An interactive model-based environment for eye-movement protocol analysis and visualization

Abstract

This paper describes EyeTracer, an interactive environment for manipulating, viewing, and analyzing eye-movement protocols. EyeTracer augments the typical functionality of such systems by incorporating model-based tracing algorithms that interpret protocols with respect to the predictions of a cognitive process model. These algorithms provide robust strategy classification and fixation assignment that help to alleviate common difficulties with eye-movement data, such as equipment noise and individual variability. Using the tracing algorithms for analysis and visualization, EyeTracer facilitates both exploratory analysis for initial understanding of behavior and confirmatory analysis for model evaluation and refinement.