Introduction
The polygraph, more often referred to as a “lie detector,” has been used for many decades to determine whether a person is speaking the truth about anything. However, the reliability of the polygraph test has been the subject of debate and controversy, with some experts arguing that the test is not a reliable means of detecting deception. In contrast, others claim that it is an accurate method. This annotated bibliography provides an overview of current research that looks at how successful the polygraph is in determining truthfulness.
Barsever, D., Singh, S., & Neftci, E. (2020, July). Building a better lie detector with BERT: The difference between truth and lies. In 2020 International Joint Conference on Neural Networks (IJCNN) (pp. 1–7). IEEE. doi: 10.1109/IJCNN48605.2020.9207005
In this study, we investigate whether or not using BERT, a pre-trained language model, can enhance the precision of lie detection. The authors contend that their method, which makes use of NLP and ML, has the potential to be more accurate than standard polygraph examinations. The research results show that BERT is superior to standard polygraph testing, demonstrating the promise of advanced machine learning methods in the quest for more accurate lie detection.
Competence in identifying dishonest or misleading writing is very desirable (Barsever et al., 2020). It is partly since the underlying patterns of misleading text are still poorly understood. On the Ott Deceptive Opinion Spam corpus, BERT achieves higher accuracy in deception classification than previous state-of-the-art methods. This ablation study’s findings suggest that certain aspects of the input, such as specific grammatical constructions, are more informative to the classifier than others. Deceptive writing is more formulaic and less diverse than genuine material, according to part-of-speech analysis in “swing” phrases deemed critical to BERT’s categorization (Barsever et al., 2020).
Bradshaw, R. (2021). Deception and detection: the use of technology in assessing witness credibility. https://academic.oup.com/arbitration/article-abstract/37/3/707/6174512
Bradshaw (2021) explores the use of technology in determining the reliability of witnesses in court proceedings. The author examines the history of the polygraph, voice stress analysis, and eye-tracking equipment, all created for this reason. However, the article cautions against placing too much stock in technological aids alone when evaluating the credibility of witnesses. In order to reach more precise judgments, technology should be utilized in tandem with other forms of evidence, such as witness testimony and physical evidence. Concerns about privacy invasion and false positives are only two examples of the ethical issues raised by the article’s discussion of the use of technology in witness credibility evaluations. The use of technology to determine a witness’s credibility is scrutinized here, and serious ethical questions are raised.
It has traditionally been a challenge for tribunals to identify reliable witnesses (Bradshaw, 2021). Tribunals of the Court of Arbitration for Sport have disagreed on whether or not polygraph evidence is acceptable to authenticate witness testimony (Bradshaw, 2021). A new generation of ‘lie detectors’ based on eye tracking, artificial intelligence, and brain imaging are now being tested by authorities in several nations. Proponents claim that these technologies can revolutionize the reliability of witness testimony since they are more precise and less subjective than conventional polygraphs. This article discusses the possibility of lie-detection technology to completely transform how arbitration tribunals assess the credibility of witnesses’ credibility. It analyzes the arguments against using lie detectors. It finds that justice and proportionality favor eliminating such evidence due to concerns about reliability, machine bias and privacy, and the right against self-incrimination.
Bhamare, A. R., Katharguppe, S., & Nancy, J. S. (2020, November). Deep Neural Networks for Lie Detection with Attention on Bio-signals. In 2020 7th International Conference on Soft Computing & Machine Intelligence (ISCMI) (pp. 143–147). IEEE DOI: 10.1109/ISCMI51225.2020.9311575
This research discusses how deep neural networks may be combined with bio-signals to create an effective liar detector. Findings from this research imply that deep neural networks and bio-signals might be employed to enhance the precision with which lie detection is carried out. More research is needed to determine the efficacy of this approach, as the study does not compare the performance of this approach to traditional polygraph tests.
Researchers have found time and time again that the human capacity to detect fraud is no better than chance (Bhamare et al., 2020). Accurate lie detection is essential for law enforcement. The traditional polygraph exam has several things that could be improved. A skilled con artist may easily fool it. Faking or controlling one’s micro-expressions is challenging. They may happen as quickly as 1/15 of a second (Bhamare et al., 2020). Another trait that may be exploited to expose a liar is their voice. Due to vocal cord tension, they often raise or lower their voice pitch or take a while to respond. The answer proposed in this study is substantial. It is a deep learning model for spotting deceit and uses facial and auditory cues. The suggested model is a sophisticated lie detector. For instance, it may help HR professionals and law enforcement authorities when conducting interviews and interrogations for businesses.
Doddamani, I., Patel, R., Singh, A. K., & Mane, R. Lie Detection Using Facial Expressions. https://d1wqtxts1xzle7.cloudfront.net/98784697/IJCRT2204164-libre.pdf?1676637652=&response-content-disposition=inline%3B+filename%3DLie_Detection_Using_Facial_Expressions.pdf&Expires=1683249735&Signature=HhndSmZSEvwK2~p~xP9fIyNnr6Uapl6-09zRpm5Mqz5G5nJodHAc5IXgTK3kPRMP3wQ4OW0GtPyOJNmlKhkz6kFWNZBQDB7RgcgN3cu-K5g1ZmxSgiP3Y7lXGFhGptRH5OK39F3x4ZD15-9zOSxJZj6Z4dx7pZeSQrT3sPChoIutjLvgcCL8q35SOAvO2mukBggsxs~nlD1er2Tqc1TcNp~44XhsNJewxebCVI4TXmpzHuOa6CbyhmLpdsNJapMAwiWCcKJfwLlJAP4GGf7e~ZYkEoLAOjFOe162zS-jzV5AiSj5-YRzMq4f-8dNRC-JuNttmKYEmNGB8iPPC3yqOA__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA
The possibility of utilizing facial expressions to detect lying is investigated in this research. The authors propose a technique for evaluating facial expressions that may be used to identify liars reliably. The research shows this method may be more effective than regular polygraph testing at identifying deception. However, the study’s findings must be further validated and replicated, and its sample size is relatively small.
Many fields, including airport security, police investigations, and anti-terrorism efforts. Rely on the ability to detect deception, as the paper explains. The only way to know for sure is to learn to recognize individuals’ brief facial speeches when attempting to conceal or repress their feelings. Measuring microexpressions manually is a laborious, error-prone process. Subtle changes in facial expression and body language Doddamani et al. (n.d.) argue that speech analysis effectively detects deception. Facial micro-expressions are a lie-proof, impossible-to-participate-in participatory reaction. Both nonverbal and verbal cues reveal whether a person is lying. Paul Ekman’s studies employing Key Analysis have led to the discovery and understanding of subtle alterations in facial expression. This research presents a cheap and painless way to identify liars among those who constitute a risk to society.
A parametric representation of the face was derived by analyzing the parameters over several frames, which may be valuable for the static analysis of facial expressions in various research domains. Using this motion with a physical model, templates based on geometric dynamics and efficiency are then applied to the face. Human emotion is also a significant area of study in psychology. Airport and homeland security, police interrogations, clinical and employment examinations, and many other areas may benefit from the proposed system (Doddamani et al., n.d.).
Nortje, A., & Tredoux, C. (2019). How good are we at detecting deception? A review of current techniques and theories. South African Journal of Psychology, 49(4), 491-504. https://journals.co.za/doi/abs/10.1177/0081246318822953
Nortje and Tredoux (2019) reviewed the literature to assess the effectiveness of existing methods for lie detection. Psychologists have long been interested in developing reliable strategies for telling the difference between fact and fiction. This article reviews and assesses several techniques used in the lab and field to detect deceit. Measurement of non-verbal behavior, verbal interview techniques, and statement assessment by humans and computers are the authors’ three overarching methodologies. Insufficient knowledge of human deception contributes to the difficulty of developing effective methods for detecting it. In this work, we examine these three theories of deceit and find that although they show promise, they still need to offer a solid basis. In the article’s second section, the writers examine ten different lie-detection techniques, although they devote the most space to the polygraph, the most used technique in South Africa. Statement analysis and other approaches for detecting lies are discussed. However, the authors conclude that more than the present methods are needed in practice (Nortje & Tredoux, 2019).
The authors observed that there is still substantial controversy about the efficacy and accuracy of different approaches to lie detection, even though numerous research has evaluated them. The methods discussed in this overview vary from the polygraph through verbal cues to nonverbal clues to physiological assessments. The authors emphasize the need for a more holistic approach to detecting deceit, which uses various methodologies, and point out the limits of these individual strategies. The paper closes with suggestions for where the field may go from here. By evaluating existing methods and ideas for lie detection rigorously and pinpointing research needs, this paper makes a significant addition to the field.
Conclusion
According to the findings of the studies that were analyzed to compile this annotated bibliography, contemporary machine learning methods, such as deep neural networks and natural language processing, can potentially enhance the reliability of lie detection. However, additional research is required to verify these findings and establish whether or not they can be implemented in clinical settings. In addition, before the polygraph or any other tool for lie detection is employed in any situation, the legal and ethical concerns involved with its usage and any other techniques for lie detection need to be thoroughly reviewed. In the end, even though the polygraph may be helpful in some circumstances, it should be treated with care and utilized in concert with other evidence to ensure that the truth is determined accurately.
References
Barsever, D., Singh, S., & Neftci, E. (2020, July). Building a better lie detector with BERT: The difference between truth and lies. In 2020 International Joint Conference on Neural Networks (IJCNN) (pp. 1–7). IEEE. doi: 10.1109/IJCNN48605.2020.9207005
Bradshaw, R. (2021). Deception and detection: the use of technology in assessing witness credibility. https://academic.oup.com/arbitration/article-abstract/37/3/707/6174512
Bhamare, A. R., Katharguppe, S., & Nancy, J. S. (2020, November). Deep Neural Networks for Lie Detection with Attention on Bio-signals. In 2020 7th International Conference on Soft Computing & Machine Intelligence (ISCMI) (pp. 143–147). IEEE DOI: 10.1109/ISCMI51225.2020.9311575
Doddamani, I., Patel, R., Singh, A. K., & Mane, R. Lie Detection Using Facial Expressions. https://d1wqtxts1xzle7.cloudfront.net/98784697/IJCRT2204164-libre.pdf?1676637652=&response-content-disposition=inline%3B+filename%3DLie_Detection_Using_Facial_Expressions.pdf&Expires=1683249735&Signature=HhndSmZSEvwK2~p~xP9fIyNnr6Uapl6-09zRpm5Mqz5G5nJodHAc5IXgTK3kPRMP3wQ4OW0GtPyOJNmlKhkz6kFWNZBQDB7RgcgN3cu-K5g1ZmxSgiP3Y7lXGFhGptRH5OK39F3x4ZD15-9zOSxJZj6Z4dx7pZeSQrT3sPChoIutjLvgcCL8q35SOAvO2mukBggsxs~nlD1er2Tqc1TcNp~44XhsNJewxebCVI4TXmpzHuOa6CbyhmLpdsNJapMAwiWCcKJfwLlJAP4GGf7e~ZYkEoLAOjFOe162zS-jzV5AiSj5-YRzMq4f-8dNRC-JuNttmKYEmNGB8iPPC3yqOA__&Key-Pair-Id=APKAJLOHF5GGSLRBV4ZA
Nortje, A., & Tredoux, C. (2019). How good are we at detecting deception? A review of current techniques and theories. South African Journal of Psychology, 49(4), 491-504. https://journals.co.za/doi/abs/10.1177/0081246318822953
write