Pesendorfer et al.’s paper represents an important contribution for a number of reasons. These are both methodological and substantive. Methodologically, the authors deserve congratulations for being the first to successfully complete a field experiment that introduces two alternative techniques, applied to the same ‘artificial fruit’, into different groups of primates and tracks subsequent behaviour. As the authors recognise, this creates much-needed opportunities to experimentally study the scope of traditions in the wild, correcting the current imbalance reflected by an existing corpus of such experiments with captive primates (their Ref. 18). Recent success in a study of this kind with wild meerkats [1] reinforces this optimism. However, it is important to recognise that a field experiment on primate traditions still needs to be done. At one point (p.3), Pesendorfer et al. say their study showed “traditions were successfully established”, but in justifying this by citing the authoritative definition of traditions owing to Fragaszy and Perry (their Ref. 5), they misquote the latter in a small but significant way. A crucial part of Fragaszy and Perry’s definition is “depends in part on socially aided learning for its generation in new practitioners” (p. 12). They went on to underline this by saying that “A particular behaviour cannot be identified as a tradition without inferring that socially aided learning supports its shared presence across individuals” (p. 12). Pesendorfer et al. subvert the definition by inserting ‘may’ - “may acquire in part by socially aided learning” (p. 1) but as Fragaszy and Perry emphasise (and as is routine in the study of traditions generally), a role for social learning is essential. The fact that Pesendorfer explicitly aimed instead to train their animals non-socially means that they should not have spoken of ‘traditions’ in this context.
Nevertheless, the results reported by Pesendorfer are important in showing how habit, rather than processes of conformity (or indeed social learning in general) can sustain behavioural differences between groups. However, the authors contrast this conclusion with that obtained in our earlier work (their Ref. 17) and it is this conclusion we particularly take issue with here.
The first point is that while Pesendorfer et al. are correct to say – indeed, they show – that habit can maintain behavioural differences of the kind they describe, in our study we inferred conformity not from the mere maintenance of a stable state, but because after individuals discovered the alternative technique to that common in their group, they later shifted their behavioural profiles back to converge on the latter (thus, ‘conformed to the group norm’) to a statistically significant extent. Conceptually, this is a crucial difference.
However, we agree that the size of the effect was small, as Pesendorfer et al. correctly note. Our study was designed principally to test whether chimpanzees can sustain alternative traditions, for which we obtained strong experimental evidence: the discovery of the conformity effect we described was by contrast unexpected and a post-hoc observation, which remains to be experimentally tested in its own right. Of course, the fact that Pesendorfer et al.’s marmosets did not show such an effect by no means implies that chimpanzees would not, or could not.
We also agree with the authors that the study by Galef and Whiskin [their Ref. 27], while making another important contribution to this field, does not correspond with some common conceptions of conformity because tests did not extend to a group context. What this underlines is that more than one definition of ‘conformity’ exists in the literature, as we recognized in [Ref. 17]. Commonly, conformity is identified with what Laland [2] described as ‘copying the majority’ and clearly [Ref 27] does not match this. In our own study [Ref. 17] we did appeal to this conception, taking the ‘norm’ as the commonest of the two techniques occurring in each group, but in addition we required that the subject had discovered an alternative technique to the group norm, implying that their conformity is all the stronger. In [Ref. 27], however, the latter condition applied (rats switched from the habit they had acquired to that of a model they watched), but (in Experiment 2) there was only a single model, so one could not sensibly refer to any ‘group norm’. That rats switched from the habit they learned themselves to a different preference they witnessed in a single other rat, therefore seems rather surprising and its functional significance remains unclear.
References
1. Thornton A, Malapert A (2009) The rise and fall of an arbitrary tradition: an experiment with wild meerkats. Proc.R. Soc. ‘B’ in press. Doi:10.1098/rspd.2008.1794
2. Laland KN (2004) Social learning strategies. Learn. Behav., 32: 4–14.

Andrew Whiten1, Victoria Horner2 and Frans de Waal2
1. Centre for Social Learning and Cognitive Evolution, and Scottish Primate Research Group, School of Psychology, University of St Andrews, St Andrews, KY16 9JU, UK. aw2@st-andrews.ac.uk
2. Living Links Center, Yerkes National Primate Research Center, Emory University, Atlanta, GA 30329, USA.