Dissertation

Empirical signatures of universality, hierarchy and clustering in culture

In this thesis, "culture" refers to the collection of subjective human traits, such as preferences an opinions, that a given, geographically bounded population has at a given moment in time.

Author: Babeanu, A.I.
Date: 24 October 2018
Links: Thesis in Leiden Repository

In this thesis, "culture" refers to the collection of subjective human traits, such as preferences an opinions, that a given, geographically bounded population has at a given moment in time. Representative samples of individuals from such populations are studied, focusing on individual opinions expressed on various topics, present in multivariate empirical data that had been previously collected, mainly via social surveys. We propose and exploit new methods for analyzing such data, relying on mathematical notions specific to statistical mechanics and information theory, but also on agent-based models/simulations of opinion/cultural dynamics driven by social influence. These methods provide new insights about how human culture is organized. They provide indications that cultural structure has universal properties, independent of the geographical region and of the set of survey questions. Furthermore, these properties suggest that culture is shaped around a small number of "rationalities", while also having a certain hierarchical organization that is robust to social influence dynamics. Finally, we propose a method of filtering the noise in the data, which seems to allow for the identification of cultural modules that are not visible otherwise. However, we also show that visible modules may well be just artifacts of survey design.