Previous research indicates that the human decision-making process is somewhat nonlinear and that nonlinear models would be more suitable than linear models for developing advanced decision-making models. In our study, we tested this generally held hypothesis by applying linear and nonlinear models to expert's decision-making behavior and measuring the predictive accuracy (predictive validity) and valid nonlinearity. As a result, we found that nonlinearity in the decision-making process is positively related to the predictive validity of the decision. Secondly, in modeling the human decision-making process, we found that valid nonlinearity is positively related to the predictive validity of nonlinear models. Thirdly, we found that the more nonlinearity is inherent in the decision-making process, the more nonlinear models are effective. Therefore, we suggest that a preliminary analysis of the characteristics of an expert's decision-making is needed when knowledge-based models...