Learning, productivity, and noise: an experimental study of cultural transmission on the Bolivian Altiplano

Abstract

The theory of cultural transmission distinguishes between biased and unbiased social learning. Biases simply mean that social learning is not completely random. The distinction is critical because biases produce effects at the aggregate level that then feed back to influence individual behavior. This study presents an economic experiment designed specifically to see if players use social information in a biased way. The experiment was conducted among a group of subsistence pastoralists in southern Bolivia. Treatments were designed to test for two widely discussed forms of biased social learning: a tendency to imitate success and a tendency to follow the majority. The analysis, based primarily on fitting specific evolutionary models to the data using maximum likelihood, found neither a clear tendency to imitate success nor conformity. Players instead seemed to rely largely on private feedback about their own personal histories of choices and payoffs. Nonetheless, improved performance in one treatment provides evidence for some important but currently unspecified social effect. Given existing experimental work on cultural transmission from other societies, the current study suggests that social learning is potentially conditional and culturally specific.