Classification of human trust in AI using brain activity data

Authors

Danushka Bandara, Fairfield UniversityFollow
Ruhuan Liao, Fairfield UniversityFollow
Fatima Chowdhury, Fairfield UniversityFollow
Leslie Abbott, Fairfield UniversityFollow

Author ORCID Identifier

0000-0002-8885-622X

Abstract

Trust plays a crucial role in human-computer interaction, particularly in scenarios involving artificial intelligence (AI) systems. This study explores the feasibility of using functional near-infrared spectroscopy (fNIRS) data to classify trust levels in human-AI interaction scenarios. A total of 18 participants completed an image classification task with an AI team member while their hemodynamic responses were recorded using fNIRS. Preprocessing of fNIRS data involved motion artifact removal, filtering, and normalization. Exploratory analysis identified significant associations between hemodynamic responses in the prefrontal cortex and trust levels. An across-subject binary trust classification model was developed using machine learning techniques, achieving an F1 score of 0.77. Receiver Operating Characteristic (ROC) analysis revealed an Area Under the Curve (AUC) of 0.81, achieving improved F1-score and AUC compared to comparable methods. The brain activity- based classifiers were found be be better at classifying the self-report trust level than the objective measure of trust. These findings demonstrate the potential of fNIRS-based approaches for real-time classification of trust levels in human-AI interaction, with implications for improving user experience and trustworthiness of AI systems.

Recommended Citation

Bandara, Danushka; Liao, Ruhuan; Chowdhury, Fatima; and Abbott, Leslie (2025) "Classification of human trust in AI using brain activity data," Northeast Journal of Complex Systems (NEJCS): Vol. 7 : No. 2 , Article 4.
DOI: https://doi.org/10.63562/2577-8439.1108
Available at: https://orb.binghamton.edu/nejcs/vol7/iss2/4