ONE WILD WORLD — NEWSLETTER

ISSUE 056 · DEEP DIVE · MYTHS & MISINFORMATION

William Moulton Marston built a blood-pressure test for lying in the 1910s, with his wife Elizabeth Holloway, whose contribution went uncredited for most of a century. He is the reason the story is fun: Marston also created Wonder Woman, and gave her a golden lasso that compels people to tell the truth. Same idea, better costume.

He also claimed to have invented the lie detector, which he had not. The polygraph proper — pulse, respiration, blood pressure and skin conductivity together — is credited to John Augustus Larson, a Berkeley police officer. In 1922 Marston was indicted for federal business fraud.

The man most associated with the world's most famous truth machine lied about having built it. You could stop the article there and it would still be worth your time, but the interesting part is what happened next.

Here is where most writing on this subject goes wrong in one direction or the other. The polygraph is not a coin toss. It is also not a lie detector.

The National Academy of Sciences reviewed the whole field in 2003 and concluded that specific-incident tests can discriminate lying from truth-telling at rates well above chance, though well below perfection. The polygraph profession's own meta-analysis of 38 studies puts decision accuracy at around 85 per cent. Fine. Something real is being measured.

The same report also concluded that almost a century of research in psychology and physiology gives little basis for expecting a polygraph test to have extremely high accuracy, and that the true error rate is unknown. A 2019 review by Iacono and Ben-Shakhar found those conclusions still standing.

The reason is simple and unfixable. There is no known physiological signature of lying. The machine measures arousal — heart rate, blood pressure, how much your skin conducts electricity. Nervousness, anger, indignation and guilt all look much the same on the chart. Which is why a frightened innocent fails and a calm liar passes.

For the job the United States mostly uses it for — screening large numbers of people to find a handful of bad ones — accuracy is almost beside the point, because the mathematics defeats it either way.

The NAS worked the example itself. Picture 10,000 government employees, ten of whom are spies. Assume an accuracy of 0.90, which is more generous than the evidence supports. Set the test sensitive enough to catch around 80 per cent of the deceivers, and roughly 1,606 employees fail. You catch eight spies and accuse about sixteen hundred innocent people.

Turn the dial the other way, so that only about 40 of the 9,990 honest employees are falsely flagged, and you now miss eight of the ten spies.

There is no setting that works. This is the base-rate problem, the same statistical trap that makes mass screening for rare diseases produce mountains of false alarms. The NAS concluded that the polygraph's accuracy in distinguishing actual security violators from innocent test takers is insufficient to justify relying on it for employee security screening in federal agencies.

And countermeasures work. The report noted that test accuracy may be degraded by them, particularly when used by major security threats — precisely the people with the strongest incentive and the most resources to learn them.

Because it was never really being used as a measuring instrument, and the agencies are refreshingly open about this in their own paperwork.

Senior officials value the polygraph for its utility in generating admissions — during the pre-test interview, the test itself, and the interrogation afterwards. It also works as a deterrent, discouraging some applicants from even applying for sensitive posts. Neither of those benefits requires the machine to be accurate.

The NAS put its finger on the mechanism in a single clause: the effect of the interrogation depends on the examinee's perceptions of polygraph accuracy. Your belief in the machine is a working component of the machine.

The comfortable version of this is that America is credulous about lie detectors and Europe is too enlightened for them. Both halves are wrong.

American courts do not trust it either. In United States v. Scheffer (1998) the Supreme Court upheld a flat ban on polygraph evidence in courts-martial, on the grounds that there is no consensus that it is reliable. Around half of US states exclude it entirely.

And Britain has become one of the most enthusiastic users on earth. The Offender Management Act 2007 made polygraph testing mandatory for high-risk sex offenders on licence, rolled out nationally from January 2014. Since then: over 8,800 probation tests and around 4,600 police tests, with roughly 65 to 70 per cent producing significant risk-related disclosures. In 2021 it was extended to convicted terrorists under the Counter-Terrorism and Sentencing Act and to domestic abuse perpetrators under the Domestic Abuse Act; in 2022, to armed forces offenders. Poland, meanwhile, has admitted polygraph results as supplementary evidence since a Supreme Court ruling in 1976.

Section 30 of the British Act contains the whole philosophy in one provision: the results cannot be used as evidence in criminal proceedings against the offender, but the information can be passed to police for further investigation. Inadmissible as proof. Perfectly good as a lead.

Continental Europe does say no, but not for the reason you would guess.

Germany's rejection is not primarily about accuracy. The Federal Court of Justice ruled against the polygraph as early as 1954, grounding it in Menschenwürde — the human dignity clause that opens the German constitution, placed there deliberately as an answer to the Nazi era. In December 1998 the court classified the polygraph as an entirely unsuitable means of evidence in criminal proceedings. France's rejection has no landmark ruling at all; it is cultural, rooted in the tradition of intime conviction and the same concern for dignity.

So America asks whether the machine works. Germany asks whether we are allowed to use it at all. Different questions, same answer, and only one of those two positions can be changed by better technology.

National Research Council (2003). The Polygraph and Lie Detection. Washington, DC: The National Academies Press — Executive Summary, Table S-1, and Chapter 7, “Uses of Polygraph Tests.”

United States v. Scheffer, 523 U.S. 303 (1998).

Iacono, W. G. & Ben-Shakhar, G. (2019). “Current status of forensic lie detection with the comparison question technique: An update of the 2003 National Academy of Sciences report.” Law and Human Behavior.

Offender Management Act 2007, ss. 28–30; Counter-Terrorism and Sentencing Act 2021, s. 32; Domestic Abuse Act 2021, s. 76; HMPPS Polygraph Examinations Policy Framework.

Grubin, D., Newcastle University — UK polygraph research programme and the 2012 evaluation of the mandatory polygraph pilot, Ministry of Justice.

Floren, D. (2025). “The Legal Status of the Polygraph in Germany.” European Polygraph 2025(1) — covering BGH 1954 (1 StR 578/53), BVerfG 1981, and BGH December 1998 (1 StR 156/98).

Britannica, biography of William Moulton Marston; PBS American Experience, “The Man Behind Wonder Woman” and “The Lie Detector.”

American Polygraph Association (2011), meta-analytic survey of validated polygraph techniques.

ONE MORE THING

You can read our MANY newsletters for free at our website:

Read our archive for free at OneWildWorld.com →