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 such as expert systems are being developed. We also verify that the lens model is effective in evaluating the predictive validity of human judgment and in analyzing the validity and nonlinearity of the human decision-making process.