Paying the price of learning independently in route choice

Abstract

In evolutionary game theory, one is normally interested in the investigation about how the distribution of strategies changes along time. Equilibrium-based methods are not appropriate for open, dynamic systems, as for instance those in which individual drivers learn to select routes. In this paper we model route choice in which many agents adapt simultaneously. We investigate the dynamics with a continuous method (replicator dynamics), and with learning methods (social and individual). We show how the convergence to one of the Nash equilibria depends on the underlying learning dynamics selected, as well as on the pace of adjustments by the driver agents. © 2013 Springer-Verlag.