Facial Behavioral Analysis: A Case Study in Deception Detection

Abstract

Aims: To establish a rich Facial Action Coding System (FACS) coded database and to investigate the use of the facial visual cues for deception detection. Study Design: A within-participants design experiment was conducted, using immigration as a scenario for asking questions of participants in controlled experimental conditions. The study design required participants to answer questions on two topics, one as themselves and one based on a learned scenario. Data regarding visible images of facial movement were collected and analyzed against cues identified as indicative of deceit. Place and Duration of Study: With the ethical approval from the University of Bradford, 32 volunteer undergraduate students and research assistants took part in the study, from March 2011– June 2011. Methodology: We included 32 participants (27 men, 5 women; age range 18-33 years). The experiment was constructed as two interview scenarios. Participants were interviewed by an ‘Examiner’ who was introduced by the ‘Facilitator’ as having recently trained in techniques to detect lie. The participants were told it was important that they appear honest throughout. For one session, they were asked to answer questions as themselves. For the other, they were given a character profile to learn and were asked to answer the questions as if they were the character in the profile. Some questions went beyond the information in the profile, requiring participants to create plausible answers. A rich Facial Action Coding System coded (FACS-coded) database was established for further analysis. Results: The Examiner’s score is 56.25% in both sensitivity and specificity. The best classification algorithm for our FACS-coded database was Logistic Regression with a sensitivity of 47.9% and a specificity of 71.2%. The findings revealed that the machine learning was biased towards truth prediction. In order to increase the sensitivity of deceit prediction, the threshold of classification was adjusted and the improved result indicates sensitivity of 70.0% and specificity of 63.3%. Conclusion: Our research established a rich FACS coded database which is expected to be important for future research development. In order to increase the detection rate, we showed that it is worthwhile to consider machine learning algorithms to aid human decision.