting directly with social learning conditions—perhaps even immediately with full demonstrations of the target behaviours. The chimpanzees in this study seemed to show a numerically low baseline level of motivation to interact with the provided materials (however there was no statistically significant decline between the interaction rates in our previous study19 and the conditions implemented in the current one). Although these chimpanzees receive regular enrichment tasks prepared by volunteers working at the sanctuary, they have not participated in research tasks like the one used in this study. Therefore, these particular subjects are relatively unfamiliar with puzzle boxes. This lack of experience may be an explanation for the chimpanzees’ low motivation to spend extended periods manipulating the testing materials we provided them. It could also be that the chimpanzees were not motivated to make cutting tools because they were not dependent on the food inside the box, as they receive a regular and varied diet from their keepers. However, according to the “the captivity effect” hypothesis, captive animals are more likely to innovate behaviours compared to their wild counterparts41, rendering this explanation unlikely. This account also does not explain why chimpanzees engage in other tasks in similar situations38,42 or why other captive ape studies—none of which deprived the apes of food—reported the development of the target abilities.
Despite insufficient levels of motivation being a potential explanation, the subjects did spend time manipulating the puzzle box and the materials. The subjects opened the puzzle box and consumed the food inside it when they could do so without tools (e.g., in the loose rope condition). Therefore, the chimpanzees clearly had some motivation to engage in the task38. Indeed, various types of manipulation were observed over the course of the study. This suggests that whilst the task was perhaps too opaque (at least for the subjects tested here) and may have limited the chimpanzees’ motivation to interact with it for extended periods, it was not ignored by the subjects. For example, one of the chimpanzees attempted to use a core on the rope in one of the earlier social learning conditions, suggesting that they had understood that the rope was an important component of the box, and perhaps even that it had to be removed to retrieve the reward. Note that the apparatus used in this study was modelled on the ones successfully used in previous similar studies14,15,16. However, future studies could provide testing apparatuses designed and tested to be even less complex and/or opaque to examine whether a simpler, or more intuitive, set-up encourages further manipulation of the materials (thus hopefully increasing the likelihood of innovation of the target behaviour).
Some of the methods to make flakes would have required the chimpanzees to manipulate and coordinate two tools at the same time (stone core and hammerstones) using one or both hands. Bimanual manipulation like this has been found to be rare in chimpanzees, indeed most of their natural behaviours require the use of only one hand at a time43 (although see also44,45). In a study on the manipulative abilities of captive chimpanzees and bonobos, it was reported that chimpanzees preferred to use only one hand, even when manipulating more than one object46. The captive bonobos tested in the same study instead showed a higher propensity towards bimanual activities, perhaps due to the fact that bonobos travel more arboreally than chimpanzees in the wild, therefore requiring the use of both hands46. In this regard, the difference in preference for unimanual and bimanual activities between species may provide yet another alternative explanation for the bonobos’ success in earlier studies and the failure of the chimpanzees in the current study. However, bimanual techniques were not the only method available to the chimpanzees to make flakes, for example, they could have used the dropping/throwing technique used by Kanzi15 and some modern humans47 (and potentially early hominins too48) which can be carried-out with one or two hands.
Another explanation for the negative results of this study could be that the chimpanzees were all too old, or more generally out of a potential sensitive learning periods, for the development of this behaviour (e.g., see49). Although the experience of the chimpanzees we tested in relation to stone tools is unknown (as we cannot fully account for some of the chimpanzees’ experience before arriving at Chimfunshi), it is likely that they were not familiar with stones as tools in general. Previous studies have found that exposure to stones as tools during a specific sensitive learning period has a positive effect on the development of stone tool-use behaviour later on in life across primates49,50. Furthermore, the development of bimanual manipulation (though see also above) has been observed to take several years of practice (and/or development) in chimpanzees (e.g.,51. One of the few behaviours that often involves the use of two hands and multiple objects in the wild is nut-cracking52. When studying the development of this behaviour in juvenile chimpanzees, Inoue-Nakamura & Matsuzawa51 found that juvenile chimpanzees were initially unable to carry out the bimanual manipulations required to crack nuts, or to correctly implement the various steps needed for the behaviour. It takes approximately 3-5 years of development and/or practice with the objects for wild chimpanzees to acquire these manipulative abilities51. Therefore, if the chimpanzees in this study were not exposed to stones during sensitive stages of their development, they may not have been able to recognize the provided stones as potential tools even after demonstrations of their properties, or to develop the bimanual manipulation abilities required for some of the knapping techniques (although, as mentioned above, only one hand is required for some techniques, such as Kanzi’s throwing/dropping technique15).
Lastly, the chimpanzees may not have recognized the human demonstrator as a salient enough model to copy but might have attended to a conspecific model showing the target behaviour instead (if such a model had been implemented). Data from an eye-tracking study with chimpanzees and human demonstrators found that chimpanzees paid more attention to conspecific social cues than to human ones, suggesting that chimpanzees may extract more referential information from conspecifics than from humans53. However, these studies do not suggest that chimpanzees will not attend human models at all (indeed, in other tasks, humans proved to be valid models; e.g.,54 and the previously tested apes learnt from humans, although this may have also been due to their enculturated status). As none of the chimpanzees showed any behaviours even approximating the target abilities, we could not explore this possibility by providing a conspecific model, but this would be an interesting avenue for future studies, perhaps using clicker training (and/or, where possible, moulding) to prepare demonstrators (see however55 in which chimpanzees failed to copy conspecific behaviours).
In conclusion, all the above are alternative explanations for the negative findings of this study. This does not mean, however, that all of these are equally likely, or mutually exclusive. Given the available data from flake production and use studies in apes, it is likely that alongside some of the explanations mentioned above, it is possible that as in the case of previously tested bonobos15,16 and one orangutan14, human training and/or demonstrations are required for the full flake manufacture and use behaviour to emerge. In addition however, there must also be a sufficiently high level of human enculturation before great apes are able to develop this behaviour or to copy these from humans. Indeed, these higher levels of enculturation are known to increase both the social and the individual learning abilities of affected apes, equipping them with the ability to acquire behaviours outside of the realm of possibilities of their unenculturated conspecifics.