Fixation Region Overlap: A quantitative method for the analysis of fixational eye movement patterns

Abstract

This article presents a new method for the quantitative analyses of fixation patterns in eye tracking data. The Fixation Region Overlap Analysis (FROA) uses thresholded spatial distributions of fixation frequency or duration to determine regions-of-interest (ROIs). The locations of these ROIs are contrasted with fixation regions of other empirically-derived, or modelled, data patterns by comparing region pixel overlap. A Monte Carlo procedure is used to assess the statistical significance of fixation region overlap based on 95% confi-dence intervals (C.I.) of the distribution of random overlap for each set of thresholded ROIs. The value of the FROA method is demonstrated by applying it to data acquired in an object recognition task to determine which of two potential models best account for the observed fixation patterns.