A Novel Experiment Design for Studying Multiple Cognitive Factors in Conditionally Automated Driving Contexts

Development of responsive automation necessitates a framework for studying human-automation interactions in a broad range of operating conditions. This study uses a novel experiment design involving multiple binary perturbations in different stimuli to elicit measurable changes in cognitive factors that affect human-decision making during conditionally-automated (SAE Level 3) driving: trust in automation, mental workload, self-confidence, and risk perception. To infer changes in these factors, psychophysiological metrics such as heart rate variability and galvanic skin response, behavioral metrics such as eye gaze and reliance on automation, and self-reports were collected. Findings from statistical tests revealed significant changes, particularly in psychophysiological and behavioral metrics, for some treatments. However, other treatments did not elicit a significant change, highlighting the complexities of a between-subject experiment design with variations in multiple independent variables. Findings also underscore the importance of collecting heterogeneous human data to infer changes in cognitive factors during interactions with automation.