Heterogeneous Mixed Populations of Coordinating, Anticoordinating, and Imitating Individuals
Decision-making individuals are typically either an imitator, who mimics the action of the most successful individual(s), a conformist (or coordinating individual), who chooses an action if enough others have done so, or a nonconformist (or anticoordinating individual), who chooses an action if few others have done so. Researchers have studied the asymptotic behavior of populations comprising one or two of these types of decision-makers, but not altogether, which we do for the first time. We consider a population of heterogeneous individuals, each either cooperates or defects, and earns payoffs according to their possibly unique payoff matrix and the total number of cooperators in the population. Over a discrete sequence of time, the individuals revise their choices asynchronously based on the best-response or imitation update rule. Those who update based on the best-response are a conformist (resp. nonconformist) if their payoff matrix is that of a coordination (resp. anticoordination) game. We take the distribution of cooperators over the three types of individuals with the same payoff matrix as the state of the system. First, we provide our simulation results, showing that a population may admit zero, one or more equilibria at the same time, and several non-singleton minimal positively invariant sets. Second, we find the necessary and sufficient condition for equilibrium existence. Third, we perform stability analysis and find that only those equilibria where the imitators either all cooperate or all defect are likely to be stable. Fourth, we proceed to the challenging problem of characterizing the minimal positively invariant sets and find conditions for the existence of such sets. Finally, we study the stochastic stability of the states under the perturbed dynamics, where the agents are allowed to make mistakes in their decisions with a certain small probability.