I know there is a lot of skepticism about lie detectors. But is it really that hard to detect lies with some degree of accuracy? How good are trained people at this? What about cameras and advanced algorithms. When people lie, they have to fake a reaction and then make something up. Is that really so easy to do? Have there been any studies to test accuracy at detecting lies?
Friend of mine has been an FBI agent for close to twenty years now and claims he can instantly tell if someone is not telling the truth. I tend to believe him.
He could probably tell when most people aren’t telling the truth, but there are some cold-blooded sociopaths that I bet could fool him.
And that is probably the sticky point. Unless the equipment is 100% fool-proof there’s no way it could be used to convict someone of a crime.
Never mind a crime, how about all this Herman Cain stuff? Was any of the stuff about Bill Clinton true? Climate change? One problem is so many people are so subjective that I think they actually believe anything favorable to them.
Actually, you’ve answered the question yourself.
“… some degree of accuracy?” – Sure! No problem!
But we want a lot more from a lie detector than merely some degree of accuracy.
Yes it is, and it’s understating things to say there is skepticism about lie detectors. Polygraph exams are inadmissible in court because they’re junk science. There’s a body of evidence that shows they simply don’t work. At all. Other claims, like the use of voice stress analysis, brain fingerprinting, micro gestures, etc, have all been debunked by peer-reviewed academic research studies.
Veteran detectives all believe this about themselves. Hundreds of exonerations (based on actual science like DNA) suggest that this is just hubris. Put one of them to an actual test in controlled conditions and the results will speak for themselves.
So what do these tests say? That they cannot detect lies at all? Or just a little bit better than random chance?
If you are testing for stress - what are the odds that someone who has been arrested and is being held in a cell for an extended is stress-free? Especially if they knew that how they answer this set of questions could make or break the next 20 to 50 years of their life? Are they thinking “I’m lying” or “Holy crap! This is the money shot question!!!”?
The prosecutor in the Duke Lacross rape case knew that a rape had been committed too; he didn’t let details like flawed ID’s, no DNA, etc. dissuade him from proceeding with the case.
The real test for your friend would be to have him do an interrogation with no preknowledge of a case. After all, the vast majority of people arrested by the police ARE guilty, so it’s a slam-dunk when they are lying; the police see all the tricks, so a not-too-bright perp trying to fool an experienced policeman who already knows what’s happened is easy to read.
One of the issues I’ve heard is that in some cultures, looking someone in the eye is not acceptable - it’s more like a challenge. Local law enforcement tend to misinterpret this failure to make eye contact as evidence of lying.
But how do you know he’s not lying?
To give a concrete example, suppose you have a lie detection method that is right 90% of the time. That’s a pretty high standard. However, suppose you apply it to a group of 100 people, 90 of whom are telling the truth and 10 of whom are lying.
- Of the 10 liars, your method will say that 9 are lying and 1 is telling the truth.
- Of the 90 truth tellers, your method will say that 81 are telling the truth and 9 are lying.
So you will end up calling 18 people liars, and half of those accusations will be false.
He is lying but since you have not been trained, you can not tell.
Lie detector results are not accepted in court because they do not work well enough. There are plenty of people who get false positives and others who can beat the machines.
That’s my understanding. Lie detectors are not junk science - they are better than random at detecting lies. But they’re not good enough to be admissible in court.
Like cjepson’s example. It’s useful in an investigation to have a means of getting information that accurate ninety percent of the time. But you wouldn’t want to use it in trials where it would be falsely convicting one out of every ten people.
Confirmation bias, on his part, at its finest.
Probably. I’m sure he can and has been lied to and fooled many times. My guess is that he is probably better at detecting it than your average person.
I ain’t going to look for it, but IIRC there was a study that came out in the past few years where some researchers found that a small subset of people (something like 1 percent?) were VERY good at detecting when other people were lying. And I’d suspect that due to the nature of this study, people were told to lie about stuff that didnt really matter, so I’d WAG these superhuman lie detectors are probably even better at detecting lies that matter.
Though, as others have pointed out, when you pit superdectors against sociopaths/delusional types it probably gets a bit more iffy.
, during training we got to play around with lie detection equipment. Here’s a good example of what can go wrong.
I had recently had a friend die of AIDs because her husband cheated on her. Another student was in an affair with a man whose “wife” was on year eight or nine in the mental institution. He didn’t divorce her because then she would lose the insurance coverage that kept her safe and comfortable.
Hook us up to the lie detecter and ask us each if we’ve ever cheated, ever slept with a married man, ever been unfaithful. (or something similar, I don’t recall the exact detailed queries.)
My readings went completely off the charts in all three. The subject was upsetting, the questions were insensitive and the implications that I might do such a thing were offensive.
My colleague was very relaxed, and comfortable with her decisions. She decided to lie to the machine, and got through it flawlessly. Why? No guilt. Not that she wasn’t technically guilty but she felt no guilt.
Scroll forward a decade or so - I was completely taken in by, and nearly married an actual full-on psychopath. To me he was one of the most relaxing people I’d been around in years. I told everyone from the moment I met him “He just has an air of integrity about him.” Why? No tells. No responses, no guilt. Ever. The most dangerous man of all was the one who never showed any trace of lying.
Real people lie quite a bit, and they show signs of guilt regularly. Do your FBI friend a favor and just check to be sure he knows that. It’s part of the training now, but it wasn’t back when.
I just read an interesting book called Quirkology that spends a good bit of time discussing the subject. The author has been studying lies and falsehoods for years. In one of his more interesting studies, he recorded two videos of the same person answering a set of questions. In one video, the subject lied; in the other he told the truth.
The author then had three different groups of people evaluate the two videos and report on which one they thought was truth and which one lies.
Group 1 saw the videos normally.
Group 2 heard the audio, but could see the video.
Group 3 saw the video with no sound.
Guess which group did the best job at detecting the lies.
Group 2. Looking for visual clues was a distraction, and conventional wisdom (liars shift their eyes more, etc.) is flat-out wrong.
He performed a lot of very interesting studies. If you’re interested in the subject, I recommend reading his book.
The OP may be interested in the article “Why Don’t We Catch Liars?” (PDF file) by Paul Ekman, who’s something of an expert on lie detection and body language.
If someone is sure that they can tell someone is lying, he ought to make a good living playing poker.
One of my wife’s friends is dating an FBI agent who doesn’t believe in evolution. If they have outright morons working there, I’m sure someone who over-estimates their lie-detecting acumen could make it.