I recently received the following e-mail:
I have a question about science into deception and detecting lies.
Particularly to do with the show, ‘Lie to Me’ and the Cal Lightman character. The show is supposedly based on some science about deception. Wikipedia says the lead character is loosely based on Dr. Paul Ekman.
The notion that because we’re hardwired to communicate our feelings through facial and bodily expressions making deception difficult (which would leave specific and unconscious tells) seems to me sensible and I would’ve liked to have believed it, bought the book by Ekman and tried to learn to read micro-expressions.
However I keep hearing that under controlled conditions that no one, not experienced detectives, scientists, judges nor average people of the street demonstrate can discern lies from truths any better effectively than pure chance.
Is this stuff about microexpressions and truth wizards based in any reality at all, and if so, how limited or useful is it really?
This is an excellent question. Many of the pseudosciences we cover are so obvious and extreme that anyone with basic critical thinking skills and a modicum of scientific literacy should be able to smell them a mile away. So I enjoy also covering topics that are on the edge of legitimate science and pseudoscience, where even an experienced skeptic would have to do some digging or ask an expert to figure our how legitimate the claim is.
Claims for lie detection are generally highly problematic, although not completely pseudoscientific. There is a body of rigorous science, but the difficulties in applying that knowledge in a practical way to the question of lie detection tend to be overwhelming.
The first problem that is often pointed out with lie detection is that the technology is not detecting lies – but the markers of behavior that correlate with lying. The polygraph, for example, measures physiological stress, which is assumed to correlate with psychological stress, which is assumed to correlate with lying. But there are nervous interviewees who display stress without lying, and their are cool liars who do not reveal stress even when they are lying.
Group vs Individual
Another primary limiting factor is that individual variability tends to be greater than commonality in terms of behavior. When doing research you can look at correlations one of two ways. You can take a group of subjects who are known to be lying and another group known to be telling the truth and compare them. When you do this you can find statistical differences between the groups.
We can therefore make statements about the kinds of behavior people display when they are lying. But this is not the same thing as indicating if one individual is lying (called single-subject truth-verification). This has been a primary (but not only) problem with the polygraph test, for example.
Therefore with this type of technology we can only make statistical statements about the probability that someone is lying or telling the truth. This renders the technique not entirely useless, but questionable, especially in some applications, like the courtroom.
If, however, we can get close to 100% accuracy in comparing groups, then it becomes more reliable to apply the technique to the individual. This is the promise of a new technology using fMRI scans to see what is happening inside the brain when people lie vs tell the truth. The idea is that we may be able to control our voice and facial expressions, but not what is happening inside our brains.
fMRI Lie Detectors
Studies with fMRI have been impressive. In a concealed information test model some studies have found a 100% correlation between lying and prefrontoparietal lie activation. Overall correlation is above 90%. It seems our brains go through a different process when fabricating a lie than when recalling the truth, which makes sense.
Since an initial publication in 2001,37 several papers on the BOLD fMRI methodology have reported differential patterns of blood flow in various brain regions in experimental paradigms in which subjects were instructed to lie or deceive in one task condition and respond truthfully in another task condition. The task paradigms included forced-choice lies (i.e., responding yes when the truth is no and vice versa)37,38; spontaneous lies (i.e., saying Chicago when the true answer is Seattle)39; rehearsed, memorized lies39; feigning memory impairment40,41; and several variations of the Guilty Knowledge Test,42,43 including lying about having a playing card,44–47 lying about having fired a pistol (loaded with blanks) before the scanning session,48 lying about the location of hidden money,49,50 and lying about having taken a watch or ring.51
The most consistent finding is activation of the certain prefrontal areas with lying. The hypothesis is that this area is needed to suppress saying the truth, which would otherwise be the default response to a question. The research summarized above, however, is of the group kind. There have only been a couple of studies looking at the accuracy of lie detection in individuals. These studies show about a 90% accuracy of this technique.
Still, fMRI lie detectors have a similar problem to that of polygraphs. In the polygraph someone who is fearful or anxious may display the same autonomic findings as someone who is lying, generating a false positive. Likewise, the fMRI is detecting (probably) not a lie but the suppression of a response. There may be many reasons in a real life situation when an innocent person would want to suppress a response, generating a false positive.
There are other problems with applying this technology as well.
Lie detector researchers have known for a long time about countermeasures – techniques that individuals can use to foil a lie detector.
Ganis et al recently explored whether lie detection countermeasures would affect the fMRI model of lie detection. Their research confirmed that 100% of subjects who are lying display characteristic changes on fMRI scanning. However, when typical countermeasures were employed the detection rate was reduced to 33%. If their findings are correct, this would put fMRI lie detection in the same camp as the polygraph.
Finally we get to the issue of microexpressions. This is the basis of the TV series Lie to Me. There is little science behind microexpressions, however. I found one peer-reviewed study specifically on microexpressions as lie detection, which found some correlation but is not sufficient to support this technique as a form of lie detector.
The basic idea, originated from Paul Eckman, is that when we conceal our true emotion, the real emotion leaks through with brief “micro”expressions. If someone is pretending to be happy when they are really sad, they will frown briefly.
If true, again we have a marker for lying, and not really a test for lying. Microexpressions would reveal that someone is concealing their true emotions, but not why they are doing so.
The microexpression concept has not been studied enough to conclude that it is legitimate. If it holds up under adequate replication then we still have all the problems inherent in lie detection – it is looking at a proxy of lying, there are likely to be effective countermeasures, and variation in individual personality and situation may overwhelm and true signal of deception.
The various lie detection technologies have collectively discovered interesting facts about human emotion and deception. They have not, in my opinion, yielded an effective lie detector, capable of high levels of accuracy despite deliberate countermeasures. The more recent technologies – fMRI and microexpressions – also need to be researched in real world applications to see how generalizable their principles are.
I would not rule out that an effective lie detection technology can be developed. The most promising approach seems to be with brain scanning. But we are not there yet.