<p>Autonomous control of traffic light timing based on the current state of the traffic queue is the ideal goal for any traffic engineering problem. Various techniques have been proposed that would utilize a dynamic system to incrementally find the optimal values, either by human programming or machine learning. However the human factor of perceived satisfaction is not considered. The paper proposes a novel utility defined as the ratio of perceived satisfaction relative to the green light time assigned. It is not only based on perceived levels of satisfaction within a queue, but also across queues. Human nature gauges satisfaction relative to others. This yields a form of competition between the traffic lights at an intersection. Game theoretical concepts are used to analyze the performance of the system, for which the resulting operating point is the well-known non-cooperative Nash equilibrium. To help improve the system performance, a mechanism is designed for taxing the individual traffic lights in proportion to the amount of harm that they cause to others, i.e. wait time. The algorithm proposed is shown to be convergent and results in a socially optimum point known as the Nash Bargaining Solution. This solution is also achieved through individual selfish non-cooperative means, yet yields a Pareto optimal operating point that Pareto dominates the original Nash Equilibrium. In essence, green light time is assigned in a way to efficiently utilize the resources of the road.</p>