Assessing Drivers' Mental Model Of Advanced Driver Assistance Systems Using Signal Detection Theory

Huang, Chunxi; Yan, Song; He, Dengbo · 2023 · Proceedings of the Human Factors and Ergonomics Society Annual Meeting

DOI: 10.1177/21695067231193671

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 addresses the limitations of current methods for assessing drivers’ mental models of Advanced Driver Assistance Systems (ADAS). Previous research relied on percent-correctness-based Mental Model Scores (MMS), which fail to capture response bias and hinder cross-study comparisons due to varying question structures. To resolve this, the authors propose using Signal Detection Theory (SDT) metrics—specifically sensitivity ($d'$) and response bias ($c$)—as more robust measures of drivers’ understanding of ADAS functions and limitations. The researchers conducted an online survey of 287 Level-2 ADAS users in China. Participants responded to 49 statements regarding ADAS capabilities using a 6-point Likert scale, with responses binarized to distinguish agreement from disagreement. The study calculated two types of MMS: a binary version (bMMS) and a continuous version accounting for confidence (cMMS). Additionally, SDT metrics were derived using both traditional SDT (for binary data) and Fuzzy SDT (to handle uncertainty in Likert responses). Linear regression models were fitted to determine how well $d'$ and $c$ predicted MMS, and mixed models identified demographic and driving factors associated with these metrics. Results indicated that $d'$ and $c$ accounted for a large variance in MMS, with adjusted $R^2$ values exceeding 0.8. Traditional SDT metrics best predicted binary MMS, while Fuzzy SDT metrics best predicted continuous MMS, suggesting that the choice of SDT method should align with how MMS is calculated. Significant interaction effects were observed, where higher response bias increased the marginal effect of sensitivity on MMS. Furthermore, while predictors of MMS (such as ADAS familiarity) also predicted $d'$ and $c$, the SDT metrics were associated with additional factors not captured by MMS, such as vehicle possession period and ADAS experience. The findings support the adoption of $d'$ and $c$ as standard metrics for assessing ADAS mental models, as they provide more comprehensive insights than percent-correctness scores alone. By separating sensitivity (objective knowledge) from bias (subjective inclination), SDT offers a clearer distinction between what drivers know and how they perceive the system. This approach facilitates better cross-study comparisons and identifies specific factors influencing driver understanding, ultimately aiding in the development of strategies to improve ADAS safety and user training.

Key finding

Signal detection theory metrics of sensitivity and response bias account for a large variance in drivers' ADAS mental model scores and capture additional predictive factors not identified by traditional percent-correctness measures.

Methodology

survey

Sample size: 287

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.

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.

• situational awareness
• mental model of traffic
• decision making risk perception
• signal detection theory

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).

• Methodological Resource: validation psychometrics
• Theoretical Contribution: computational model, theory or model