Theoretical Research On MPT Models And Its Simulation

Memory research is an important part in the field of cognitive psychology. According to the related theory of cognitive psychology, memory can be divided into two different parts, one is the main part of the memory called the item memory; the other part is the background information of the memory called the source memory, it contains the details of the main part of the memory of in the scene which the memory occurred. So far, the most direct and effective method to investigate the source memory is source monitoring experiment. Because of the traditional empirical method to analysis the experimental data can not meet the reality monitoring theory very well, psychologists proposed the multinomial processing tree (MPT) modeling method to handle this problem. MPT modeling method is a simple and effective approach developed from the polynomial distribution theory associated with some related concepts of cognitive psychology. Its basic idea is treat the latent psychological processes as several processing trees and each of which include three different psychological processes. In MPT modeling method, the participants'response bias will be considered, and the polynomial model will be employed to fit the latent psychology processes.Firstly, this paper introduces the related theories and development of the source memory investigation in the field of cognitive psychology briefly. Secondly, on the basis of a brief introduction of the multinomial processing tree modeling theory's background and development process, the strict definitions and basic characters of the MPT model were given and then followed with an example. Thirdly, the paper focused on the application of MPT modeling method in source monitoring, discussed its theoretical foundation, listed specific algorithm processing steps, accomplished the simulation, and then analyzed the simulated results in field of cognitive psychology. Finally, the original model were discussed and expanded from the following two parts, one is the simplification stand of different level sub models, and the other is the development of the original algorithm, and the effect of the improved algorithm was discussed later.