Reliability of oculomotor and driving simulation performance metrics

Background: Driving simulators assess how human and technical factors influence operational performance. Examples of driving performance measures include speed and lane position variability and their associated physical inputs, e.g., throttle and steering input variability. Secondary cognitive tasks, designed to assess oculomotor function and attentional capacity, are often integrated into driving simulations. However, the ‘test-retest’ reliability of driving performance and oculomotor performance measures when obtained simultaneously are yet to be established. This study assesses the reliability of oculomotor and driving performance measures evaluated within the same paradigm. Methods: Participants (n = 14) completed a familiarisation and two driving simulator sessions. Each session entailed drives through two scenarios, ‘urban’ and ‘country’, each demanding different levels of driving skill and cognitive complexity. Driving performance was measured in both scenarios. The country scenario included an integrated spatial attention task where performance was assessed via saccade kinematics and psychomotor reaction time. Within- and between-session reliability was calculated for all measurements. Results: Lane position and steering input exhibited good (ICC 0.75-0.9) to excellent (ICC > 0.9) within-session agreement, whereas between-session reliability was poorer. Speed and throttle variability had moderate to poor (ICC 0.5-0.75) within-session reliability and poor agreement between-sessions. Saccade velocity and psychomotor reaction time exhibited better reliability within-session and satisfactory reliability between-sessions. Conclusion: When combined within a driving simulation paradigm, driving performance and oculomotor measures are a reliable research tool. For the most confidence in results, we recommend that intervention studies conduct baseline measurements for multi-day study designs.