Telling lies is stressful. That’s the basic logic of a polygraph test: that the stress of deceiving others will manifest itself through fleeting physical responses that may be imperceptible to another person but can be measured by a machine. Typically, a polygraph records blood pressure, galvanic skin response (a proxy for sweat), breathing and pulse rate.
There is a fairly standard protocol for the lie detector examination. The examiner will mix specific questions relevant to a case – “Did you commit a robbery on 29 March?” – with a series of control questions. Crucially, the control questions are also designed to be anxiety-inducing – for instance: “Have you ever stolen from a friend?” Along the way, the subject will be reminded that the machine can distinguish truth from lies and that they must respond truthfully.
In theory the control questions, designed to be difficult to answer with absolute honesty, will generate some baseline level of stress. For an innocent subject, the assumption is that these questions will be more stressful than the relevant ones, where a straightforward denial can be given. For a guilty party, the relevant questions are expected to be more stressful to answer.
So a liar is expected to have higher physiological responses to relevant questions than to control questions, and someone telling the truth will show the reverse pattern. A similar response to each set is judged inconclusive.
Many experts question whether this works with any reliability in practice. “Polygraphs work very well as physiological measures – scientific measures of changes in your body as you experience different emotions,” said Prof Sophie Scott, a neuroscientist at University College London. “However, they are not scientifically validated as reliable measures that someone is lying.
“Indeed, there are no reliable measures of lying, full stop, which is why the police use cognitive measures such as asking people to describe events backwards, to spot inconsistencies as liars start to get things wrong.”
Prof Chris Chambers, a psychologist at Cardiff University, put it more bluntly: “Polygraphs are bullshit. They have always been bullshit and they will always be bullshit.”
An exhaustive review of the scientific evidence by the US National Research Council in 2003 indicated that although the polygraph performs above chance, studies had found wildly differing accuracy rates. Some gave accuracies of 85% when evaluating genuinely guilty people, which proponents of the polygraph say underlines its utility.
However, the review also highlighted the potential for high false positive rates ( some studies found almost half of innocent people were identified as liars), and it pointed out that people can train themselves to beat a polygraph. It concluded that the US government should not rely on polygraph examinations for screening prospective employees or to identify spies or other national security risks, because the test results were simply too inaccurate. Another analysis published last year reached broadly the same conclusions.
Prof Albert Vrij, a psychologist at the University of Portsmouth whose research focuses on deception and lie detection, said overconfidence of examiners in the accuracy of lie detectors was a common theme. “Although they often acknowledge that the test is not always accurate, they often seem to think that other examiners make incorrect judgments rather than they themselves. You never hear stories from polygraph examiners where they got it wrong.”
Despite their notable shortcomings, lie detectors have had something of a renaissance in the UK in the past decade. Last year the Ministry of Justice launched a three-year pilot of mandatory polygraph tests on convicted domestic abuse offenders released on licence. Tests have also been given to serious sex offenders on parole in England and Wales since 2007, and since 2014 mandatory tests have been added to some offenders’ release conditions.
Vrij said the reliability of lie detectors in these latest applications was even more unclear. Polygraphs were designed to test someone’s involvement in a known crime that occurred in the past. In some cases, sex offender testing could involve open questions about future intentions.
“To use sex offender testing as a justification to use it in terrorist cases is odd,” said Vrij. “Of course, it gets further away from its original design as the ‘crime in question’ is no longer a crime committed in the past. To simply introduce the test in an entirely different setting is too far stretched.”
So the science is shaky. But in the real world, the question of whether lie-detector tests work has another component: irrespective of their accuracy, do they make people tell the truth? There is evidence to support this, including a 2007 study by scientists at the University of Kent which showed that sex offenders who were attached to a fake polygraph admitted to far more thoughts that would place them at risk of future offending than in a standard interview.
Perhaps these observations, rather than a disregard for scientific evidence, are the basis of the latest polygraph rollout.
Prof Thomas Ormerod, a polygraph expert at the University of Sussex, said: “The use of lie detector tests in this context will, at best, be a waste of police resources, and at worst will exacerbate problems associated with terrorism. Their use will give the public false confidence that they are being protected, while for terrorists who take and pass the tests, it gives them a free pass out of the legal system and cover to carry out attacks unimpeded. Moreover, for genuinely reformed offenders or innocent people implicated in terrorist activity, use of a technique that cannot be used in UK courts as admissible evidence will likely appear discriminatory.”
“It seems a missed opportunity that the UK government is failing to use the research results it has contributed to funding, preferring instead to rely on techniques that seem to offer a magic technology but in fact are deeply flawed and potentially dangerous.”