| Abstract | Under incomplete information and repeated decision-making, belief adjustment is in accord with more “rational” belief learning, or reinforcement learning that is in short-sighted to be more affected by the immediate rewards, while the subject of intensive study? This paper attempts to answer this question. To this end, double tournamentp game here, a simplified experiment to study the spot reward of faith adjustment - that is, the impact on learning. Experimental results show that there is no immediate reward in the case of individual beliefs usually better to start learning quickly approach the optimal decision, and the entire virtual game model parameters EWA model can better reflect this process. By contrast, there are immediate rewards, more attention to the individual or the more recent round before the round of selection and payment of the game, by strengthening the impact of the more obvious stimulus, interferes with the individual convergence to the optimal decision, belief learning characteristics being undermined. All parameters of EWA model can explain the process that exist strengthened incentives, which combine the reinforcement learning and belief learning. |