Using fNIRS to Verify Trust in Highly Automated Driving

Perello-March, Jaume R.; Burns, Christopher G.; Woodman, Roger; Elliott, Mark T.; Birrell, Stewart A. · 2023 · Crossref

DOI: 10.1109/tits.2022.3211089

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 critical need for objective, real-time measurement of trust in automation (TiA) within highly automated driving (HAD) contexts. Current methods rely heavily on subjective self-report scales, which lack the temporal resolution and objectivity required for real-time driver state monitoring. The authors propose using functional near-infrared spectroscopy (fNIRS) to identify neural correlates of trust and distrust, aiming to distinguish between the cognitive mechanisms underlying these distinct states. The research is motivated by the safety implications of inappropriate reliance or over-trust in automated systems, seeking to establish a neuroergonomic basis for future monitoring technologies. The experimental design involved 27 participants in a high-fidelity driving simulator, randomly assigned to either a low credibility (LC) or high credibility (HC) group based on induced expectations of system reliability, while actual vehicle performance remained identical across groups. Participants engaged in various driving scenarios, including highway driving, urban traffic, and a risky emergency maneuver, while performing a verbal 2-back working memory task as a control for mental workload. fNIRS data were collected from the prefrontal cortex using a wearable NIRSport device, focusing on regions such as the dorsolateral, ventrolateral, and orbitofrontal prefrontal cortices. Data were pre-processed to calculate hemodynamic responses, specifically oxygenated (HbO) and deoxygenated (HbR) hemoglobin concentrations, and analyzed using analysis of variance to detect localized brain activation differences between groups and conditions. The results demonstrate that trust and distrust are mediated by separate yet interrelated cortical mechanisms. Trust was associated with decreased monitoring and working memory load, evidenced by lower cortical activation in the LC group during complex scenarios compared to the HC group, who maintained higher vigilance. Conversely, distrust was found to be event-related and strongly tied to affective mechanisms, with the LC group showing greater brain activation in the orbitofrontal, ventrolateral, and dorsolateral prefrontal cortices during risky situations. The study confirms that situational TiA calibrates based on induced credibility expectations, with distrust eliciting a complex, resource-intensive top-down response involving emotional processing networks, whereas trust involves a more deliberate, cumulative process. These findings are significant for the development of driver state monitoring systems in automated vehicles. By identifying distinct neural signatures for trust and distrust, the study provides a methodological foundation for using fNIRS to objectively measure situational TiA in real-time. This capability could mitigate the risks associated with over-trust or inappropriate reliance by detecting shifts in driver engagement and cognitive load. The research underscores the importance of considering both cognitive and affective components in human-automation interaction, offering a pathway toward safer integration of highly automated driving technologies through neurophysiological feedback.

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-nvidia on 2026-08-10; verification: verified.

Topics

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

• trust calibration
• trust in automation foundations
• automation
• automation surprise

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, physiological data
• Methodological Resource: validation psychometrics