Do Lie Detectors Work? (2024)

The term “lie-detector” suggests that a polygraph machine can detect lies. However, this device, which is commonly used in criminal investigations, actually measures nervous excitement. It operates on the premise that if a person is telling the truth, they will remain calm.

Trial by ordeal

The lie detector can be considered a modern variant of the old technique of trial by ordeal. A suspected witch was thrown into a raging river on the premise that if she floated, she was harnessing demonic powers.

Such techniques never had much credibility. A subject who passed the test had a reasonably high probability of drowning that was at least preferable to getting burned at the stake as a witch.

No one dies of a polygraph test, of course, and the results are mostly inadmissible in courts of law. Yet, a person who is otherwise exonerated by the evidence may endure the legal jeopardy of being unfairly incriminated in the eyes of investigators.

Although polygraph machines look scientific and measure responses such as sweating and increased pulse rate with exquisite accuracy, they are crude in their conception. Indeed, they are no more sophisticated than an ancient Arab ordeal for detecting liars.

In the Arab test, a heated knife blade was pressed to the subject’s tongue. If he was telling the truth, his tongue would not get burned. The idea is that when people are nervously excited, their mouth goes dry because nervousness suppresses salivation. In principle, the lie detection system involved is exactly the same as for a polygraph test.

Does the polygraph work?

There has been a lot of controversy about whether lie detectors work. Some experts claimed that a high proportion of persons who “failed” the polygraph subsequently confessed to crimes. On the other hand, the test generates a lot of false positives, i.e., people who are telling the truth but whose polygraph test suggests they are lying.

Whereas the American Polygraph Association claims accuracy rates of over 90 percent, leading critics, such as David Lykken¹ put the polygraph accuracy rate at around 65 percent that is only slightly better than the 50 percent correct one would get by flipping a coin.

Interestingly, the polygraph is quite good at identifying liars but does no better than chance at detecting honest people according to Lykken. In other words, there is a 50:50 chance that a polygraph test will say an honest person is lying (a 50 percent “false positive” rate).

It is bad enough that polygraph tests are so indirect, prompting some researchers to look for more direct evidence of lying through analysis of brain scans. Another major weakness is that the test can be faked.

In the normal administration of the test, technicians rely upon responses that they know to be true to provide a baseline against which deceptive answers can be judged as an increase in nervous arousal. One of the most effective means of faking the test is to enhance arousal accompanying honest answers so that it is difficult to detect increased arousal theoretically accompanying lies.

The fact that the test can be fooled in this way also highlights the subjectivity with which examiners judge the results, for there is little standardization of procedures as to how much of a polygraph change indicates lying.

Doubts about polygraph tests grew in the scientific community until the National Research Council – an organization of scientists – conducted a systematic evaluation and concluded that the test is lacking in scientific validity².

In 1998, the U.S. Supreme court acted to restrict their use in legal proceedings. In particular, defense attorneys can no longer use evidence that their client passed a polygraph test as establishing innocence of a crime.

Even as the polygraph test is discredited in legal proceedings, its use also declined in other settings and most employers are legally barred from using it as a technique to recruit honest employees. The government is one exception.

From a scientific perspective, there is no rationale for administering a polygraph test. So there is certainly no good reason to take one — if you can avoid it. Otherwise, you expose yourself to the nightmare of false self-incrimination.

References