Biases in cultural transmission shape the turnover of popular traits

Abstract

The neutral model of cultural evolution, which assumes that copying is unbiased, provides precise predictions regarding frequency distributions of traits and the turnover within a popularity-ranked list. Here we study turnover in ranked lists and identify where the turnover departs from neutral model predictions to detect transmission biases in three different domains: color terms usage in English language 20th century books, popularity of early (1880–1930) and recent (1960–2010) USA baby names, and musical preferences of users of the Web site Last.fm. To help characterize the type of transmission bias, we modify the neutral model to include a content-based bias and two context-based biases (conformity and anti-conformity). How these modified models match real data helps us to infer, from population scale observations, when cultural transmission is biased, and, to some extent, what kind of biases are operating at individual level.