Evolution of Opinions on Social Networks

A fundamental aspect of society is the exchange and discussion of opinions between individuals, occurring in situations as varied as company boardrooms, elementary school classrooms and online social media. Opinion dynamics is the development and analysis of mathematical models that describe and capture how individuals' opinions evolve as they interact with each other in a social network. It is part of the wider research area of social network analysis, which is a rich and diverse area of interdisciplinary research that has been tackled by many different scientific communities. A fundamental challenge is to develop models which are grounded in socio-psychological concepts, and the models should accurately capture phenomena widely accepted in the social science literature.

A number of different and exciting questions are currently being asked. New models are emerging to describe how a discrepancy can arise between an individual's private and expressed opinion on the same topic. Such models have enormous potential to be used to provide a mathematical framework for studying phenomena such as "the spiral of silence", where individuals stop voicing opinions during a discussion due to fear of appearing different. In a different direction, we seek to answer the following question: how does an individual's self-confidence in his or her opinion change over time if that individual is able to observe the amount of contribution he or she has on the group discussion? Preliminary analysis shows that the way self-confidence changes is tightly related to the network structure. However, further investigations are needed to establish what might happen when an individual underrates or overrates his or her contribution to the discussion because of e.g., humbleness or arrogance. Using rigorous mathematical analysis, we aim to establish theoretical conclusions about opinion evolution that have meaningful social interpretations, so that our understanding of society can be furthered systematically.