Lie Detection: Can Technology Uncover Deception?
Lying is a part of life that we cannot evade. Whether it is intended or a white lie, we come across lying every day. Evolution didn’t provide us with some kind of protection from being deceived. Though people can randomly detect that their interlocutor might conceal something, scientists work on technologies that will make it easier. The accuracy of a polygraph test that’s been around for years has for long been debated, so there must be some alternative.
In our new series of articles, we decided to deal with the notion of lie and talk about how technologies can uncover if somebody is trying to deceive you.
Can we recognize a liar without any technology?
Well, there are some things you might consider when you suspect that someone is not sincere with you. Researchers claim that a liar’s speech has distinct linguistic patterns mostly reflected in lexicology and grammar [1]. For example, liars tend to use simple sentences and avoid specific facts [2]. They can use a lot of vivid descriptions to set the atmosphere but avoid the facts that can be verified. To distance themselves from their lie, there will be a lot of negations, emotional and obscene words. Well, there will be a lot of words at all. (Or maybe quite the opposite. Liars understand that providing too much false information will increase the chances of being caught. This is called the liars dilemma hypothesis [2].
Whether we truly can rely on linguistic features is hard to say. But according to the theory of lie detection — ALIED [3] — if we have reliable clues that someone is lying or not, we follow them and chances are we are more likely to detect what is going on. However, when there are no clues like these we make a rational assumption based on our understanding of the situation. In our opinion, the way we understand the situation is rational and relying on these assumptions is right, however, we cannot predict the future or realize how our thinking actually fits the context. Therefore, we can misplace or even ignore the context, making the probability of detecting a lie lower.
When trying to spot a liar, you can also take into account these statistics: highly intelligent extroverts are likely to behave deceptively, while highly intelligent introverts as well as extroverts with lower intelligence are the least likely to fool around. While personality traits define common behavioral trends, it is the person’s intelligence and the environment that set the boundaries of his or her behavior.
When it comes to numbers, the data is quite disturbing. “Is this man a liar or is he telling the truth?” — meta-analyses of many studies have shown that the average accuracy of a lie detection in person is 54% (just a bit more than a chance would be), correctly classifying 47% of a 100% lie as a lie and 61% of a 100% truth as a truth. [4] This means that we can recognize if somebody is lying, but this assumption will be unreliable.
Ol’ good polygraph
The traditional polygraph is almost 100 years old. It was introduced in 1939 and was accompanied by respiration tracking at the time. As it was believed,changes in breathing patterns were a sensitive indicator of stress and emotional changes. The accuracy of a polygraph test is based on many external factors, but in laboratory conditions it can reach up to 90%. [5]
A traditional lie detection procedure that aims to successfully identify deception consists of two crucial components: a polygraph and a polygraphist. According to American Psychological Association (APA), polygraphs can record several physiological indicators, e.g. heart rate, respiration, skin conductivity (to measure sweating). Measurements of skin galvanic response show changes in our reaction to stress, but it is unclear whether it is psychological, physiological or emotional. A polygraphist then goes through a very elaborate questioning procedure and matches the answers to people’s physiological reactions.
However, the link between physiological reactions and indicators of dishonesty has not been confirmed and there is a way to cheat during the test, or to interpret physiological responses as false positives. What’s more? The tests are not always conducted under proper conditions.
Nervousness, fear and emotional imbalance, these very conditions can be measured by the polygraph, but there are still no physiological indicators that can be attributed to lying. A classic lie detector is more likely to be called a fear detector, but not all criminals are worried about their harmfulness of their actions. Besides, there are now many techniques that can be used to bypass the test.
Anything new in lie detection?
In 100 years the technology has moved forward. There are not only methods that presuppose contact but also new remote tools that can become groundbreaking for the industry. All of them deal with the human physiology or brain analysis.
Unlike polygraph, brain-based lie detection can be even more accurate. When we lie, we intentionally conceal the information we know. Turns out the brain reacts differently to the known and unknown and the measurement of brain activity in such moments can advance the accuracy of lie detection to up to 90%.
Brain reactions are so subtle and nuanced that by analyzing them during crime interrogation, it is even possible to determine whether the suspect knew if the victim was killed with a knife or a bullet. The main difficulty is: if the details of the case are made public, then everyone who was interrogated would know what was the murder weapon, and it will be impossible to predict guilt.
Physiological reactions such as pulse, respiration rate, and even changes in skin resistance can be learned to change consciously, but people have not yet learned how to control the blood flow to different brain regions. That is why the idea of detecting lies using neuroimaging techniques remains very attractive to scientists. One good example — lying affects the amount of oxygen that goes into different parts of the brain. These changes can be registered with functional magnetic resonance imaging (fMRI) [6].
The main issue with brain-based lie detection though is that each particular situation of deception consists of many variables, such as motivation to lie, remorse or shame, or maybe a complete absence of these. A person may be afraid that people will find out the truth and there will be consequences and punishment. Otherwise, he may be completely indifferent to it. These might lead to potential false positives. Expensive tomography or EEG studies also require highly qualified staff to collect, analyze and interpret brain data, and the guarantee that the guilty person’s specific gyrus are being activated is quite low. And it is probably not the case that all people have the same neural structures responsible for lying, thus it might take years to spot one and only deception neural network, if such exists.
Contactless methods for lie detection
Generally, lying is considered to be an intensive and negative experience that takes a lot of cognitive effort. The process of lying can cause stress on the liar himself that becomes visually noticeable. The technique called thermal imaging, or thermography, is capable to track these manifestations.
Thermography has recently been introduced to the lie detection industry. The face and body of a person are recorded on an infrared camera to track changes in skin temperature.
Remember Pinocchio? Yes, that wooden guy whose nose grew when he was lying. Turns out, it might not really be a fiction. Lots of studies mention the so-called Pinocchio effect. When we lie, the temperature of the tip of the nose drops, and the forehead on the contrary becomes warmer. [7]
Why the forehead temperature might be so revealing? There is an assumption that at the increased level of stress associated with deception, humans involuntarily frown eyebrows, which is accompanied by an influx of blood and a corresponding increase in the forehead temperature. Researchers report that for 38 subjects, the forehead deception detector yields 76.3% success rate. [8]
Thermography is primarily more convenient than a polygraph test because it is a contact‐free method to record the response of the autonomic nervous system. The suspect won’t even know that he is being recorded and its average accuracy may range between 70% and 90%.
This method of lie detection can be quite reliable, but there is still a problem of acquiring expensive thermal cameras with high temporal and spatial resolution to track the smallest temperature fluctuations in different areas of the face. In addition, there is a hypothesis that for the people who do not feel remorse (with a high level of Machiavellism and/or psychopathy) the body’s physiological responses to deception will be minimal and unidentifiable [9].
Conclusion
Alternative methods of lie detection have a number of drawbacks. There is no universal checklist to determine the cheater. The perpetrator’s pulse rate increases, blood pressure rises, respiratory rhythm is disturbed, the eyebrows become frowned upon, the eyes run, some areas of the brain (the anterior prefrontal cortex and cingulate gyrus) are activated, the temperature of the nose changes, the voice trembles, and the pupil dilates. Or not.
Physiological parameters depend on many uncontrollable factors that cannot be accommodated in a simple experimental technique. Perhaps AI will be able to solve the mystery. A neural network will analyze thousands of true and false fragments in different conditions and find regularities that people have not been able to detect.
However, we must not forget that lying is contextual. It will be hard to create some universal technology that, like a magic wand, would recognize a lie in all its manifestations. It’s more possible that there will be lots of algorithms with a specific set of applications. But only time can tell.
References
- Evidence for the Pinocchio effect: Linguistic differences between lies, deception by omissions, and truths. Discourse Processes (2012), Van Swol, L. M., Braun, M. T., & Malhotra, D
- Digging Further Into the Speech of Liars: Future Research Prospects in Verbal Lie Detection (2019), Galit Nahari and Zvi Nisin
- ALIED: A THEORY OF LIE DETECTION (2016), Chris N. H. Street, Ciencia Cognitiva
- Accuracy of deception judgments. Personality and social psychology Review, 10(3), 214–234) (2006), Bond Jr, C. F., & DePaulo, B. M.
- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4873061/
- Historical Techniques of Lie Detection (2015), Martina Vicianova
- The mental nose and the Pinocchio effect: Thermography, planning, anxiety, and lies (2018), A. Moliné, E. Dominguez, E. Salazar‐López, G. Gálvez‐García, J. Fernández‐Gómez, J. De la Fuente, O. Iborra, F.J. Tornay, E. Gómez Milán
- Forehead Thermal Signature Extraction in Lie Detection (2007), Zhen Zhu, Panagiotis Tsiamyrtzis and Ioannis Pavlidis
- Thermal signatures of voluntary deception in ecological conditions (2013), Maria Serena Panasiti, Daniela Cardone, Enea F. Pavone, Alessandra Mancini, Arcangelo Merla, Salvatore M. Agliotib
Authors: Valeria Karpova, Research Scientist at Neurodata Lab, Nadezhda Glebko, Research Scientist at Neurodata Lab, Elizaveta Zaitseva, SMM Specialist at Neurodata Lab.
