Abstract
This study investigated changes in pupil diameter when computer users were confused or surprised, and proposed a practical method to detect those situations. Pupil diameters of 5 male subjects aged between 20 and 23 were recorded while doing a memory task on a computer. Among the trials of primary memory task, 8 special trials were presented; 4 expected to cause confusion and 4 to cause surprise. Results demonstrated that if distributions of pupil diameter for given task conditions were known, subjects' confusion and surprise could be detected at rates of 25% to 65%. Detection rates varied depending on the settings of boundaries and the nature of the confusion and surprise conditions. The causes of this variation were discussed. Potential applications of this study include intelligent training systems that provide users with appropriate advice when users are troubled.
