Revealing patterns of cultural transmission from frequency data: equilibrium and non-equilibrium assumptions.

Enrico R Crema,Stephen Shennan,Anne Kandler

doi:10.1038/srep39122

Abstract

A long tradition of cultural evolutionary studies has developed a rich repertoire of mathematical models of social learning. Early studies have laid the foundation of more recent endeavours to infer patterns of cultural transmission from observed frequencies of a variety of cultural data, from decorative motifs on potsherds to baby names and musical preferences. While this wide range of applications provides an opportunity for the development of generalisable analytical workflows, archaeological data present new questions and challenges that require further methodological and theoretical discussion. Here we examine the decorative motifs of Neolithic pottery from an archaeological assemblage in Western Germany, and argue that the widely used (and relatively undiscussed) assumption that observed frequencies are the result of a system in equilibrium conditions is unwarranted, and can lead to incorrect conclusions. We analyse our data with a simulation-based inferential framework that can overcome some of the intrinsic limitations in archaeological data, as well as handle both equilibrium conditions and instances where the mode of cultural transmission is time-variant. Results suggest that none of the models examined can produce the observed pattern under equilibrium conditions, and suggest. instead temporal shifts in the patterns of cultural transmission.

Highlights

A long tradition of cultural evolutionary studies has developed a rich repertoire of mathematical models of social learning
Our focus is on historical case studies and we assume the existence of records of changes in the frequencies of cultural variants over time, but note that 1) observed frequencies only describe the composition of a sample and not of the whole population of cultural variants; 2) the frequencies do not represent instantaneous distributions but rather an accumulation over a certain time period; and 3)
In the equilibrium version of the model, the median posterior density of b was 0.028 which would suggest the presence of some degree of weak anti-conformist bias, albeit the 95% highest posterior density interval (HPDI) covers a range between −0.005 and 0.102

Summary

Introduction

A long tradition of cultural evolutionary studies has developed a rich repertoire of mathematical models of social learning. Studies have laid the foundation of more recent endeavours to infer patterns of cultural transmission from observed frequencies of a variety of cultural data, from decorative motifs on potsherds to baby names and musical preferences While this wide range of applications provides an opportunity for the development of generalisable analytical workflows, archaeological data present new questions and challenges that require further methodological and theoretical discussion. Attempts have been made to develop non-equilbrium models of cultural evolution[13] and to couple those models with Bayesian inference techniques[14,15,16] The foundation of this inference framework consists of formulating cultural hypotheses as mathematical models, and deriving those hypotheses that can produce theoretical data similar to observed data using approximate Bayesian computation This set-up allows for an exploration of the effects of equilibrium and non-equilibrium assumptions on the inference of underlying cultural processes

Objectives

Methods

Results

Conclusion