Enhancing Deception Detection with Cognitive Load Features: An Audio-Visual Approach
Keywords: deception detection, cognitive load, audio visual
Abstract: Deception ranges from minor mischief to serious fraud, often leading to significant psychological and financial harm. Effective deception detection is crucial to mitigate these risks and preserve societal trust. Cognitive load is a useful indicator for detecting deception, as lying causes individuals to experience greater mental strain. While prior research leveraged cognitive load features, typically measured through physiological signals such as pupil dilation, these methods often require specialized equipment and can be subject to human bias. These limitations hinder the scalability and automation of deception detection systems. Thus, we propose a novel deception detection framework that automatically extracts cognitive load features from audio-visual data, eliminating the need for specialized hardware or subjective human input. Our approach integrates these features into the deception detection pipeline, enhancing its robustness. Moreover, we introduce a focal loss to address the inherent complexity of deception detection. This objective function enables the model to focus on harder-to-detect instances of deception, thereby improving the performance. Our approach achieves state-of-the-art results on benchmark audio-visual datasets, demonstrating significant improvements in automated deception detection. Extensive experiments validate the effectiveness of both our cognitive load feature extraction and the proposed objective function in advancing the field.
Supplementary Material: zip
Primary Area: applications to neuroscience & cognitive science
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 7974
Loading