Electronic Supplementary Material (Whiten et al.)
1. Detailed comparative comments on studies
(a) Tennie et al. (2009).
In the otherwise excellent review paper that is meant to be a ‘(sparring?) partner’ to our own, and indeed in numerous respects complements it very well, Tennie et al. make a number of points that we wish to take issue with.
(i) Tennie et al. [henceforth TCT] comment on an absence of evidence of cultural traditions in bonobos, a state of affairs not consistent with our inference that the shared ancestor of humans and chimpanzees (and thus also bonobos) had a rich cultural repertoire. On this point TCT neglect a prior publication by Hohman and Fruth (2003) specifically on this topic. It is true that little putative cultural behaviour has been recorded for bonobos, but TCT are correct that the lives of bonobos remain relatively little known, and are in any case restricted to a small area of central Africa, unlike the common chimpanzees that have revealed so much inter-population variation in behaviour. What is more relevant is that cultural variations defined by multiple differences in traditions of different types (social, foraging etc) have been documented in another great ape, the orangutan (van Schaik et al., 2003, as cited by TCT), from which van Schaik et al. and Whiten and van Schaik (2007) drew the parsimonious inference that this complexity, distinctively shared by three great apes (orangutans, chimpanzees and humans) derived from their common ape ancestor of approximately 14 Myr ago. Gorillas and bonobos may thus have lost some of this richness, or it may be that they have not yet been sufficiently investigated from this perspective.
References:
Hohmann, G. and Fruth, B. 2003 Culture in bonobos? Between-species and within-species variation in behaviour. Curr. Anthropol.44, 563-71.
Whiten, A. & van Schaik, C. P 2007 The evolution of animal ‘cultures’ and social intelligence. Phil. Trans. R. Soc. ‘B’, 362, 603-620.
(ii) TCT refer to the Whiten et al. (1999) study rather disparagingly as being based on ‘discussions’ among the co-authors. As the original papers made clear, however, (Whiten et al. 1999, 2001), the analysis was based on a systematic, two-step procedure that involved all research groups in each step, and in the second step required coauthors and their colleagues to systematically code each of 65 consensually defined behaviour categories for patterns of occurrence at their study site.
(iii) TCT describe the study of Humle and Matsuzawa (2002) as ‘the most systematic study to date’ of ant-dipping. As TCT underline, that study did indeed show that environmental factors, notably the species of ant and the aggressivity of their bite at nest hole versus trail, influenced chimpanzees’s choice of tool and dipping technique at Bossou, casting some doubt on the cultural basis of similar regional variations in chimpanzees’ ant-dipping. However this study cannot be described as the most systematic to date, given the follow-up studies by Humle and other colleagues (Shöning et al., 2008; Möbius et al. 2008) that have included numerous careful measures and experiments concerning ants, ant behaviour and chimpanzee ant-dipping across 14 sites in eastern, western and central Africa. TCT offer only a selective quotation from the Shöning et al. paper. Thankfully TCT complete the quotation in their Endnotes, which converges on a different conclusion to that implied in the part quoted in their main text, yet TCT contrive to interpret Shöning et al. as concluding that the variations are “not based on sociocultural processes” (Endnote 2).
In fact both of these admirably careful and extensive follow up studies offer conclusions different to the portrayal of TCT. Shöning et al. say that “although the origin and evolution of army-ant-eating are obviously difficult to elucidate, our data contradict both genetic and environmental deterministic hypotheses for the observed variation” (p. 57). They conclude by saying that “In summary, we have examined alternative environmental hypotheses for across-site variation in army-ant-dipping by chimpanzees. Our results show that the geographic pattern of presence/absence is not caused by availability of suitable prey species. Current data on the availability of alternative animal prey likewise seem not to explain this pattern. The similarities and differences in tool use, tool length, and technique are driven only partly by the identity and characteristics of the prey species. Our data do not show that variation in army-ant-eating is sociocultural, but our findings are consistent with the interpretation that army-ant-eating by chimpanzees varies culturally” (p. 57). On the basis of a very detailed numerical and experimental study focused on the two key sites of Bossou and Tai, Möbius et al. are led to say that “We hence suggest that cultural differences based on social learning remain the most parsimonious explanations for the observed differences in tool length when dipping for intermediate species at nests and for the variation in predation strategies on epigaeic species between the two populations” (p. 44). Echoing Shöning et al., they conclude that there is a “complex interplay between ecological and cultural influences in army ant predation by wild chimpanzees”.
The field research has therefore built up from broad-brush surveys like that of Whiten et al. (1999), through more focused studies such as that of Humle and Matsuzawa (2002), to those of Shöning et al. and Möbius et al., which truly represent “the most systematic studies to date”, achieving sophisticated results and conclusions.
(iv) TCT describe results of the experiments on leaf-swallowing behaviour by Huffman and Hirata (2004) as “undermining the social transmission hypothesis”. Yet this is another case where the data were consistent with a role for each of individual discovery (in some chimpanzees) and social learning (by others). Huffman and Hirata concluded their abstract by saying that “This study provides support for the hypothesis that leaf swallowing originated in the wild from opportunistic feeding behavior and was later passed down in the form of a self-medicative behavioral tradition.” (p. 113).
(v) TCT worry that the indications of multiple origins reduce the likelihood that regional variations among chimpanzees are culturally based, as in the case of nut-cracking. However, in west Africa where multiple sites have been monitored, the distribution of nut-cracking appears continuous over several hundred kilometres; and while a second ‘centre’ has been discovered in Cameroon as TCT note, nut-cracking has been documented as absent across large areas where nuts are available. This distribution fits a model with as few as two origin points, consistent with it being a rare chimpanzee invention, followed by cultural diffusion across large swathes of west Africa (the distribution around the Cameroon site remains to be mapped). Note, in any case, that the existence of multiple origins in human cultures, such as the pyramids of the Egyptians and the Maya, undermines neither the general conclusion that our species display cultures, nor that these are examples of cultural transmission, once invented.
(vi) TCT argue that chimpanzees in our pan-pipes experiments “did not learn anything radically new or improbable”. Yet as we note in our paper, and as in TCT’s ‘loop’ experiment, we ran no-model control tests precisely to discover what was improbable for naïve chimpanzees. As many as 56 chimpanzees have been tested without benefit of a model, and of these, only one was seen to invent the ‘poke’ technique: none invented ‘lift’. Unlike TCT’s loop technique, however, the ‘lift’ technique spread by social learning in our study – probably because we designed it appropriately for chimpanzees, to lie in our Zone 2 (Fig. 3). This appears to be what TCT’s ZLS hypothesis predicts should not be possible in chimpanzees (although it is difficult to be sure, pending a more formal definition of the hypothesis). Other of our experiments, notably those described in Whiten et al. (2007) likewise showed social transmission of alternative techniques that no-model control individuals failed to achieve. Moreover, given that TCT make a distinction between ‘copying products’ versus ‘copying processes’ it remains unclear why chimpanzees replicating, say, the pan-pipes ‘poke’ technique versus ‘lift’ technique would be seen as copying a ‘product’ rather than copying one of the two different ‘processes’ that lead to gaining food.
(vii) TCT worry that the occasional rewards that the observer chimpanzees in our ghost studies obtained may have distracted them. However, these intermittent rewards were provided so as to mimic as exactly as possible what occurs in the normal social situation, through occasional scrounging, where presumably they are equally ‘distracting’, or not. In fact these intermittent rewards likely helped ensure that observer chimpanzees did watch the ghost displays and Hopper et al. (2007) provided data showing that they did indeed attend.
(viii) TCT express a further, related concern about our ghost experiments results: that ‘it is not at all clear that chimpanzees attend to a moving box in the same way they attend to the results of other’ actions. If this were true, then it would in any case shift the sense of ‘emulation’ away from the stark form that we quote from Tomasello (1998) in which emulation is compared to the effect of ‘wind’ rolling a log (a true ‘ghost’ scenario), and closer to copying what others chimpanzees ‘do’ – as we emphasised in our paper in citing Thorndike. We note that Tennie et al. (2006) themselves embraced the logic of the ghost experiment, which by its very nature equates to the Tomasello impersonal ‘wind’ scenario.
(b) Lyons et al. 2007.
Lyons et al. (2007) incorrectly impute to Horner and Whiten (2005) the hypothesis that children copy to satisfy a ‘social demand’ (as is indeed proposed by others such as Nielsen, 2006), because Horner and Whiten suggested that part of the explanation may be that children see the adult’s actions as intentional, given they repeatedly poke the tool in the top of the puzzle box, and this over-rides the bizarreness of the actions. In fact we argued against a ‘social function’ explanation because we found children performed their copy even if the adult had quit the scene. Instead, the interpretation of Lyons et al., presented as novel and contrasting with our own, appears essentially similar, in proposing that children “may automatically (and potentially erroneously) encode all the adult’s purposeful actions as causally necessary” (p. 19751: italics added here). Likewise, Lyons et al. interpret our hypothesis that over-copying occurs because a normally functional copying motivation ‘remains habitual’ in inappropriate contexts as suggesting over-copying is just a learned ‘habit’, and therefore reversible. In fact our usage of ‘habitual’ in this context is closer to being a synonym of the automaticity referred to by Lyons et al. In sum, we see our theoretical perspective on our results as highly similar to that of Lyons et al. What we see as their important original contributions are summarized in section 6, and the greater contrast is perhaps between the results of Lyons and ourselves, and those of other researchers summarized in the final paragraph of that section.