Anthropologists have long argued that fear of victimization through witchcraft accusations promotes cooperation in small-scale societies1. Others have argued that witchcraft beliefs undermine trust and therefore reduce social cohesion2. However, there are very few, if any, quantified empirical examples demonstrating how witchcraft labels can structure cooperation in real human communities. Here we show a case from a farming community in China where people labelled zhu were thought capable of supernatural activity, particularly poisoning food. The label was usually applied to adult women heads of household and often inherited down the female line. We found that those in zhu households were less likely to give or receive gifts or farm help to or from non-zhu households; nor did they have sexual partnerships or children with those in non-zhu households. However, those in zhu households did preferentially help and reproduce with each other. Although the tag is common knowledge to other villagers and used in cooperative and reproductive partner choice, we found no evidence that this assortment was based on cooperativeness or quality. We favour the explanation that stigmatization originally arose as a mechanism to harm female competitors. Once established, fear that the trait is transmissible may help explain the persistence of this deep-rooted cultural belief.
R.M., M.G.T., T.J. and J.W. were all funded by the European Research Council grant to R.M. (AdG 249347). R.M. and J.W. were also funded by Lanzhou University and the 1000 Talent plan of China, T.J. by the National Natural Science Foundation (NSFC no. 31470453), Q.Q.H. by the China Postdoctoral Science Foundation (nos. 2013M541036 and 2014T70122) and the NSFC (no. 31600305), and Y.T. by the NSFC (nos. 31270439 and 11471311). All authors were also funded by an International Partnership grant from the British Academy (IPM 120180) to R.M. and T.J. The funders had no roles in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors thank the many people who participated in this research and those who gave assistance with the demographic surveys and/or games experiments. The authors thank H. Zhang and others at HEEG Lanzhou, S. Peacey and others at HEEG UCL and M. Dyble, M. Hoffman and others at IAST, for discussions.
Supplementary Note, Supplementary Methods, Supplementary Figures 13, Supplementary Tables 1–8, Supplementary References 1–8
Witch: a tag that shapes social networks. A video abstract explaining the context and content of the research. An incorrect version of this video was originally published owing to an administrative error by the publisher. The correct version has now been uploaded.