Slashdot Mirror
Lie Detector Glasses Coming Soon
Zelphyr writes "The EE Times is reporting on a product soon to be released by an Israeli company that allows the wearer of special glasses to tell whether the person they are talking to is telling a lie. Not only that, they can tell you whether someone loves you! Apparently a PC version of the 'love detector' is in the works as well. Think my Windows box will be upset when it knows how much I hate it?"
I can just imagine what these things will do to the poker world! sheesh
Remember back in 96-97 there was a big rage in "lie detection software" which supposedly would analyze audio input of someone speaking and then match their voice stress level to either "True" or "False" indicators?
It was crap. I think more than a few morning radio shows tried to use it on their callers with failure after failure. I tried a copy myself and found that not only was it horribly written, but even if you were able to get the subject to "train" it (by answering several questions that are known to be true) it gave inncorrect responses virtually half the time.
Come to think of it, the software might have been made by an Israeli company too. Maybe the same one, I don't know. Can't remember the name but it was horrid.
Do I think the FBI/CIA might have technology like this, to analyze voice stress or facial temperature and determine if you are lying? Sure, why not. But there's a reason why lie detection technology is not admissible in court. It just doesn't work. Too many experts can beat it and too many amateurs get nervous and give false positives.
-JoeShmoe
.
-- I wonder which will go down in history as the bigger failure: the War on Drugs or the War on Filesharing
This is not a polygraph. Take your stats elsewhere, or supply some stats relevant to lie detection based on voice. (NO, that is not a polygraph.)
I Mean it is hard, but what if ANY other country with such a horrible Human Rights record as Isreal came up with a lie detector 'prototype' and claimed to market it.
Would you buy a detector from N.Korea (or S.Korea), Pakistan, Syria, Jordan, Saudi Arabia, Chile, Argentina, Turkey, Belarus, Angola, Guatamala, Uganda, Or The *n*t**d St**t*s.
Even the Best polygraph tools are only 50% accurate. It isn't anti-semetic to call a country which has never kept a treaty or accord in my entire life 'Isreal' but it is a little more than shady to use their tools and methodology in any (so called) War on Terror
--Shaddup and support your local PBS station Plan for it
So let's assume the 90% accuracy figure is not bullshit (which it probably is). That means 1 out of 10 innocent passengers will be harrassed as suspected terrorists and 1 out of 10 terrorists will be allowed through. Not especially comforting thoughts in either case.
Want to learn how to lie well? Just practice your campaign speech in front of this thing.
Yeah, but the psychological power of being hooked up to a machine that can tell if you are lying is huge. Sure, the system can be beaten, and that has been proven. But most people don't know that, and furthermore don't know how to beat it. So they might be willing to divulge the truth more readily if they believe that if they lie they will be caught. That is why the term "lie detector" is much more ominous than "polygraph".
Is there a way to programmatically tell if someone is lying? I think there are general "tells" that most people do when they lie, and a computer can be taught to recognize these. But I don't think it will generally be accurate enough to escape harsh (and well deserved) scrutiny from the scientific community.
My beliefs do not require that you agree with them.
It reminds me of the old joke about a mental patient that believed that he was Napoleon Bonaparte. After many years of treatment, he was ready to be released. As a final test, they gave him a polygraph examination. When asked whether he was Napoleon Bonaparte, he said no. The polygraph examiner concluded that he was lying.
Mea navis aericumbens anguillis abundat
Actually, that *is* a fairly interesting proposition. Even if it were terribly inaccurate at reading the subtleties of emotional responses, maybe it could be used by machines as a source of additional input. Really now, imagine a kiosk at a retail clothing chain which offers you selections on what you might want, and gauges from its love-o-meter readings how strongly you hate the silk-sheen mauve stretch-fabric tee shirts and love the traditional white polo. At an even more granular level, such a kiosk would be able to gather tiny bits of information on what shades of which colors, what fabrics in what cuts, and such that you prefer. And all that with less interaction than would have been required otherwise.
If you think about it, this kind of technology, if moderately effective and economically manufacturable, could be applied to any expert-system-type application that guides users to recommendations. Just imagine: An interactive porn catalogue that requires NO hands to operate (now *both* hands can be free)! Okay, that isn't my ideal application of the technology, but it's worth consideration.
Polygraphs have valid uses. Just ask anyone who's going through a background investigation starting with the EPSQ. I'm sure that some people have learned to "fool" the system, but a trained operator will detect most.
Sciam is a nice science enthusiast magazine. As is the nature of the publication, the article is light on details. The magazine a good tool to be introduced to new information, but I would prefer to read about some hard studies, such as in JAMA
(They do have this to say)
Something I feel compelled to point out, that is a common irritant in much I read (Yeah, I admit it. I try not to, but guilty of it too): You have a logical fallacy in your assertion that I'd say they're as objective as you get, unless of course you believe in some kind of "science-conspiracy".. (Check out Wikipedia logical fallacy
You make the assumption that one needs to believe in a "science conspiracy" in order to presume that the magazine is not "as objective as you get."
Bollocks. I don't believe in science conspiracy, yet I don't know the credentials of a particular journalist, so I can't assume that that particular journalist is completely objective or knowledgeable enough to report fully and accurately.
Having said that, I personally dislike polygraphing, I think it is intrusive, like a mental form of body cavity search.
Unfortunately, we don't live in a nice world, and sometimes the polygraph is a tempting, and if it IS accurate, then a useful tool. A problem of polygraphing is potential abuse. I hear of abuse stories a good deal. How many are true, how many are fabricated? I don't know.
Dogma - "let's just say we'd like to avoid any empirical entanglements."
Polygraph tests work great if the guy is dumb.
And only a few smart criminals get caught.
Metafilter pointed me towards a really interesting model for managing deception: Recognition detection. The idea is, rather than find out if someone is lying or not, simply find out if they recognize an object or scene they could only recognize if they were guilty. A certain brainwave, coined the P300 Wave, is emitted within a certain number of milliseconds of seeing an item one recognizes. One study, done by a group called Brain Wave Science, was able to reliably (and perfectly) separate FBI agents from average civilians by showing pictures of items from FBI training courses and operations. Detailed information may be found here.
I, of course, make no claims as to the veracity or accuracy of this material. But this wave is not pure pseudoscience -- the NYT has an article showing how weak P300's correspond to weak signal recognition. And BWS isn't the only group looking into P300 and deception.
There are other approaches -- blood flow and PET scans come to mind -- but this has the advantage of involving just a few electrodes.
So -- we may yet see a lie detector functional in our lifetime. Of course, it won't always be trusted, for reasons similar to the legalistic need for occasional exceptions to the rule of unique suspect DNA identifiers. But it'll be there.
--Dan