ScanMatch Versus MultiMatch: A Comparison of the Sensitivity of Scanpath Similarity Metrics to Changes in Workload

In complex domains such as aviation, military, and healthcare, the increasing complexity of technology and automation is leading to a growing demand for operators to complete more tasks that rarely stay at one level of cognitive workload. The goal of this work is to apply scanpath similarity algorithms such as ScanMatch and MultiMatch to eye tracking data of pairs during UAVs tasks while they are subject to workload changes. The aim is to assess whether these metrics are sensitive to workload changes and whether these metrics correlate with performance. Twenty-six pairs of college students were recruited for the study. Despite observing slightly higher ScanMatch scores in the high workload condition compared to the low workload condition, the paired t-test did not reveal a statistically significant difference. On the other hand, MultiMatch analysis showed more promising results. All MultiMatch metrics except "Position" showed sensitivity to workload changes. In this analysis, we had predicted that a higher similarity between participants' scanpaths would result in better performance. Our hypothesis held true for three dimensions of the Multi-Match algorithm: shape, length, and duration similarity. How teammates scan has a higher impact on performance than where exactly they are looking. These findings carry important implications for the design of interfaces, as metrics showing sensitivity to workload changes could be leveraged for real-time monitoring and adaptive task adjustments. Additionally, the identified significant correlations provide a basis for targeted training programs aimed at improving cognitive processes associated with specific MultiMatch metrics to enhance overall task performance.