Game Learning Model And Its Experimental Study Based On Fairness Considered

Traditional economic theory based on rational economic-manassumptions is questioned by more and more people, experiments someeconomists do show that the participants depart from self-rationalassumption partially or completely. Participants demonstrate fairness,reciprocity, altruism, learning tendencies in experiments. Economistshave developed a theoretical model to explain the participants' non-rational self-regard behavior. Among them, the fair model, the learningmodel is widely used, but those two models have some Shortcomings,the fair model consider the Participants' psychological preferences (fair)how to influence decision-making, but it didn't explained that participantsuse available information to adjust their strategy in iterative game.In thelearning model, Participants didn't consider psychological factors willimpact the game-movement.Sowe develop a model which is composed offairness and learning model in this paper, called Game learning modelbased fairness considered (ab. FGL model).FGL model is an improved by adding the fairness factor in the EWAmodel,so the FGL model consider the participants fair reflection on theimpact of the equilibrium movement in dynamic Games.It replacematerial effect with psychological function changes when participantshave fair thinking,and modify strategy's profit and strategy'sattraction.FGL model also reserve the EWA learning model's advantage,So it can accurately forecast the probability of each strategy insequential game and correctly predict the equilibrium equemergence.FGL model make decision after it have learned all strategy income, soeven if there are mutil-equilibrium, FGL model can choose the bestequilibrium from equilibriums.It resovle the Rabin'smodel problem. FGLmodel has more interpretative ability than EWA learning model.Ice-cream cake experiment verificated that the participants have fairthinking, learning behavior.They are not perfectly rational self-regard.used the experiment data,we Estimate the FGL model and EWA learning model's initial parameter values and strategies attractions. Wesimulate decision-making process in Experiment 1 and Experiment 2. Thesimulation results show that the two models all captured balancedstrategy after some rounds game,but Proponent's decision-makingprocess is closer to reality in the FGL model than in EWA learning model.FGL model has more forcast ability than EWA learning model, moreexplain ability.