Game theory is commonly used to model problems from a large spectrum of fields including: economics, engineering, social science, and theoretical biology. A two-person game can be viewed as a two-sided optimization, where each side wants to optimize its payoffs. Many practical games involve multiple objectives (payoffs), which might be contradicting. Multi-Objective Games (MOGs), aim to deal with such problems. Most studies on MOGs, involves scalarization (e.g., by a weighted sum). This is problematic since that such an approach does not allow a comprehensive investigation of the available strategies of the players. Moreover, it requires a-priori decision on preferences towards objectives. Here we suggest a novel approach for MOGs in which the players postpone their decision on objective preferences, which enables an examination of performance tradeoffs. The proposed approach is based on an amalgamation of game theory with solution methods from Multi-Objective Optimization (MOO). This results in a selection procedure, which leads to the optimal strategies. Moreover, a co-evolutionary algorithm is suggested to solve the MOG, with postponed objectives, by means of heuristic methods. To demonstrate the proposed approach, the classical tug-of-war game is re-defined as a MOG. This game is characterized as a game with continuous spaces of the strategies. Its appeal is that the optimal strategies are known in advance and therefore it can be used to evaluate the proposed algorithm. Next, the co-evolutionary algorithm is applied on the selected example. The results demonstrate the applicability of the suggested method and algorithm.
