Estimation of expected human attention weights based on a decision field theory model

Modeling human decision making behavior is of great interest in understanding how a decision maker weights different decision attributes when making a decision. Such knowledge is critically important in helping predict future decisions, evaluating human decision performance, and improving the design of human and machine interface systems. Decision field theory (DFT) provides a psychological representation of the cognitive deliberation process, which is driven by the fluctuations of a person's attention among decision attributes. In this research area, the most common use of a DFT model is to estimate or predict the human decisions by using a set of pre-specified expected attention weights (EAWs) in the DFT model. Unlike other research, this paper extends the capabilities of DFT in a complementary direction, showing how to fit or train a DFT model by estimating the EAW based on sequentially obtained samples of decision trials. Furthermore, the inherent connection between the EAW and the decision choice uncertainty is investigated. The proposed modeling method is discussed in detail for a two-alternative decision scenario based on two attributes. Both simulations and a case study are conducted in the paper to demonstrate the effectiveness of the proposed modeling approach.