$N$-player game formulation of the majority-vote model of opinion dynamics

Abstract

From a self-centered perspective, it can be assumed that people only hold opinions that can benefit them. If opinions have no intrinsic value, and acquire their value when held by the majority of individuals in a discussion group, then we have a situation that can be modeled as an $N$-player game. Here we explore the dynamics of (binary) opinion formation using a game-theoretic framework to study an $N$-player game version of Galam's local majority-vote model. The opinion dynamics is modeled by a stochastic imitation dynamics in which the individuals copy the opinion of more successful peers. In the infinite population limit, this dynamics is described by the classical replicator equation of evolutionary game theory. The equilibrium solution shows a threshold separating the initial frequencies that lead to the fixation of one opinion or the other. A comparison with Galam's deterministic model reveals contrasting results, especially in the presence of inflexible individuals, who never change their opinions. In particular, the $N$-player game predicts a polarized equilibrium consisting only of extremists. Using finite-size scaling analysis, we evaluate the critical exponents that determine the population size dependence of the opinion's fixation probability and mean fixation times near the threshold. The results underscore the usefulness of combining evolutionary game theory with opinion dynamics and the importance of statistical physics tools to summarize the results of Monte Carlo simulations.