Confirmation Bias and Convergence of Beliefs: an Agent-Based Model Approach

Abstract

We simulated societal opinion dynamics when there is confirmation bias in information gathering and spread. If decision making is influenced by confirmation bias, an agent puts more weight on positive information to confirm hypothesis or reservation in the learning process, which renders selectivity in information gathering. If the utility discovered post-purchase is low, it is externalised rather than internalised (i.e. self-blame) for the selectivity of information. This causes the agent to outweigh the negative information. These two mechanisms are simulated to investigate societal opinion dynamics and explain behavioural patterns such as over-confidence and convergence of opinions.