The impact of partial automation and cognitive load on drivers’ visual search strategies in an urban driving environment

Non-visual cognitive tasks (e.g., hands-free phone conversations) are well known to impair drivers' visual search in manual driving, but evidence in partially automated driving (PAD) remains inconsistent and limited, especially in complex driving environments that demand active visual scanning. In a driving-simulator study, we examined how automation and cognitive load (2-back task) shape visual-search strategies in a busy urban setting with frequent distractors and potential hazards, using both temporal and sequential scanning measures. Unlike findings from monotonous driving contexts, PAD drivers broadened their visual search without trading off attention to the forward road. Under cognitive load, their attention to peripheral roadway declined more sharply than in manual driving, leading to restricted visual search across both driving modes. Our results underscore the influence of non-visual distraction on drivers' visual scanning during PAD in complex environments and highlight the value of eye-tracking-based features for monitoring cognitive load.