An Investigation of Drivers' Dynamic Situational Trust in Conditionally Automated Driving
DOI: 10.1109/thms.2021.3131676
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This study investigates the dynamics of situational trust in conditionally automated vehicles (AVs), specifically examining how system performance and initial trust preconditions influence drivers' trust during takeover transitions. The research is motivated by the safety risks associated with undesirable trust levels, such as overtrust leading to delayed takeovers or undertrust causing unnecessary interventions. While previous studies often measured trust statically, this work focuses on how trust evolves dynamically over time, which is critical for designing effective trust calibration systems. The researchers employed a 3 (system performance: 95%, 80%, 70% accuracy) by 2 (trust precondition: overtrust, undertrust) mixed-subjects design with 42 participants. Participants viewed 30 video scenarios of AV driving with potential takeover requests. To establish preconditions, participants first watched 10 consecutive success videos (overtrust) or failure videos (undertrust). They then viewed 20 additional videos with varying failure rates corresponding to the accuracy levels. Trust was measured using two methods: self-reported situational trust (SST) via the Situational Trust Scale for Automated Driving (STS-AD) and behavioral situational trust (BST) via agreement and switch fractions, which quantified participants' willingness to take over control before and after observing the AV’s decision. The results revealed a divergence between self-reported and behavioral trust measures. Participants’ self-reported trust adjusted quickly and consistently with the system’s accuracy levels, regardless of their initial precondition. Statistical analysis showed a significant main effect of accuracy on SST, with higher accuracy yielding higher trust, but no significant effect of the trust precondition. In contrast, behavioral trust was significantly influenced by the precondition. Participants in the overtrust condition exhibited a significantly higher agreement fraction (willingness to let the AV drive) compared to those in the undertrust condition. Conversely, the undertrust precondition significantly decreased the switch fraction, indicating less willingness to change their initial takeover decision after seeing the AV’s action. These behavioral effects persisted across different accuracy levels, suggesting that initial trust states can trap drivers in specific behavioral patterns despite changes in system performance. The findings highlight the importance of distinguishing between self-reported and behavioral trust when evaluating human-AV interaction. While drivers may cognitively adjust their trust ratings to match system reliability, their behavioral responses remain anchored to their initial trust precondition. This implies that overtrust and undertrust can create persistent biases in driver behavior that are not immediately corrected by observing system performance. The study provides critical insights for designing in-vehicle trust calibration systems, suggesting that interventions must address behavioral inertia to ensure safe and appropriate takeover transitions in conditionally automated driving.
Key finding
Adding the combined MCMS message produced no seat belt use gain beyond the single-issue campaigns, as observed belt-use increases were generally larger in control areas than in MCMS program areas.
Methodology
field_study
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via openalex_abstract on 2026-05-08 (8 acquisition events logged).
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 3 | 2026-05-29 |
| archive | success | semantic_scholar | — | — | 21 | 2026-06-10 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | — | — | — | 1 | 2026-06-01 |
| chunk | success | — | — | — | 1 | 2026-06-01 |
| embed | success | — | — | — | 1 | 2026-06-02 |
| enrich | success | openalex | — | — | 2 | 2026-05-08 |
| promote | success | — | — | — | 2 | 2026-06-10 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 3 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 18 | 2026-06-11 |
| verify | partial | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified_with_issues.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- trust calibration
- trust in automation foundations
- automation
- automation surprise
- situational awareness
- acceptance adoption
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: self report data
- Methodological Resource: validation psychometrics
- Theoretical Contribution: computational model