The age of the lie detector

Lying is a stressful business. Can police trust the polygraph to root out the liars?

19th October 2017

Most of us have a certain vision of how a lie-detector test works. It’s a common theme for daytime tabloid shows where it’s hailed as the ultimate authority on whether a person is being truthful or not.

Picture it now: the guilty-shift of the defendant’s eyes, the anxious, emotionally charged apprehension of the other guests, the cutaway shot to the audience waiting with baited breath, and the host—note cards in hand—waiting until the last excruciating moment for the big reveal. It manifests in the moment the host trills out: “the lie-detector test determined that was … a lie,” and subsequently, everyone loses their shit. The set becomes a deluge of audiences yelling, fists flying, people displaying anger, despair, devastation, disbelief, righteousness and, finally, validation.

But that’s a theatrical, overly embellished version of what the polygraph actually is, what it does, and the effects that it can have for people stuck between being branded lawfully honest or criminally dishonest once and for all.

How do polygraph tests work?

Polygraph testing usually occurs in phases. The first phase involves a pre-test interview in which clients can disclose things that might incriminate them or they can confess their culpability, and then the polygraph examiner formulates the questions they want to ask. Following that, the examiner asks the client one (or more) control question to determine their base level of response—physiognomic signifiers such as skin conductivity (sweating), respiration, and blood pressure—which will later be compared to their response levels in regards to the relevant questions, so as to determine whether they are being truthful. The examiner then asks the three specifically detailed questions, which the client answers. And then, the examiner mixes up the questions, asks the client again and they answer. Rinse and repeat, once more.


During its long history the polygraph has been a contentious character in the quest for truth and efficiency; its validity has been called into question, as has its ethical and moral impacts. Former US police officer Douglas Williams brands it “the last message of witchcraft”, a pseudo-science that puts dangerous precedent on mistaking nervous, honest people for liars. Fresh from his two-year stint in prison, with charges related to teaching clients (some of whom turned out to be undercover FBI agents who confessed “crimes” to him) how to cheat polygraph tests, he tells me that he’s in the midst of putting together a lawsuit that seeks compensation for the damages inflicted upon innocent persons whose credibility has been shattered by their polygraph results.


The Ink polygraph
Image: Wellcome Library, London

Part of Williams’ vehemence towards polygraphs is that he says they’re not much more accurate than the toss of a coin and even then polygraphy lacks scientific basis. But this is not necessarily true.


The American Polygraph Association (APA) is one of the principal authorities within this small, insular industry. It has its own board of directors, a set of guidelines and an eponymous journal hosting a series of scientifically backed research regarding polygraphy. Within the APA’s website there is a report by Raymond Nelson titled ‘Scientific Basis for Polygraph Testing’ which reviews a bounteous amount of research—albeit by a small number of researchers—and concludes that polygraph testing is, on average, accurate 95% of the time. This is a conclusion that is reiterated by the British Polygraph Association (BPA), an affiliate of the APA, and the private detective agency Lie Detectors UK, whose examiners graduated from and conduct their work according to the APA.

Sex offenders

This data can be a little misleading, critics say. While it is true that the experiments reviewed in Nelson’s paper offer a high probability of accuracy, the stakes are not the same when comparing volunteer test subjects in a controlled scientific environment to people who face serious criminal convictions depending on the outcome of their tests.

In the UK, polygraph tests have been rolled out nationwide as part of the post-conviction management of sex offenders on the sex offender register. It is mostly on a voluntary basis, but it is mandatory for high-risk sex offenders. As per a spokesperson for the National Police Chief’s Council (NPCC), there is no national policy for police use of the polygraph in such cases, but independently operational local forces opt to use them in order to aid their processes. The Ministry of Justice, in collaboration with researchers based at the University of Kent, published an evaluation of a mandatory polygraph pilot between April 2010 and December 2011.

Testing two groups (one who undertook polygraphs, and one who didn’t) the publicly available review finds that, out of 303 offenders, the probation officers of the polygraph-tested group engaged in almost double the amount of actions compared to the non-polygraph-tested control group. These actions include more increased “high-risk” assessments, more warnings issued to offenders, more recommended recalls of offenders and more decreases in risk assessment.

Notably, however, the review also found there is a 64% increase in offenders admitting to problematic behaviours when they are subjected to a polygraph test.

Intimidation tactic

Seemingly the polygraph can be used not only as scientific instrument but also as an intimidation tactic. This is certainly the way that Williams sees it. He calls it nothing more than a “psychological prop used to coerce confessions out of people”.

the polygraph’s accuracy isn’t as important as the suspect’s belief that it works

George Maschke, who runs, agrees with this. During our discussion he gives me the name of six people wrongfully coerced into confessing a crime they did not commit because of failed polygraph results: Peter Reilly, Kevin Fox, Jeffrey Mark Deskovic, Anteletto Jones, Byron Halsey and Matthew Livers. In some of these cases, the failed result had been falsified by investigators in order to provoke a response.

He concedes that the police can use the polygraph effectively, but he also adds: “In this context the polygraph’s accuracy isn’t as important as the suspect’s belief that it works.”

lie detector
Can a machine help police understand who is lying and who isn’t?

As a scientific instrument that measures physiological response, the polygraph does work, but that’s not really what is being contested here. Rather, the contention lies on the notion that these signifiers always correspond to the person telling a lie. The polygraph doesn’t detect dishonesty, it detects bodily response and it is the examiner who then interprets whether those responses mean that the person is being dishonest. There has yet to be scientific research that conclusively identifies a unique physiological reaction that occurs solely when a person is lying, so detecting whether or not someone is lying falls upon how the examiner interprets the results of the polygraph test, opening up a greater capacity for human error.

Professor Don Grubin of Newcastle University, an instrumental figure in introducing polygraph testing in the probation management of sex offenders in the UK, says this is one of the biggest problems with the system.

“In my experience, objections to polygraphy arise either because of exposure to poor practice, or because of misconceptions about how it works, how it’s used, and how it’s interpreted,” he says.

Polygraphs don’t detect lies

Neither in the UK nor in the US is the polygraph used in isolation. Since evidence obtained by polygraph is not usually admissible in court, it instead becomes a tool that instructs officers of the law in how to move forward and how to prioritise or approach a case.

But there are certain things polygraph subjects can do to produce similar physiological signs that examiners are looking for in a “truthful” candidate. These are termed countermeasures; Maschke provides special detail to these in his free online book The Lie Behind the Lie Detector. And even though Jason Hubble—founder and lead polygraph examiner of Lie Detectors UK—assures me that it is “straightforward to identify [a countermeasure] and [it] usually follows a pattern,” there’s evidence to suggest that that’s not always the case. After all, there have been multiple instances throughout history of people seemingly “cheating” a polygraph test. Famously, the CIA agent and KGB spy Aldrich Ames passed two polygraph tests during his time as a mole.

Polygraph test
Image: Lwp Kommunikáció

Williams actually was a polygraph examiner in the 1970s, and he recalls being taught interrogation techniques meant for intimidation more than being taught how to correctly interpret a polygraph result.

What we can gather from this, then, is that a polygraph result is only as good as its examiner. Its common name, the “lie detector”, is a misnomer. Polygraph testing doesn’t detect lies, but rather detects a probable case of lying.

The difference is significant because it means that we rely on examiners to accurately assess whether or not a person is lying, which can be obviously problematic depending on their expertise, their experience, and their intentions. These are all factors that can drastically change a person’s life; it can besmirch the innocent and liberate the guilty.

But is polygraph testing significantly less reliable than an eye-witness account, a panel of randomly allocated jurors, or a judge tasked with reviewing all possible evidence and making a verdict within, usually, a few hours or days?

In the UK, it has proven to be an advantageous asset in the managing of criminal offenders. All police officers utilising the polygraph as an investigative tool must adhere to the national standards of interviewing as set by the College of Policing. In the US, however, the case in favour of polygraphy is fraught with many more contentious variables, and the path towards the truth isn’t so clear. The BPA’s website slogan proclaims vincit omnia veritas, or, the truth conquers all; but it doesn’t, not always.

19th October 2017