‘A simple heuristic for distinguishing lie from truth’
A few days ago, the forensic psychologist Bruno Verschuere and a group of co-authors published a paper on lie detectionin Nature Human Behavior, which received ample attention in Dutch (1 2 3) and Belgian (1 2) popular media. After all, we all want to know how to detect lies!
People (including police officers and judges) tend to be very bad at detecting whether someone is lying or speaking the truth. Even if they attend to all sorts of supposed clues (such as facial expression, eye contact, voice stress, gesturing and more), they are hardly better at it than a monkey pushing random buttons. So instead of using this variety of techniques to detect lies, Bruno Verschuere and colleagues argue, people should use only one, the best one: attending to the level of detail in a statement. People who speak the truth generally give much more detailed reports of their doings than people who are lying. If people only look at this specific aspect, the authors find, they are significantly better at detecting lies (66% accuracy in one of their experiments).
The novelty of their argument is in the claim that attending to one aspect (level of detail) is better than combining several means to assess whether someone is lying. But what interests me is that the technique itself has been around for centuries, and the method for testing it for decades. And so have their critics.
Premodern lie detectors
Detecting lies has been the core business of criminal investigation since there was such a thing as criminal investigation. Already in the first manuals for inquisitors tasked with finding heretics, mostly written between the thirteenth and fifteenth centuries, clarity and detail were mentioned as the keys to a good interrogation. You know you’re dealing with lying heretics, Bernard Gui wrote in the 1320s, “when they do not answer clearly and directly, when, in order to answer besides the question, they seek obscurities and subterfuge” .
Conversely, detailed answers guaranteed that what a suspect said was the truth. This is why many early modern interrogation manuals counseled to ask for and record every detail when a suspected confessed. These details could be verified to prove that the confession was true, thus guarding justice against both false confessions or later retractions .
Details were not the only means early modern judges possessed to assess whether someone was lying. As today, they were aware of many physical signs people could display that could indicate their guilt, even if they were denying. Blushing, turning pale, trembling, gesturing: all these corporeal behaviours could reveal guilt . At the same time, also like today, many scholars warned against relying too much upon such signs. In the beginning of the nineteenth century, for instance, Jeremy Bentham outlined how common symptoms of guilt could be a result of both a bad conscience, or of fear of being unjustly condemned .
Forensic psychologists enter the fray
The authors of interrogation manuals up to the nineteenth century differed from Bruno Verschuere and his colleagues in many respects. For one, they were more concerned with innocence or guilt than with truth or falsehood. The two often aligned, but they are not the same. Moreover, they mainly drew upon their extensive ‘experience’ in the field, rather than on controlled experiments. That changed around the beginning of the twentieth century.
Around 1900, several mostly German legal scholars and practicing magistrates took an interest in the developments in the upcoming field of experimental psychology. They read about psychologists’ experiments on attention and memory, collaborated with them and devised their own experiments to try out new methods of interrogation. Much like Verschuere et al, they found willing test subjects among their students.
‘Forensic psychologists’ tried out all sort of techniques to detect guilt: association tests, hypnosis, instruments that measured blood pressure or breathing rate. By the 1920s, US psychologists started to focus on the detection of lying rather than guilt. This had more extensive applications (marriage counseling, for instance). From these experiments, the ‘lie detector’ was born, a ‘polygraph’ that combined multiple measuring instruments .
More research is needed
As soon as the results of these experiments started to be published and new techniques were recommended for application in legal practice, the critiques came. The difference between a guilty and an innocent person, between a lying and a truthful suspect, was not absolute, the German psychologist William Stern objected in 1904. Experience (and Jeremy Bentham) had shown that people who were falsely accused often exhibited the same symptoms as guilty people, for instance out of fear. There was no reason to assume that this would not be the case with the new, less visible symptoms that machines or association tests revealed .
Moreover, students were not (usually) criminals, pretend crimes were not real crimes and experiments were not criminal investigations. There was very little at stake for the test subjects, and very much for real people involved in criminal investigations. This might influence their behaviour, their memory and their attitude .
Finally, the technique was mostly tested on people who did not know how it worked. After Carl Jung boasted of his success in using an association test that showed that a young servant was guilty of theft, Max Lederer was sceptical. What does this tell us about the use of this method in criminal practice? ‘Nichts, gar nichts!’ The young suspect did not even know that he was suspected of anything. Once people know how a test works, they adapt their behaviour .
In short, while there were many enthusiasts for using psychological techniques to detect lies or guilt, its applications in criminal justice remained contentious. The criticisms voiced against experimental techniques around 1900 still resonate today, as Verschuere et al recognize in their article: classrooms or laboratories are not interrogation rooms; experimental conditions are not real life . It remains to be seen whether the techniques is reliable in practice.
The long history of the attention for detail as a tool for detecting lies or the history of experimental psychology and its critiques do not invalidate the findings of the new study. They do perhaps curb our enthusiasm. In the hundreds of years of criminal investigation, in the decades of experimental research on lie detection, no single reliable technique has been found that can help us in all sorts of circumstances. The search for a ‘simple heuristic’ to detect lies seems never-ending. Guilt and lying are concepts that cover such diverse practices that I’m optimistic that we will never find it.
Notes Gui Bernard. Manuel de l’inquisiteur, ed. G. Mollat, Paris, 1964, 155-57.  E.g. Jousse, Daniel. Traité de la justice criminelle de France. Paris: Debure, 1771, volume 1, 677-8; Brunnemann, Johann. Anleitung zu vorsichtiger Anstellung des Inquisitions-Processes. Halle: Renger, 1697, 23.  See Hofman, Elwin. “Corporeal Truth: Conscience, Fear and the Body in French Criminal Interrogations, 1750-1850.” Cultural and Social History 18, no. 1 (2021): 61–78. https://doi.org/10.1080/14780038.2020.1752107.  Bentham, Jeremy. Traité des preuves judiciaires. Edited by Étienne Dumont. Paris: Bossange, 1823, volume 1, 362-5.  See Bunn, Geoffrey C. The Truth Machine: A Social History of the Lie Detector. Baltimore: Johns Hopkins University Press, 2012.  Stern, William. “Psychologische Tatbestandsdiagnostik.” Beiträge zur Psychologie der Aussage 2, no. 2 (1904): 276.  See a summary of several debates on this in Lipmann, Otto. Die Spuren interessebetonter Erlebnisse und ihre Symptome. Theorie, Methoden und Ergebnisse der ‘Tatbestandsdiagnostik.’ Beihefte zur Zeitschrift für angewandte Psychologie und psychologische Sammelforschung 1. Leipzig: Johann Ambrosius Barth, 1911,
81-5.  Lederer, Max. “Die Verwendung der psychologischen Tatbestandsdiagnostik in der Strafrechtspraxis.” Monatsschrift für Kriminalpsychologie und Strafrechtsreform 3 (1906): 164.  ‘Our findings should be extended to more realistic settings, including self-chosen and/or high-stake lies. Raising the stakes will also increase the odds that liars attempt to alter their message to increase its credibility, for example, by enrichen their lies with details. Second, we demonstrated the success of the use-the-best heuristic only in the particular context where statements are about episodic memory and truth tellers were both willing and able to provide specific details. Cues other than detailedness may be more valid in different contexts, and a context-contingent approach may provide guidance on which cues will be most efficient in other contexts such as the detection of fake news.’ (Verschuere et al., p. 5).
Leave a Reply