When automation fails - Investigating cognitive stability and flexibility in a multitasking scenario

Managing multiple tasks simultaneously often results in performance decrements due to limited cognitive resources. Task prioritization, requiring effective cognitive control, is a strategy to mitigate these effects and is influenced by the stability-flexibility dilemma. While previous studies have investigated the stability-flexibility dilemma in fully manual multitasking environments, this study explores how cognitive control modes interact with automation reliability. While no significant interaction between control mode and automation reliability was observed in single multitasking performance, our findings demonstrate that overall task performance benefits from a flexible cognitive control mode when automation is reliable. However, when automation is unreliable, a stable cognitive control mode improves manual takeover performance, though this comes at the expense of secondary task performance. Furthermore, cognitive control modes and automation reliability independently affect various eye-tracking metrics and mental workload. These findings underscore the need to integrate cognitive control and automation reliability into adaptive assistance systems, particularly during the perceive stage, to enhance safety in human-machine systems.