Assessing the performance of visual identification tasks using time window-based eye inter-fixation duration

The objective of this research was to develop a new eye-tracking metric to evaluate human performance of a visual identification task. According to prior research on eye tracking, eye movement metrics could examine relationships between task performance and eye gaze behaviors. However, little is known about the relationship between the eye movements and the performance on the identification task. In this study, by using the time window-based human-in-the-loop simulation representing an anti-air warfare coordinator, we studied the performance on the identification task and eye tracking data during the experiment. The performance was evaluated by the total number of correctly identified aircraft. Saccade duration (SD) and time windows-based inter-fixation duration (TWID) were used to analyze participants' eye tracking data. Transition number (TN) was also used as a metric to compare SD with TWID. The results indicated that both TWID and SD showed a significant difference between excellent performers and others. The participants who perfectly identified all unknown aircraft had much shorter inter-fixation duration compared to others who made mistakes. However, only the well-skilled performers' average TWID was significantly related to TN. The findings of this research supported that TWID could be used as one of the eye movement measures to assess the performance of the identification task in a computer-based environment.