A methodological framework for capturing relative eyetracking coordinate data to determine gaze patterns and fixations fromtwo or more observers

摘要

Few studies have published methodologies that can be used to analyze simultaneous gaze behaviors and recurrent fixations while multiple observers are viewing dynamic scenes and moving their heads. In this study, we aimed to develop a methodological framework to assess simultaneous gaze behaviors and recurrent fixations in predetermined areas of interest, while accounting for head movement and nonstandard observer positioning. Gaze coordinates were recorded during six trials in which a single participant focused on the center of a video image and moved his head in six degrees of freedom. Markers were positioned at the image corners. Eyetracking equipment recorded the video image and gaze behaviors (crosshair), which were then uploaded to SIMI Motion Analysis software. The corner markers were digitized in order to determine image position as the head moved, and were used to calculate new gaze coordinates relative to this head movement. Calculations accounted for the perspective error due to nonstandard participant positioning. Across all trials, the error between the measured and calculated coordinates was acceptable (<3.5 %CV). The frequencies and durations of fixations (≥100 ms within 1 ° of visual angle) within six areas of interest are reported, and they compared well to manual calculations. This methodology was then assessed using participant dyads (N=5), each simultaneously observing the same images. Recurrent fixations were determined using a hierarchical model, and these also compared well to manual analysis. This article presents a valid and reliable methodological framework for determining fixation frequency, duration, and location from multiple observers, while accounting for head movement and nonstandard positioning. This framework facilitates the analysis of simultaneous oculomotor variables, improving ecological validity and reducing environmental constraints.