基于公平的博弈学习模型及其实验研究

A Modeling and Experimental Study of Game Learning Theory Based on Fairness

在"经验权重魅力值学习模型"(EWA)中引入公平因素而构建公平博弈学习模型,模型用公平思考引起的心理效用来代替策略物质效应,以此来修正决策参与者的策略收益,改变策略魅力值,进而研究动态博弈过程中博弈均衡的移动;通过对公平博弈学习模型和EWA学习模型的最后通牒博弈决策模拟比较,结果发现公平博弈学习模型能更快地收敛于均衡策略;我们继而设计并进行了冰淇淋蛋糕分配实验,实验证明了参与者在实验过程中存在公平思考及学习等行为;最后用实验数据检验公平博弈学习模型与EWA学习模型,结果显示:两个模型都可以收敛于均衡策略,但公平博弈学习模型具有更快的收敛速度和对参与者决策行为更好的预测能力,实验结果表明人类的公平思考行为影响着决策行为。

By incorporating fairness factor in the EWA （experience-weighted attraction） learning model, we develop an extended game learning model called FLG model. We use psychological effect instead of material effect to modify strategy＇ s payoff and attraction, then, to study the equilibrium movement in dynamic Games. Psychological function changes when participants have fair thinking. Compared with EWA learning model by simulating the decision-making in Ultimatum Game, we find FLG model converges to equilibrium strategy faster. Then we design and carry out an experiment of ice-cream cake distribution bargaining. The experiment data prove that participants behave with fair thinking, learning and so on. Using the experiment data, we test the modified model and traditional EWA learning model, the results indicate that both models can converges to equilibrium strategy, but the former model moves to equilibrium faster and has more forecast ability on participants＇ strategy behavior than the latter. The experimental results indicate that people＇s fairness thinking is an important factor in the deciding process.