Presentation
Identifying Informative Physiological Indicators and Time Scales for Inferring Human Cognition During Automated Driving: An Information-Theoretic Approach
DescriptionWe present an information-theoretic framework for identifying which physiological indicators, and at what temporal scales, are most informative of changes in human cognitive states (trust, workload, risk perception) during conditionally automated (SAE Level 3) driving. We analyze data collected from an in-person driving simulator study in which participants interact with a conditionally automated vehicle in a single continuous drive. Evaluating physiological signals, including heart rate and galvanic skin response, through mutual information enables assessment of informational relevance without assuming linear relationships or predefined model structures. This approach is then suitable for continuous, non-event-based settings. The results suggest that informative physiological responses emerge over intermediate time scales (50-70 seconds) and that no single indicator is universally optimal across individuals. These findings highlight the importance of using prediction models that are personalized to the user for effective cognitive state estimation.
Event Type
Lecture
TimeTuesday, October 20th3pm - 3:20pm PDT
Location
Similar Presentations