European Researchers Develop More Accurate Full-Body Polygraph

jfruh writes: Despite their widespread use in industry and law enforcement, traditional lie-detector polygraphs give accurate results only about 60% of the time, barely better than the 55% accuracy people can get just by following their gut instincts. Now researchers in the UK and the Netherlands are trying to improve that. They claim a full-body polygraph based on motion-capture suits used for movie special effects can detect lies with 75% accuracy.

Comments

Accuracy

[Fire_Wraith]

75% of the time, it works all of the time!

Re:Accuracy

[Dorianny]

polygraphs are as much a science as astrology.

Re:Accuracy

[Jumunquo]

Exactly. There's no reliable body response for a lie. All they are measuring is nervousness, which you could have for a variety of reasons. It's the same thing the border agent does.

The purpose of the polygraph is to bully the victim into a confession. The unknowing victim thinks they are undergoing a scientific test, but they are actually being drilled by a skilled interrogator w/ no lawyer present to defuse the loaded questions.

Re:Accuracy

[Fire_Wraith]

The difference is that we don't as a society (generally) rely on astrology for anything of serious consequence. With polygraph tests though, they're used to screen for employment in critical defense and intelligence functions, and in legal proceedings. Even though it's not compulsory, the gross inaccuracy should rule them out for any serious consideration even when someone agrees to take it. Even 75% means a 1 in 4 failure rate, and regardless of how many of those are false positives vs false negatives, that's still way, way too high to be anywhere close to considered effective.

And yet, so many people have the erroneous impression (from Hollywood or elsewhere) that these devices are 100% effective.

Re:Accuracy

[s.petry]

Yup, the basics of the "classes" people give to beat a polygraph are summarized as relaxation techniques. (Oh noes, will I go to jail for revealing the "secret" that's all over the internet? Oh, I'm okay because I didn't charge for it.. *whew!*). Mask the nervousness and you can spin some wild tales while hooked up and look to be absolutely truthful. Don't get me wrong, it takes a bit of practice but has been proven to work repeatedly. Most often by former "experts" in polygraphs that want to prove what a sham they are.

Re:Accuracy

[Jumunquo]

Well, that sort of illustrates why the test is wrong so often, but if you TRY to be bored and aggressive, you'll probably think too much ;)

The way they teach you to beat it is pretty simple. Before they ask the hard questions, they need to calibrate. They will ask you something easy to get a base truth response. Likewise, they will get a base lie response. You want your base lie response to go sky high so that nothing can ever match that. Bite your tongue. Tighten you ass. Whatever you like. Once you're past that, then strain yourself a little and answer all the questions like that. Just remember that you're not in the clear yet because you've handled the problem with the junk science handing out false positives, but don't forget the person across from you is not a scientist but a skilled interrogator. They can still ask all sorts of loaded questions. For example, have you ever done something personal during work time? The interrogator is much better at this game than you are and probably has more mental stamina than you do. That's why they say if you have a choice, don't take a polygraph - tell them you know how it works, and that it's a sham.

Re:Accuracy

[ShanghaiBill]

Even 75% means a 1 in 4 failure rate

Ever been to Las Vegas? All those sparkling lights and tall buildings were paid for by winning 52% of the time.

Re:Accuracy

[donscarletti]

Exactly. There's no reliable body response for a lie. All they are measuring is nervousness, which you could have for a variety of reasons.

A polygraph measures nervousness on one axis and time on the other.

The point is not to measure if the subject is generally nervous, it is to measure a nervous reaction to stimulus, usually a question posed by an interrogator.

Polygraph results are not admissible in court, they do not override a suspect's right to not answer questions and unlike torture there is no real evidence that they lead to false confessions. A stupid but innocent suspect could only believe that the polygraph will exonerate them. It's not that judges, prosecutors, the police or anyone in power believes they are more than 60% accurate, what a jury believes about them is irrelevant since if they receive the outcome it's grounds for a mistrial.

A polygraph test is like an IQ test, it does measure something that is very useful for some people in some cases, but what it does measure is very different to what the general public understands it to be measuring. If you however are one of the people who need the information that it actually does measure, it's extremely useful. It seems however that the same group of people have their panties in a knot about both things.

Re:Accuracy

[AK Marc]

And if you think too much, you'll "fail" every question. "fail" the baseline questions, like your name, and they give it to you a few times, then treat it like a pass. They have no other choice. Repeated "inconclusive" is a pass.

They can still ask all sorts of loaded questions. For example, have you ever done something personal during work time?

And they give you the questions ahead of time so you won't be confused/blindsided by them. Have you ever done something personal on work time? Yes. Everyone has. I peed today. Oh, and I stood around the water cooler and talked non-work things to my coworkers. I even took off 2 hours early once when school called to say the 6 year old fell and was in the hospital getting stitches. The manager said to make it up later.

They ask the stupid questions to try to guess if you are lying.

I applied for a minimum wage job at a video store (funny, I've applied to two in my life, didn't get either). They gave a personality profile test. They asked whether I've done drugs. I said no. They asked if I thought drugs should be legal. I said yes. I presume the results (wrongly) indicated I was a lying stoner. The other time, the manager who took my application was a douchebag. I let him know as much. The ironic thing is that I know more about movies than anyone else I know, and I've known quite a few movie buffs.

so it can still be trained for

[Friar_MJK]

just add full body motion to the list of do's and don'ts when you're taking the test. ever notice how often joe biden scratches his nose when he's talking out of his ass? yea, that's called a tell. i remember hearing about an indiana guy that was arrested for training people to beat polygraphs.

75% Accuracy is touted on Slashdot

[Bob_Who]

I have a dowsing rod with better accuracy. My coin flip is 50% accurate. But lets convince everyone here that our standards for the truth are low enough to buy a bunch of polygraph apparatus that is 75% accurate, because technology is just not good enough to get to the truth. The truth is, this is totally stupid.

As if getting a polygraph isn't humiliating..

[mpthompson]

...and stressful enough already. Now they'll tell you to strip and get into a silly motion capture suit. Next up is sticking a probe up our anus to measure the contraction of the sphincter muscles. After all, it's for our own good. How else will our overlords prevent another Snowden fiasco?

Re:As if getting a polygraph isn't humiliating..

[wbr1]

Think that's invasive, try a plethysmograph. Required of sex offenders everywhere.

MRI

[manu0601]

I wonder why they don't use MRI or some other brain activity visualization technique; Recalling memory and forging a new story must be more distinguishable there than on body movements.

Focus on the brain

[righteousness]

Instead broadening the area of interest, why not focus on where lies are actually formed, i.e. the brain? Put more effort in understanding the brain and on finding a way to detect changes to the brain when someone lies.

Re:Focus on the brain

[fnj]

I know what would be even better. Let's use mutant precogs to find people guilty of precrime. Apologies to Philip K. Dick.

Re:50% accuracy is as good as blind guessing

[AK Marc]

Even worse, they don't give the failure rate in a useful manner. 100% for pass (zero false negatives) and 0% for fail (100% of fails reported as a false positive) is easy. You just say "he passed" no matter what the results are. Or the reverse as well. But "75% accurate" is mostly meaningless.

Accuracy of traditional lie-detector polygraphs?

[lippydude]

"Despite their widespread use in industry and law enforcement, traditional lie-detector polygraphs give accurate results only about 60% of the time"

There is no verifiable scientific evidence that polygraphs actually work.

Re:Great news

[Aighearach]

Now we need to make it compulsory for all politicians to use these when discussing their political manifestos to get in power...

So, let me get this straight... it isn't bad enough most politicians are already sociopaths, you want to actually institute a formal litmus test that is inaccurate, but favors sociopaths?

Pseudoscientific nonsense

[Required Snark]

Yes, fitting a random person into a full body rig will have zero impact on the false positive/false negative rates. No problem.

They tested this on 75 volunteers. This is an example of the kind of bogus "proof" that is used to justify the utility of polygraphs in the first place.

It's in the same territory as drug companies excluding tests that show problems with their drugs. I'm sure if they ran enough small groups that they could find one with better then 90% and report only that.

Why do polygraph advocates lie so much?

Voodoo/2.0

[J'raxis]

In other news, the same research group has improved the accuracy of entrail reading by including other internal organs, doubled the accuracy of palm reading by using both hands, and are now hard at work devising ever-larger crystal balls in the hopes of refining their accuracy beyond "total bullshit."

polygraph=bullshit

[ihtoit]

File it with "Scientology bunkum".

Ob

[Hognoxious]

They claim a full-body polygraph based on motion-capture suits used for movie special effects can detect lies with 75% accuracy.

I don't believe them.

Re:Accuracy of traditional lie-detector polygraphs

[Cochonou]

I would also like to know if there is really a "widespread use" of polygraphs. I understood they were almost exclusively used in the US, and that most other countries actually forbade its use as evidence in courts - which would make the use of polygraph a local idiosyncracy rather than a widespread practice.

Re:Accuracy of traditional lie-detector polygraphs

[Carewolf]

I would also like to know if there is really a "widespread use" of polygraphs. I understood they were almost exclusively used in the US, and that most other countries actually forbade its use as evidence in courts - which would make the use of polygraph a local idiosyncracy rather than a widespread practice.

I think even in the US they are not allowed as evidence. They are used as interogation, but yes, I have never heard of any use outside of the US, at least they fell out of favor around the same time as phrenology.