A paper out today in IEEE Transactions on Cybernetics features QS2 research advancing understanding of game-theoretic agent behaviors in networked systems.
Modern, mathematical game theory has been used as a formal language to study behavior since the mid-1900s. Of course, the accuracy of resulting models (and all of today’s models of human behavior, for that matter) is unclear. But, the core ideas of game theory – that people are working to optimize some payoff, that individual payoffs are influenced by the behavior of other people also trying to optimize their own payoffs – are important and have proven highly influential.
The trick is that formal understanding of mathematical games typically leans on simplifying assumptions, e.g., that the population is fully mixed (everyone in the population has equal likelihood of interacting with everyone else). Outcomes of games on structured populations, like social networks, are much harder to characterize. In these cases, simulations are used. But those again rely on parameter and initialization assumptions.
“Opinion Dynamics in the Presence of Increasing Agreement Pressure” takes steps, both formally and through agent-based simulations, to move the dial – if slightly – on the set of questions related to the behavior of game-theoretic agents in networked populations. Specifically, we study these systems under influence to come to consensus (peer pressure, if you will). We envision this work as foundational to mathematical studies of peer influence on opinion dynamics in social networks.