EEG-based deception detection using weighted dual perspective visibility graph analysis.

Deception detection is a critical aspect across various domains. Integrating advanced signal processing techniques, particularly in neuroscientific studies, has opened new avenues for exploring deception at a deeper level. This study uses electroencephalogram (EEG) signals from a balanced cohort of 22 participants, consisting of both males and females, aged between 22 and 29, engaged in a visual task for instructed deception. We propose a novel approach in the realm of deception detection utilizing the Weighted Dual Perspective Visibility Graph (WDPVG) method to decode instructed deception by converting average epochs from each EEG channel into a complex network. Six graph-based features, including average and deviation of strength, weighted clustering coefficient, weighted clustering coefficient entropy, average weighted shortest path length, and modularity, are extracted, comprehensively representing the underlying brain dynamics associated with deception. Subsequently, these features are employed for classification using three distinct algorithms: K Nearest Neighbors (KNN), Support Vector Machine (SVM), and Decision Tree (DT). Experimental results reveal promising accuracy rates for KNN (66.64%), SVM (86.25%), and DT (82.46%). Furthermore, the features distributions of EEG networks are analyzed across different brain lobes, comparing truth-telling and lying modes. These analyses reveal the frontal and parietal lobes' potential in distinguishing between truth and deception, highlighting their active role during deceptive behavior. The findings demonstrate the WDPVG method's effectiveness in decoding deception from EEG signals, offering insights into the neural basis of deceptive behavior. This research could enhance the understanding of neuroscience and deception detection, providing a framework for future real-world applications.