Cutting through the myths and flawed research within the field of body language and behaviour analysis
Social media and conflicting research in body language prompted me to help researchers, and organisations cut through the myths floating around – many based on research with flawed logic. There seem to be four popular myths which I have posted on LinkedIn and assembled here for ease. These myth-busting statements have triggered some interesting comments on social media (use #BodyLanguageMythBuster on LinkedIn if you wish to review or add to the commentary).
Myth 1
The reliability of humans in detecting deception from body language is only 54%, no better than chance, so you can’t detect lies from body language.
Myth 2
We can make truth/lie decisions based on micro-facial expressions and body language.
Myth 3
Facial expressions and basic emotions are neither linked nor universal.
Myth 4
Micro-facial expressions of emotion don’t exist because I have never seen them.
I have added ‘endnotes’ to each section where needed, using different formats per section to prevent overlap.
Myth 1: The reliability of humans in detecting deception from body language is only 54%, no better than chance, so you can’t detect lies from body language.
This is a ‘straw-person argument’ [1] that has been used in many articles and publications, often by individuals with an agenda to sell a different theory of basic emotions or lie detection. It is misleading. The source of this “54%” statistic is a research publication by Bond & DePaulo (2006)[2]. This was a meta-study that considered 206 documents and 24,483 individuals and their “attempt to discriminate lies from truths in real-time with no special aids or training. In these circumstances, [average[3]] people achieve an average of 54% correct lie-truth judgments” (ibid:214).
So this 54% does not represent what levels of accuracy are possible from people trained in scientifically validated approaches to lie detection. The study deliberately excluded those “studies in which individuals could incorporate into their judgments systematic aids to lie detection (e.g., polygraph records, Criterion-Based Content Analysis, or behaviour codings from repeated viewings of a videotape)” (ibid:218).
Trained behaviour analysts using SCAnS (Six Channel Analysis System) identified 27 research-validated Points of Interest (‘PIns’) that occur more frequently from those who are lying in high-stake crimes than those who were truth-tellers – “the number of PIns identified by the four coders which averaged 11.25 when it came to the deceptive accounts (within a range from 9 to 15) and only 1.5 for Sharp (the truthful account) over similar time periods (within a range of 1 to 2).” (Archer & Lansley, 2015)[4]. Training based on SCAnS was deployed with 23 security/intelligence professionals in Europe by the EIA Group and increased the accuracy of the attendees from 55.1% to 84.67% after only 72 hrs training.[5]
Endnotes Myth 1:
[1] An argument (originally labelled ‘straw-man’) that fails to address the proposition in question by intentionally misrepresenting the opposing position.
[2] Bond CF Jr, DePaulo BM. Accuracy of deception judgments. Pers Soc Psychol Rev. 2006;10(3):214-34. doi: 10.1207/s15327957pspr1003_2. PMID: 16859438.
[3] ‘Average’ described the main sample group for this meta-study. From the Bond & DePaulo paper, page 230 it is stated that 88.1% of the ‘receivers’(judges) were students; 11.9% were classified as ‘experts’. This means they “have a background researchers deem relevant to detecting deception. They include police officers. detectives. judges. interrogators, criminals. customs officials. mental health professionals. polygraph examiners, job interviewers. federal agents, and auditors.” (ibid:230).
[4] “Public appeals, news interviews and crocodile tears: an argument for multi-channel analysis | Corpora” (2022), p. Available at: https://www.euppublishing.com/doi/full/10.3366/cor.2015.0075 (Accessed: 6 August 2022).
[5] The Impact of Behaviour Training in High-Stake Airport Contexts (2016). Available at: https://www.eiagroup.com/airport-security-training/ (Accessed: 6 August 2022).
Myth 2: We can make truth/lie decisions based on micro-facial expressions and body language?
No, we can’t. Micro-facial expressions are a reliable indicator of what a person may be feeling because they are activated by the 7th cranial nerve that comes from the midbrain. This means that they occur below consciousness. They are ‘micro’ (appearing for less than half a second on the face) often because we are choosing to suppress the display once we sense the muscle movement on the face, for example, when we want to hide what we are feeling; for privacy, as a social lubricant, or when we are trying to mislead others.
The problem with relying on the face alone in behaviour analysis is that it is unreliable as a single indicator. There is no reliable single indicator of deception. At the Emotional Intelligence Academy, we have evidenced that we need 3 indicators (from 27 reliable cues) across 2 or more of the 6 channels (face/body/voice/psychophysiology/verbal style/verbal content) within 7 seconds of a stimulus or probe. The 3/2/7 rule. Our peer-reviewed research has proven this is crucial in high-stake lies (see https://lnkd.in/gx9W7fe ).
We need to corroborate the hypothesis of ‘deception’ with signals from the body or words that are equally strong but only reliable when they occur as a cluster of 3.
The same goes for those who only rely on statement analysis (the words), or those who only use body language (movement of hands/body/head/legs/eyes). We need a Six Channel Analysis System (SCAnS) to be confident in our behaviour analysis. This is why we have had no rebuttals having used it extensively with intelligence, law enforcement, military and business clients. Nor have we had any challenges or litigation from our conclusions on 60+ murder cases documented publically on the Discovery Channel. (https://lnkd.in/e2tCa7AV ).
Myth 3: Facial expressions and basic emotions are neither linked nor universal.
Psychologists have been in agreement about basic emotions and how they affect the body and motivate behaviour. Over the last decade or so, however, we have seen articles and criticisms of the Basic Emotion Theory (BET)[i], particularly from James Russell and Lisa Feldman Barratt, arguing an alternative ‘Constructivist Theory of Emotion (CTE)[ii]. Ekman responded with a rebuttal in 2014 “so that the public is not misled”.[iii]
So, what do the broader scientific community agree on? To answer that question, Paul Ekman conducted a peer-reviewed study in 2016[iv]. He engaged Russell to check the selection of consultees was free from bias and to also vet the survey questions. The study revealed that the ‘existence of “compelling evidence for universals in any aspect of emotion” was endorsed by 88% of the respondents. The evidence supporting universal signals (face or voice) was endorsed by 80%.’ (Ekman, 2016:32). There was high agreement (from 74% of those surveyed) that five basic emotions have been empirically established… “anger (91%), fear (90%), disgust (86%), sadness (80%), and happiness (76%).” (ibid:32).
What is muddying the water?
It may be the fuzzy logic that we see in some articles. This includes:
- Confusing universality of facial expression from felt emotions with the lack of universality in how people pose/signal emotions and how they perceive posed emotions (often from actors).
- Assuming the face can only be manipulated by emotions – and not weaning out conscious gestures and actions (screaming/crying/shouting/running/jumping) that might follow emotions or co-occur with them.
- (Mis)-judging photos out of context – e.g., comparing surprise with putting on eyeliner. Seriously!
- Assuming the expression must match the activity – e.g., winners don’t always smile (they might be sad, thinking about their late parent who isn’t there to see the victory).
- Not recognising that ‘brows-down’ isn’t always ‘anger’ – e.g., it might be about cognitive load, squinting, or pain.
- Language – differences in the labels/words or language used to compare emotions across cultures to ensure we compare apples with apples.
The last point of language needs careful attention. Facial expressions associated with American anger (often viewed as constructive if directed at an offensive action or principle [to them] to drive change) cannot be compared to the Tibetan counterpart, ‘lung-lang’, which is viewed as “a fundamentally destructive sentiment, equally harmful to self and others” (Shweder et al., 2008:416).[v] So, this is not a cultural difference of the same emotion. We must be careful around variations in what makes different individuals angry (for example), with cultural differences in the basic emotion itself. Culture, the way things are done around here, can clearly lead to differences in the emotion we feel when we see actions that conflict with local values. I witnessed this during Kurban Bayram (Eid-al-Adha), a holiday period in Turkey when sheep are sacrificed (and offered to poorer communities), often in public, leading to different emotional reactions from locals and European observers.
Endnotes Myth #3:
[i] Basic Emotion Theory proposes that human beings have a limited number of emotions (e.g., fear, anger, joy, sadness) that are biologically and psychologically “basic” (Wilson-Mendenhall et al., 2013), each manifested in an organised recurring pattern of associated behavioural components (Ekman, 1992a; Russell, 2006).
[ii] Constructivist Theory of Emotion is “any theory holding that emotions are not innate but constructed through social and cultural experience” (APA Dictionary of Psychology (2022). Available at: https://dictionary.apa.org/constructivist-theory-of-emotion [Accessed: 9 August 2022]).
[iii] Darwin’s Claim of Universals in Facial Expression Not Challenged (2014). Available at: https://www.paulekman.com/blog/darwins-claim-universals-facial-expression-challenged/ (Accessed: 9 August 2022).
[iv] Ekman, P. (2016) “What Scientists Who Study Emotion Agree About”, Perspectives on Psychological Science, 11(1), pp. 31-34. doi: 10.1177/1745691615596992.
[iv] Shweder, R. A., et. al. (2008). The cultural psychology of the emotions: Ancient and renewed. In M. Lewis, J. M. Haviland-Jones, & L. F. Barrett (Eds.), Handbook of emotions (pp. 409–427). The Guilford Press.
Myth 4: Micro-facial expressions of emotion don’t exist because I have never seen them.
Sadly, I have heard variations of this too often. So, I would like to offer a few points of clarification for those interested in this topic.
How fast is a ‘micro’ facial expression?
Some commentators suggest micro-facial expressions (MFEs) have a duration of less than 1/25th of a second, and others say less than 1/5th of a second. We thought that too, ten years ago, but our own research, and other current research, suggests that MFEs should be defined as expressions in the face that last for up to 1/2 a second.
One factor is that emotional episodes usually last for between 1/2 a second and 4 seconds – ‘macro’ expressions. So anything quicker suggests an emotion is being suppressed/repressed/masked and therefore should be labelled ‘micro’ since there have been no suggestions of expressions (such as ‘medium’) falling between macro and micro.
Our own primary research in 2016ii used a 200fps camera and good methodology for generating real MFEs (not samples created by a one-frame image of a posed macro expression planted in 23 or 24 frames of a neutral expression). Our results showed that the majority of the expressions that subjects are incentivised to hide have a duration of between <1/4s and >1/2s. Here is a subsequent paper (iii) from a respected fellow scientist who also published the first peer-reviewed paper on MFEs.
Do untrained people miss Micro Facial Expressions?
Many do miss them. Maybe because they aren’t attentive, perhaps they are focusing on a different channel, or maybe they haven’t learned to identify the MFEs as such (misjudging movements as twitches?).
Can trained people see Micro Facial Expressions?
Yes. Paul Ekman Group and David Matsumoto’s ‘Humintell’ offer competing MFE training tools that improve MFE detection. On our courses, we regularly see improvements from around 50% (group average) to around 90% accuracy.
Do they happen with real (suppressed) emotions? Are there any critic bias/agendas?
Yes. See our SAMM dataset(footnote ii) and also our YouTube channel (https://www.youtube.com/c/emotionalintelligenceacademy) and Website (https://www.eiagroup.com) if you need some none-research, real-life examples.
Are there any critic bias/agendas?
Yes. Those denying the existence of MFEs seem to be either:
- Social media uninformed trolls.
- Researchers who have been critiqued by MFE advocates.
- Biased researchers’ selling’ alternative emotion theories or single-channel body-language/statement-analysis/cognitive-load based behavioural analysis constructs or products.
Remember, though, that MFEs alone aren’t reliable for judging emotions or deception. We advocate a multi-channel approach so behaviour analysts can corroborate clusters of cues across six channels (face/body/psychophysiology/voice/verbal style/verbal content). See SCANS model in Section 5 of this article (iv) and in this peer-reviewed article (v).
Endnotes Myth #4:
i Frank MG, Ekman P, Friesen WV. Behavioural markers and recognizability of the smile of enjoyment. J Pers Soc Psychol. 1993 Jan;64(1):83-93. doi: 10.1037//0022-3514.64.1.83. PMID: 8421253.
ii https://e-space.mmu.ac.uk/617069/1/IEEETAC_final.pdf
iii Matsumoto, D. and Hwang, H. (2018) “Microexpressions Differentiate Truths From Lies About Future Malicious Intent”, Frontiers in Psychology, 9. doi: 10.3389/fpsyg.2018.02545.
iv https://www.eiagroup.com/study/facial-expressions/how-are-facial-expressions-measured-and-analyzed/
