An improved evidential Markov decision making model

The disjunction effect is an interesting phenomenon of the violation of the law of total probability, and many studies have shown that the categorization decision-making experiment (CDME) can produce the disjunction effect. Recently an Evidential Markov decision making modeling (EM) has been proposed to predict the disjunction effect in CDME. A critical step in EM is that it uses Deng entropy to measure the amount of information of mass function. However, Deng entropy just considers the partial amount of information, which results in information loss in the prediction of the disjunction effect. To address this problem, an Improved Evidential Markov decision making model, named IEM, is proposed in this paper. Instead of Deng entropy, another method, i.e., Information volume of mass function (IV), is applied by the proposed IEM. IV can measure the total amount of information by superposing information volume at different orders. Hence IV is more reasonable than Deng entropy in measuring information volume. Experimental results show that IEM can predict the disjunction effect more accurately than the existing methods.