Abstract
The ability to objectively compare scanpaths enables researchers to study the adoption and evolution of visual search strategies within complex task environments. Current eye tracking technology is enhancing the researcher's ability to collect scanpaths. However, a drawback of using eye tracking technology is the difficulty in objectively analyzing portions of the obtained data. For example, the ability to compare two complete scanpaths, two subsections within the same scanpath, or two subsections of two different scanpaths has been elusive.
Recent research has applied objective measures to scanpaths (see Goldberg & Kotval, 1999; Ponsoda, Scott, & Findlay, 1995). However, such measures fail to utilize the temporal flow of dwells composing scanpaths (e.g., scanpath length or convex hull area). I present a novel method to objectively determine scanpath similarity that includes the temporal flow of dwells. A sequence alignment algorithm adopted from bioinformatics is used to determine whether areas of interest in a complex task environment are foveated in the same temporal order over repeated trials and between participants. The algorithm determines the minimum number of edits necessary to change one scanpath into another - the smaller the number of edits the greater the similarity between the compared scanpaths.
To demonstrate the usefulness of scanpath comparisons, all scanpaths from two conditions of an empirical study were compared for two participants (one per condition). Results demonstrate that prototypical scanpaths are determinable through sequence alignment, and represent search strategies. Results also demonstrate that such strategies are attained and settled on early in the task, and that two or more strategies likely compete over the course of the task. Furthermore, prototypical search strategies for the two subjects were similar, but differed systematically, suggesting that minute changes in task environments subtly change visual search strategies.
The work reported was supported by grants from the Air Force Office of Scientific Research AFOSR F49620-03-1-0143, as well as the Office of Naval Research ONR N000140310046, to Wayne D. Gray.