Bugs and features. A reply to Smaldino et al. (2017)

A conformity bias is, in cultural evolutionary terms, an individual-level tendency to disproportionately copy the majority. Collectively, at population-level, this individual behaviour results in a characteristic sigmoidal relationship between the frequency of the behaviour in the population and the probability to copy it (“the sigmoid”). In two recent papers (Acerbi et al. 2016, van Leeuwen et al. 2016), we, among other things, presented a series of simple models showing that, in some conditions, populationlevel outcomes produced without a conformist bias could also generate a sigmoid (“generate a sigmoid” here means that, when fitting the data both with a linear and a sigmoid model, the latter provides a better fit). We did not claim that the sigmoid is never evidence for an individual-level conformist bias, but that empiricists might want to pay attention to the possibility that mechanisms other than conformity can explain its presence. More generally, we invited to be cautious when inferring individual-level behaviours from population-level patterns. In particular, we identified three scenarios in which the sigmoid could be generated in the absence of conformity bias: