GAMESlab

Cooperation is an emergent social state related to the dynamics and complexity of road traffic and is reinforced through adaptive learning. Game theory and research in behavioural economics provide ample evidence that cooperation can efficiently solve social dilemmas similar to traffic congestion in dynamic settings. Traffic theory, asserts User Equilibrium, is both a stable and equitable, albeit inefficient, network state, which is a behavioural outcome of the selfish uncoordinated decision of drivers. In contrast, the System Optimum is an efficient network state that minimizes the total travel costs but is hard to maintain due to the inherent cost inequalities drivers will incur. In this paper, we describe how the principles of game-theory in a simple 2-player game allow the emergence of a stable system optimum through cooperation. We then investigate what happens in n-player games by applying an agent-based route-choice model. The model shows how reinforced learning and different behavioural specifications regarding agents’ cognition – selfish or cooperative - brings a simple road network from User Equilibrium towards the system optimum while preserving sufficient equity amongst drivers. The results suggest that a sufficient number of route alternations between drivers and a certain degree of altruism allow for a self-organizing formation of a fairness equilibrium that can maintain the network in the system optimum. The implications of future congestion management strategies that can be implemented with information and communication technologies are discussed.