Eye Gaze and Mouse Cursor Relationship in a Debugging Task

Abstract

This study evaluates the relationship between eye gaze and mouse cursor movements in a debugging task. 9 students with relevant programming experience participated in this study. Debugging difficulty was manipulated (error types; lexical, logical and syntactic) in order to measure the effects on debugging performance (accuracy and reaction time), eye gaze and mouse cursor behavior (frequency and duration of visits of target areas vs. non-target areas). Performance data showed that lexical errors are easier to find than logical errors. Mouse cursor behavior was a significant indication of the level of difficulty and therefore performance. The general pattern of mouse movements was comparable with eye gaze patterns. This study indicates that mouse data does add relevant information on top of eye gaze. These results support the idea of using mouse tracking as an alternative for inferring intentions.