Building Contextualized Trust Profiles in Conditionally Automated Driving

Avetisyan, Lilit; Ayoub, Jackie; Yang, X. Jessie; Zhou, Feng · 2024 · IEEE Transactions on Human-Machine Systems

DOI: 10.1109/thms.2024.3452411

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 formation of contextualized trust profiles in conditionally automated driving (SAE Level 3) to facilitate personalized user experiences and improve safety. While prior research has examined factors influencing trust in automated vehicles (AVs), such as system reliability and transparency, there is a lack of comprehensive understanding regarding how individual differences—such as personality, emotions, and dispositional trust—interact with dynamic trust during actual driving interactions. The authors aim to address this gap by identifying distinct trust profiles and validating the factors that predict them. The researchers conducted a between-subjects experiment using a desktop-based driving simulator with 70 university students. Participants were assigned to one of three conditions: a control condition with valid takeover requests (TORs), a false alarm condition, or a miss condition where the AV failed to detect obstacles. Throughout the simulation, the study measured dispositional trust, initial learned trust, personality traits (using the Big Five Inventory), and emotional responses. Crucially, dynamic trust was monitored in real-time via single-item ratings every 25 seconds. The data, comprising 48 normalized features, were analyzed using a K-means clustering algorithm to identify trust profiles. To validate these profiles, the authors employed a multinomial logistic regression model interpreted through SHapley Additive exPlanations (SHAP) to determine feature importance. The analysis identified three distinct trust profiles: "oscillators," "believers," and "disbelievers." Oscillators (23 participants) exhibited high dispositional and initial learned trust but showed rapid fluctuations in dynamic trust based on immediate experiences. Believers (31 participants) demonstrated moderate initial trust that increased over time, maintaining relatively high dynamic trust. Disbelievers (16 participants) displayed the lowest levels of dynamic, dispositional, and initial learned trust, with trust levels either decreasing or remaining low throughout the experiment. Statistical analysis revealed significant differences in personality and emotions across these profiles. Specifically, disbelievers scored significantly higher on the "carper" personality trait (finding fault with others) and reported higher levels of negative emotions such as resentment, hostility, and fear. Conversely, believers reported significantly higher levels of confidence, security, and happiness. The validation model achieved an F1-score of 0.90 and an accuracy of 0.89, confirming that personality traits, particularly agreeableness, and emotional states are strong predictors of trust profiles. The findings underscore the importance of considering individual differences in the design of AV trust calibration systems. By identifying specific profiles and their associated psychological drivers, designers can develop targeted interventions to adjust trust levels, thereby enhancing safety and user acceptance. The study provides a framework for moving beyond generic trust models toward personalized, context-aware systems that account for the complex interplay between driver personality, emotional state, and system performance.

Key finding

Three distinct trust profiles (oscillators, believers, and disbelievers) were identified in automated driving, characterized by significant differences in personality traits, emotional responses, and dynamic trust patterns.

Methodology

simulator

Sample size: 70

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 author_sweep_intake on 2026-05-28.

StageOutcomeToolModelPromptAttemptsCompleted
discover success author_sweep 2 2026-05-28
archive success canonical_url 1 2026-06-06
extract success cached 3 2026-06-10
clean success clean 1 2026-06-04
chunk success chunk 1 2026-06-04
embed success embed Qwen/Qwen3-Embedding-8B 1 2026-06-04
enrich success 1 2026-05-28
promote success 1 2026-06-04
summarize success llm qwen3.6-27b-prismaquant summ-v5 2 2026-06-10
tag success vector_similarity 15 2026-06-11
verify success 2 2026-06-10

Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

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).