Traffic flow simulation dovetailed with evolutionary game theory

Previous work (Yamauchi et al., PRE 79 and Nakata et al., Physica A 389) found several social-dilemma structures in phases of traffic flows created by bottlenecks resulting from lane closings. We investigate another question: whether or not such structures can be formed from frequent lane changes in a usual traffic flow without any explicit bottlenecks. In our model system, two classes of driver-agents coexist: C-agents (cooperative strategy) always remain in the lane they are initially assigned, while D-agents (defective strategy) try to change lanes to move ahead. In relatively high-density flows, such as the metastable and high-density phases, we found structures that correspond to either n-person Prisoner's Dilemma (n-PD) games or to quasi-PD games. In these situations, lane changes by D-agents create heavy traffic jams that reduce social efficiency.