Investigating the dynamic interaction between trust and mental workload in conditionally automated driving: an exploratory naturalistic study

Drivers’ trust in automated vehicles critically shapes technology adoption and road safety. However, the dynamic interaction between drivers’ trust levels and mental workload in automated driving remains insufficiently explored. This study investigates how trust in automation modulates mental workload under different driving modes and non-driving-related task (NDRT) complexities. Twenty participants completed a naturalistic driving experiment involving automated and manual driving under no-task, 0-back, and 1-back conditions. Multimodal data were collected through subjective assessments, including a trust questionnaire and NASA-TLX, and objective measures, including NDRT error rates and pupil diameter. Participants were classified into high- and low-trust groups based on questionnaire responses and validated using objective measures. Analysis of variance (ANOVA) was conducted to examine the effects of trust level, driving mode, and NDRT complexity on workload-related measures. A composite mental workload model integrating grey relation analysis and entropy-weighted TOPSIS was developed to quantify workload dynamics. Results showed that driving mode and NDRT complexity significantly affected subjective workload, while trust level and NDRT complexity significantly influenced NDRT errors. Trust level, driving mode, NDRT complexity, and their interactions significantly affected pupil diameter. The model revealed divergent workload patterns: low-trust drivers exhibited higher and more variable workload during automation (0-back: 0.513; 1-back: 0.947), whereas high-trust drivers maintained lower and more stable workload (0-back: 0.046; 1-back: 0.590). These findings provide a trust-sensitive framework for mental workload assessment and support personalized interface design and trust calibration in conditionally automated driving and safer human–machine collaboration in automated driving.