By Dr Llewellyn Cox, Principal, LieuLabs
The young gentleman on the Maury Show is looking uncomfortable. He shuffles uncomfortably in his seat as the genial Maury Povich reads aloud the results of his recent polygraph test to the audience in the studio, who dutifully greet each new result with a hearty chorus of boos.
More booing. Tiffany, a heavy-set young woman sitting in the next seat along, jumps to her feet and screams a continual stream of expletives at the young man that lasts for at least a minute. The censorious *bleep* edit will be working overtime on this episode. An immense bodyguard effortlessly holds her back as she flails madly around him with her fists, trying to make contact with the young man’s face. Eventually, she gives up and storms off the stage, a mobile camera crew hot in pursuit as she flees backstage to the Green Room.
There is one problem with the polygraph test — it doesn’t actually work. Despite decades of use in the field, the machine’s effectiveness cannot be scientifically substantiated, and it is not widely considered allowable in courts as evidence. The National Academies of Sciences found in 2003 that the majority of polygraph research that had been done to date was “unreliable, unscientific and biased”. It is well known that there are established techniques to cheat the test — as the polygraph essentially measures physiological stress as a corollary for lying, a skilled liar can easily pass the test by maintaining calm and focus.
If polygraphs are so unreliable, then why are they still used? Although they are not a reliable scientific method to establish for sure whether or not a subject is being truthful, polygraphs are still very useful as an investigative tool. Among the best-established interrogative methods for outing a liar is to unsettle the subject, to mislead them into thinking that the interrogator knows a truth that will ultimately uncover their lie anyway. The theater of the polygraph can have a powerful effect on the weak-minded — like a voodoo curse, it only works if the recipient thinks it does. A focused, relaxed subject may be able to beat the test, but its very use is often enough to induce the unsuspecting into confessing the truth.
Any test that requires human interpretation can be biased. No matter how high-tech the test’s technical underpinnings, as long as a human being is responsible for determining the outcome, then it will always have a flawed quality of individual judgement built-in. If a more advanced lie detector test in the future is ever to be considered reliable, it will need to operate and analyze its data autonomously. To be considered accurate and reliable, it would also need to be extensively validated by experts other than the police or prosecutors it will serve.
Ultimately, even with the best of tools and a free hand to use them, we will only ever be able to find out if a subject thinks they are telling a true and accurate account of what occurred. Human memory is far from accurate, and so by itself is still an unreliable witness to the truth, regardless of how expert we grow able to harvest it from the minds of others.