Abstract

We present a discrete-time, non-Markovian, agent-based model of opinion dynamics, which enhances theoretical frameworks and corroborates experimental observations. A key aspect of the model’s novelty lies in a memory-dependent measure of pairwise affinity, and it is found that if interactions are severed whenever the affinity falls below some threshold, segregated opinion clusters tend to emerge. Clusters are found to undergo ‘opinion drift’ under certain conditions, leading to collective extremisation of the population. Oscillatory clusters are proven to exist and verified numerically. By carefully interpreting these results, we posit explanations of socio-psychological phenomena such as emergent cooperation and group polarisation.