École Nationale Supérieure de
Techniques Avancées

Abstract

The coming work aims at giving a R-package for modeling cultural change. The code implements the generative inference framework developped in Kandler and Shennan (2015). Cultural change can be quantified by temporal changes in frequency of different cultural variants. The goal of this framework is to identify which underlying cultural transmission processes could have produced the data. The framework models the temporal changes in frequencies of cultural variants between two time points t1 and t2 conditionned on a specific hypothethis of cultural transmission. Two hypotheses are currently implemented : neutral evolution and frequency-dependant evolution. In order to determine these hypotheses, some parameters need to be infered. We use a statistical inference procedure in form of approximate Bayesian computation (ABC). The code is then used to study the impact of population size uncertainty on the final hypothethis we choose. In most of the cases, all we know are two samples, one for t1 and one for t2, without any information on the population size between t1 and t2. The question is to know if we are still able to get infomation on the transmission processes with this lack of information.