The American Psychological Association doesn’t like the term used by most people, lie detector, because the machine does not detect lies. The APA prefers the more scientific term, polygraph, a Greek word that meant “writing much.” There is no detecting going on because all the polygraph machine does is record several of the test taker’s reactions, the ones that make the needle jump in movies and on television. There are no lies discovered either, just physical changes that are taken by some to indicate when someone is lying.
Why Can’t Lie Detectors Detect Lies?
“Build a better mousetrap and the world will beat a path to your door.”
To “build a better mousetrap” means to improve an existing product. For more than a century, people have been creating and refining machines to catch people when they lie.

As is so often the case with inventions, the polygraph evolved over time as someone added a bit here and a bit there in efforts to make the machine more efficient
Long before the first machine, people determining the guilt or innocence of someone suspected of a crime thought that liars produced less saliva. Suspected liars had to eat pieces of dried bread and hard cheese in a single bite. The Latins said in wine, there is truth.
Things got tougher when accused liars were tortured by fire and boiling water
Some believed God would spare the righteous and those harmed would be liars. Lombroso’s Glove measured changes in the accused’s blood pressure. Phrenologists believed liars could be identified by the shape of their skulls.

The first machine measured only blood pressure because it was thought to be a sufficient indicator of deception
The inflatable cuff we know today was pumped up and the attendant listened through a stethoscope as the heart muscle contracted and relaxed.

Allison Marsh is a professor of science, technology, and engineering at the University of South Carolina
She says the first actual polygraph (first called a cardio-pneumo-psychogram) was a medical device for recording patients’ vital signs. It was created by a mash-up between two medical devices. One measured pulse and blood pressure. The other was a sleep monitor.

This version added sensors that measured heart rate, breathing, perspiration, and the tiny little changes in our skin’s electric field created by activated sweat glands
Each of these physical reactions is assumed to measure autonomic arousal**. In other words, the machine only indirectly measures what it is supposed to, so lying is implied by physical reactions that are associated with lying, not lying itself.
Smithsonian says “It’s not lie detection, it’s lie inference”
All the machine does is record changes in heart rate, blood pressure, respiration, and perspiration. The determination of whether those results show you’re lying is left up to the examiner administering the test. So the data are science and the interpretation of the data is an art.

Naturally, others would come along and find ways to capitalize on the a.i.* wave
Subjects are interviewed while the camera records how they act and react during the interview. The digitized audios and videos are fed to a computer program that calculates their brain activity by interpreting body language, posture, changes in voice, volume, and tone. This fully tricked-out bells-and-whistles device records and interprets eye movements that appear for only thousandths of a second.

This type of pre-programed computer software isn’t artificial or intelligent
It’s real and it’s only as smart as what some human feeds it.
MIT Technology Review says artificial intelligence-powered lie detectors beat humans at identifying lies
That’s a pretty low bar, because humans are no good at telling when someone is lying. Otherwise, why do we vote for politicians who feed us lies and why do we buy products and services from manipulative advertisers and marketers who care more about sales than facts?
The MIT researchers reported studies where artificial intelligence-powered lie detectors chose the right true or the right false two times out of three
Better, but hardly conclusive, wouldn’t you say?
Here’s the sentence that sums up the MIT article:
In reality, the psychological work that undergirds these new a.i. systems is even flimsier than the research underlying the polygraph.
Then along came MRIs
Real-time brain imaging successfully classified truths and lies three times out of four by recording electrical activity in specific areas of the brain. This beats the polygraph’s just more than half and a.i.’s two out of three.

Whatever ability these machines have to detect lies is negated by the fact that innocent subjects can fail the tests and some guilty subjects are able to pass with flying colors. And it’s impossible to tell the difference.
No lie detectors produce reliable results because of the complexity of human feelings and emotions
Current technology is limited in its ability to accurately interpret subtle and presumably subconscious cues. What lie detectors really measure is anxiety and fear.
So why do people use lie detectors if they don’t work?
It’s the same reason they use the screwy Myers-Briggs Personality Inventory. People use lie-detector tests because they’re popular and they’re popular because people use them.
Who uses these tests?
The Federal Government, the U.S. Military, FBI, CIA, and others use polygraph tests to screen applicants, check backgrounds, determine security clearances, and make hiring decisions.
Until I was researching this subject, I did not know that except in rare cases, civilian business are prevented by law from using lie detection machines.

Did you know there are a.i. Checker, Detector, and Humanizer companies that want your business?
- Checkers determine if written materials were written by a.i. or humans.
- Detectors check your a.i. version to make sure you won’t be flagged for plagiarizing someone else’s a.i. Some promise to make your a.i.-written content undetectable.
- Humanizers give you control over style and tone to make your robo-text sound more natural, flowing, and “humanlike.”
If you think that you will prove your innocence by taking a lie detector test, stop right there
Many factors influence polygraph test results. The room, the time of day, the situation, medications taken, the test examiner, and how long the test takes all affect the outcome.
More than a few honest people seem to be lying because they’re nervous about the test and the possible negative outcomes
At the same time, there are many dishonest people who seem to be telling the truth during polygraph exams when they’re actually lying.
Sociopaths typically pass lie detector tests with ease because they:
- Are cunning, manipulative, and deceptive.
- Remain calm under pressure.
- Are emotionally detached.
- Lack remorse and the ability to empathize.
- Don’t exhibit the typical signs of anxiety.
- Distort baseline measures.
The big question
What if lying is just too complex for any machine to reliably detect, no matter how advanced its algorithm is? Theorists say no evidence exists that demonstrates deception can be detected by a unique set of physiological reactions.

This week’s article was written by David Allan Van Nostrand
* I use the lower-case a.i. as my abbreviation because it is never confused with Al or AL when using upper-case sans serif fonts.
** What the heck is autonomic arousal?
Autonomic is science-speak for involuntary or unconscious, meaning it happens by itself. Arousal is a term that covers strong feelings and excitement. It applies in other situations, but is mostly associated with sexual desire.

The physical symptoms of autonomic arousal include rapid breathing, increased heart rate, and sweating. The emotional symptoms are associated with tension, fear, and excitement.
Bonus
It’s been said that Onan was a habitual autonomic arouser. Onan is a character in the Book of Genesis. His offense was refusing to fulfill his religious obligation to impregnate his brother’s widow. He did so by committing coitus interruptus to “spill his seed on the ground.” Today the term onanism has a broad definition that includes using the withdrawal method, masturbation, and any “improper emission of seed.”
None of these three conditions is yet a part of lie detector protocol.
