Abstract
The recognition-primed decision model is renowned for its adaptability and superior performance, making it extensively utilized across various domains, including artificial agents, social systems, medical diagnostic systems, and industrial operations. This paper presents a novel approach, namely the probabilistic memory model, which integrates the probabilistic characteristics inherent in human memory into the recognition-primed decision model and enhances the recognition-primed decision model’s ability to process continuous information effectively. The memory structure in the proposed model is represented by a succession of joint probability density functions, each delineating a prototype in human memory. Leveraging these prototypes, four memory functions, i.e., memory encoding, prototype abstraction, memory indexing, and memory retrieval, are designed. As a case study, the proposed model is applied to analyze pilot decision-making in midair encounter scenarios. The findings substantiate the superior suitability of the proposed model in handling continuous information. Furthermore, empirical evidence supports the notion that the proposed model excels in retaining information from experiences, thereby exhibiting improved accuracy compared to existing models. Additionally, this paper studies the sensitivity of the model to input noises and the prototype abstraction number. Finally, the proposed model is evaluated against realistic midair encounter scenarios. The evaluation results demonstrate that the proposed model can effectively predict pilot behavior during midair encounters, outperforming existing models in realistic midair situations.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call