A New Lie Detection Method Based on Small-number of P300 Responses

A large number of stimuli were often used for identify the guilty subjects in lie detection. In this paper, we proposed a novel lie detection approach to enhance the identification accuracy of the subjects. The independent component analysis (ICA) was used to separate the EEG signals, and a reconstruction algorithm was presented to recognize the P300 components automatically, and then reconstruct the P300 in the Pz electrode. We extracted two groups of features based on time-, frequency-domains from the reconstructed Pz signals. SVM was employed to classify the two kinds of feature vectors. The balance test accuracy of 82.45% suggests that the presented method could effectively detect the deceptive and truthful responses. In addition, the related result shows that identifying the subjects in lie detection could be implemented based on small-number P300 responses using the proposed method.