Impact of Confirmation Bias on Group Dynamics

Human behaviour is complex and driven by many internal and external factors. Predicting an individual’s behaviour may require modelling complicated dynamics in order to capture the many facets that influence a human. However, in this paper we show that even simple dynamics can lead to complex, interesting results. The addition of confirmation bias into a standard behavioural change model gives rise to dynamics of polarisation and clustering. Additionally, the simulation of these confirmation bias effects in a classroom setting displays unforeseen responses. These simulations show that having low bias can be beneficial if an instructor is trying to sway beliefs about behaviours, such as note taking. However, the simulations also show that low biases make it difficult to ensure that students stick to these changes once the instructor is no longer able to encourage better note taking when a school term has ended. Students that now take notes because of the encouragement of their instructor may stop taking notes when that encouragement is no longer there. These results display compelling evidence of the usefulness of simple modelling techniques and pose intriguing ideas for future behavioural change intervention techniques.