A novel framework to classify opinion dynamics of mobile agents under the bounded confidence model

Abstract

The formation and evolution of public opinion have been widely studied to understand how consensus forms due to atomic interactions between individuals. While many studies have paid attention to modelling influence and interaction, most of the literature assumes static agents, ignoring the frequent changes in physical locations expected in real life. This feature naturally allows humans to interact with diverse people and avoid disagreement, which heavily impacts the co-evolution of opinions, communities or isolation in human societies. Our previous work proposed an extension of the bounded confidence model inspired by the theories of homophily and cognitive dissonance, which concern humans’ natural behaviours of attraction and disagreement. Although this demonstrated a marked difference to a static opinion model and purely random mobility, the limited experiments gave little insight into the causes or the resulting structures of consensus. This article addresses these shortcomings through a thorough investigation of the impact of mobility modelled by different mechanisms. Through extensive simulation, we observe a transition from multiple stable opinion clusters to complete consensus and a shift from a geographically based organisation to isolated structure-less agents. Lastly, we propose a novel classification of the different outcomes of self-organisation in opinion models, highlighting the patterns of emerging behaviours across the spectrum of interaction range and influence parameters.