Examining the Validity of the Exemplar-Based Classifier in Identifying Decision Strategy with Eye-Movement Data

In this study, an exemplar-based classifier was developed to predict which decision strategy may underlie an empirical ocular search behavior. Our rationale was mainly inspired by the exemplar-based models of categorization; that is, different decision strategies are conceived as different concepts, with the exemplar referring to the sequence of empirical fixations on decision information during a decision process. In order to ascertain the best exemplar of each strategy for our classifier, the Tabu search algorithm was applied. An eyetracking based experiment was conducted to collect fixation data for training and validation. Our result showed that the classifier has significant accuracy in identifying underlying strategies, achieving an average hit-ratio of 76%. This indicated to us that the integration of the exemplar classifier with fixation data has certain applicable value for leveraging the adaptability of DSSs. Our result also has some important implications for the direction and methodology of behavioral decision research