Reading Minds

Is commercial lie detection set to go?

Evolutionary psychologists suggest that human cooperation is the result of evolved brain mechanisms that enabled our ancestors to detect cheaters. Broadly speaking, cheaters are people who accept a benefit from someone on the understanding that they will reciprocate, but then fail to give back. A robust finding of game theorists is that the ability to detect cheaters is necessary for cooperation evolved. In the constrained situations that characterize games it's easy for players to detect cheaters because their lack of reciprocation becomes immediately obvious. But what about the real world? Are evolutionary psychologists right when they claim that human beings have evolved into natural lie detectors?

Most of us think that we're pretty good at identifying liars. However, a lot of experimental data says that we're wrong. Most people can distinguish truth from lies at a rate no better than chance. Not even professionals, such as cops and judges, do much better. Of course, humanity has been ceaselessly seeking the fool-proof lie detector, ranging from thumbscrews to polygraph testing. With regard to the latter, the National Academies of Science issued a comprehensive report in 2003 on polygraphy that concluded, "There is essentially no evidence on the incremental validity of polygraph testing, that is, its ability to add predictive value to that which can be achieved by other methods."

In the meantime, some social psychologists claim that they can train people to more accurately detect deception up to 78 percent of the time. Trainees focus on cues like fleeting facial expressions that betray what others are really thinking. However well this kind of training may work as a way to screen people passing through airports and other transit points for potential criminal activity, it will not do for presenting evidence in a court of law.

Deception arises in our brains. The utility of finding a way to look under the hood directly for the source of deception is undeniable. Not surprisingly, a number of researchers have been trying to find correlates in the brain for truth and lies. Using functional magnetic resonance imaging (fMRI), Temple University radiologist Feroze Mohamed says that he has identified 14 regions in the brain that become active when someone is telling a lie. By contrast, only 7 regions are active when subjects are telling the truth. Interestingly, no neural imaging studies have found brain regions that become more active during truth telling (compared with lying). According to British researchers this finding, "supports the hypothesis that truthfulness comprises a relative baseline in human cognition and communication." Evidently the researchers have not spent much time in Washington, DC. On the other hand, perhaps Mark Twain was right when he observed, "If you tell the truth you don't have to remember anything."

A machine that could reliably identify the neural correlates of truth and deception would be the ultimate lie detector. Now a couple of American companies are claiming to be able to do just that. No Lie MRI in Tarzana, Calif., and Cephos Corporation in Pepperell, Mass. use fMRI scanning to uncover deception. No Lie MRI asserts that its technology, "represents the first and only direct measure of truth verification and lie detection in human history." Both companies say that their technology can distinguish lies from truth with an accuracy rate of 90 percent. The New Scientist cites the case of a No Lie MRI client, Harvey Nathan, whose deli burnt down in 2003. His insurance company refuses to pay him because of suspicions that Nathan may have set the fire himself. In order to prove his innocence and thus collect his insurance money, Harvey had No Lie MRI scan his brain. The result? The scan says Nathan is innocent. No word yet on how impressed his insurance company is.

What evidence does No Lie MRI and Cephos Corporation offer for their assertion of 90 percent accuracy in detecting lies? A look at the studies cited on No Lie MRI's website is not reassuring. The company links to one done using 26 right-handed male undergraduates; to another with 22 right-handed male undergraduates; and to a third one with 23 right-handed participants (11 men and 12 women). Oddly, the company links to a 2005 review article from a bioethics journal that is against commercializing fMRI lie detection now. The article actually warns, "premature application of these technologies outside of research settings should be resisted."

Cephos links to just three fMRI studies, one using a total of 61 subjects (29 male and 32 female of whom 52 were right-handed); another using 14 right-handed adults who did not smoke or drink coffee; and a third one that tested 8 men. So adding up the studies cited by these two companies, we get a total of 154 subjects whose brains have been probed for lying in controlled laboratory settings. And on top of that most of the subjects studied were right-handed men. Of course, these are not the only fMRI deception studies, but it would be surprising if the number of subjects whose brains have been scanned to date for deception exceeds a couple of thousand.

While fMRI deception detection looks very promising, a lot more and different kinds of brains need to be scanned in order to validate it. Right now its accuracy has not yet been proven beyond a reasonable doubt. Or as Stanford law professor Hank Greeley succinctly put it: "I want proof before this gets used, and proof is not three studies of 40 college students lying about whether or not they are holding the three of spades."

Ronald Bailey is Reason's science correspondent. His book Liberation Biology: The Scientific and Moral Case for the Biotech Revolution is now available from Prometheus Books.