IE 11 is not supported. For an optimal experience visit our site on another browser.

Getting at the truth behind lie detectors

Do lie detectors tell the truth, the whole truth and nothing but the truth? Not exactly, says “Gee-Whiz Science” columnist David Ropeik.
/ Source: Special to

Wouldn’t it be great if there was a machine that could reliably detect liars?

There is. But even lie detectors don’t always yield the truth, the whole truth, and nothing but the truth.

The lie detector machinery of today is based on the same premise the ancient Chinese used to catch liars; the stress of being caught lying causes changes in bodily functions, changes that can be detected.

The Chinese used rice. An examination for truthfulness might go something like this:

“Is your name Chiang?” (They know the guy’s name is, in fact, Chiang.)


The interrogators hand Mr. Chiang some rice. They have already counted the number of rice grains.

“OK. Put this handful of rice in your mouth. Hold it for three seconds. Spit it out.”

Then they count how many rice grains come out.

“Did you steal the chicken?”


“OK. Put this handful of rice in your mouth. Hold it for three seconds. Spit it out.”

Again, they knew how many grains went in, and they count how many come out. If more grains come out after the question about the stolen chicken than came out after the “easy” question, where the suspect truthfully gave his name, they know he’s lying. How? The stress of being caught lying makes the suspect’s mouth drier. Fewer grains stick. More come out. Mr. Chiang stole the chicken.

A device instead of rice
Fast forward a few millennia, to today. The technology has advanced, but the premise holds. The stress of lying is going to cause measurable bodily changes. A modern polygraph (“poly,” for multiple; “graph,” for the writings made on the measurement chart during the test) goes something like this:

First, the person being questioned has to agree to the test. Interrogators don’t want a person stressed by having to take the test in the first place. That stress could skew the results.

Next, the test subject is told the questions. In advance. Interrogators don’t want the suspect surprised by questions. The surprise, rather than the questions, could cause the stress measured by the apparatus. When the suspect knows a question about their truthfulness is coming, that induces the stress measured by the test.

Then the suspect is connected to three measuring devices. One tracks blood pressure. One tracks the breathing rate and depth of each breath. One tracks electrodermal response … the flow of tiny amounts of electricity on the skin, which goes up when there is more sweat on the skin … like, when somebody’s lying.

These bodily functions are all controlled by the autonomic nervous system. So they’re tougher for the test taker to manipulate.

“Is your name John Doe?”


“Are we in Washington, D.C.?”


Meanwhile, the machine is running a paper tape, and needles are making little wiggly lines on the chart measuring the responses of the blood pressure, breathing and electrical levels on the skin. The test subject is allowed to see the little wiggly lines … to make him nervous about how those lines will look when “the” question comes up.

“Did you steal the $500?”

“No.” What matters here isn’t what the lines do, as much as what they do compared with the lines made when the control questions were asked. Significant differences indicate “Liar, Liar, pants on fire.”

Sometimes the examiner asks the question directly; “Did you do it?” But sometimes they do a concealed information test. Let’s say there’s some clue about the crime that the police discovered that only the perpetrator could have known, like a item of clothing — say, a blue hat — left at the crime scene.

“Did you know the robber was wearing a blue hat?”


If the person shows signs of stress, it’s “Liar, liar, pants on fire.”

There are as many as 50 physiological responses that can be measured suggesting stress. Eye movements, dilation of the pupil, changes in blood volume in the fingertips as measured by something called a plethysmograph — all suggest stress. But research shows that these measures are no more telling than the basic polygraph measures of blood pressure, respiration and skin electrical activity. (A penile plethysmograph is used to test sex offenders to see if they have been cured of their arousal enough to be released from prison.)

Do they really work?
There remains controversy over how conclusive the results of polygraph tests are. Some evidence suggests that a test subject can fool the detection apparatus. Also, standards for administering polygraph tests vary widely, and the results are only as good as the way the test is administered … the way the questions are worded, or the tone of voice of the examiner, and so on. So polygraph results are not admissible in open formal court testimony.

But polygraphs are commonly used in investigations. Prosecutors offer them to suspects, and if the suspect passes the test, the investigators might redirect their investigation to somebody else. Defense attorneys give private tests to their clients, and if they pass, they offer to have their clients take one for the police.

The police say this is fine. The suspect will still react nervously to the second test, being conducted now by police, even if he’s already heard the question from his friendly defense attorney. Also, police tests might word questions differently, evoking a different response. Finally, if a suspect does register as truthful on the police test, that provides useful information to investigators. It’s estimated that more than two-thirds of police departments in America use polygraphs to help solve crimes.

Sixty-two percent of the nation’s police forces use polygraphs to screen job applicants. So does the CIA, the National Security Agency, and now the FBI. Private employers are not allowed to subject job candidates to polygraph examinations except in a few high security industries like drugs and money manufacturing.

Other lie detectors
There are two other ways of detecting liars. One is technological — the voice stress analyzer, or VSA — which is based on the premise that there are normal subaudible tremors in the voice of someone talking while not under stress. These tremors are gone in the speaking voice of a person who is under stress. The voice stress analyzer can detect the changes. Research indicates this technology is not very precise at picking up deceitfulness.

The other way isn’t technological at all, but very human. A group of people known as aphasics have proven to be accurate natural lie detectors. Aphasics have lost the ability to understand or use words due to brain damage. They can still hear. They just can’t understand. These people have proven highly accurate at detecting liars through changes in facial expression and in the fundamental audible frequency of the voice. (Remember, the microtremors being detected by VSEs are subaudible.) Apparently something changes to the frequency of our audible voice under stress. Aphasics, without interference from the meaning of the words to get in the way, apparently can detect that subtle change in frequency.

This report was originally published on July 24, 2001, and has since been updated.