Quantum-like evidence networks decision-making model

Abstract

For the purpose of addressing the deviations of the Sure Thing Principle observed in cognitive experiments, numerous artistic models have been developed and refined within the probabilistic framework. In spite of this, it is clear that these models still have room for improvement in terms of effectively expressing and processing global ignorance information and local ignorance information. As a result, a quantum-like evidence networks decision-making model (QLENDM) is proposed. The work we operates within the framework of evidence theory, rather than probability theory, which is different from those art models, such as quantum dynamics Markov model, quantum-like approach, quantum prospect decision theory, and quantum-like Bayesian networks. Quantum-like basic probability assignment can better model the global ignorance or local ignorance information presented by cognitive experiments than the probability distribution, which solves the issue of inadequate modeling of uncertain information within a probabilistic framework. Moreover, from a heuristic perspective, QLENDM automatically fits the parameters related to interference effect using the distance of quantum-like basic probability assignments based on focal element structure. Therefore, QLENDM with these two characteristics, in addition to its universality, has better predictive performance than the above art models. We then apply QLENDM to cognitive science and information retrieval. As a result of the experiment, it appears that compared with other models, QLENDM has the minimum average and standard deviation of fitting errors, indicating that it is better suited for addressing deviations of the Sure Thing Principle as well as predicting the behavior of participants more accurately in decision-making experiments.