Althouse: "Speech-based 'lie detection'? I don't think so."

November 11, 2011

"Speech-based 'lie detection'? I don't think so."

Mark Liberman excoriates the press for believing some guy who purports to know that Herman Cain is not lying.

20 comments:

Anonymous said...: When I worked for the State Department, there was a bunch of training about something called micro expressions. This was for consular work -- giving out visas.

The basic idea behind it that you can detect facial expressions in people who are lying because their face shows what they really are thinking before they are able to mask it. It happens quickly and instantaneously in real time. But when you look at people in slow motion, you can really see certain expressions pass across their face that show what they are feeling.

It never worked with me when I was giving out visas. I didn't have video. But maybe somebody should try that with Cain.; 11/11/11, 1:58 PM
YoungHegelian said...: Wait! Are you trying to tell me that someone who's using free news publicity to hawk his product is stretching the truth?

Go On! That never happens!; 11/11/11, 1:58 PM
Scott M said...: Hmm. So, you take some drugs and you can tell if someone is lying?; 11/11/11, 2:07 PM
Nora said...: It's old news that it's easier to spot a lie when listening to a lier then when also looking at a lier. However, it's the first time I hear about listening device for detecting lies.

I guess at the moment it would be difficult to find out whether there is science behind this thing, but I'm curious.; 11/11/11, 2:26 PM
Anonymous said...: Hmmmm, I wonder if it still works when he is singing.; 11/11/11, 2:28 PM
TMink said...: One of the problems with the scientific voice stress analyzer literature is that so much of it is conducted by people who are paid by or practice polygraphy. Or using polygraphs, or polygraphics.

Whatever, you see the potential for bias. But even if the devices work and are reliable Mr. Cain was not involved in a controllled environment that the machines need to work effectively.

I support Mr. Cain, and I am skeptical of the quoted critiques, but I am more skeptical of the machine to work by analyzing someone talking on television.

Trey; 11/11/11, 2:36 PM
edutcher said...: Sorry, the report doesn't sound all that "credulous", but they're not dismissing it as "fake but accurate".

For the Lefty media, maybe that does count as "credulous".; 11/11/11, 2:41 PM
Anonymous said...: It is my understanding that if you clench your asshole while responding to every polygraph test question, you will pass.; 11/11/11, 2:43 PM
Jason said...: It is true that you can, in a controlled environment, assess a greater or likelihood of truthiness by analyzing eye movement, facial tics, and other unconscious means of expression.

However, normally, you would make the assessment after having gone over a number of control questions, whether unbeknownst or beknownst to the subject or not. For example, most of us are "right-brained" for creative and left-brained for logic. When we're making things up, we tend to look to one side of our heads, rather than the other. A certain number of people have the opposite wiring, though, so this particular tell has limited validity until you have first established whether the subject is, indeed, wired in the usual way.

It's more than just polygraph. A good, professional interrogator doesn't need machinery, except to the extent the machinery lends credibility in a court case.; 11/11/11, 3:48 PM
traditionalguy said...: It sounds like very Sour grapes because Cain got the immediate benefit of a Jury listening to live witnesses under testimony and cross all without the show trial and years of delays.

That is why Juries get it right. It is a human skill to know a liar by a 6th sense. Now a computer can be programmed to do it too.

What hath God wrought?

Will Mark Lieberman also call for a riot over this exposure of carefully concealed lies like Paterno and his boy lover sex ring called for on Penn State's campus last night?; 11/11/11, 4:14 PM
TMink said...: Seven, I had heard a similar story but it involved putting a thumb tac in your shoe and stepping on it after random questions to introduce noise and make the polygraph unusable.

I think the pain response would be more confounding than a clinch, but I have no info about whether the tac strategy actually works.

Trey; 11/11/11, 5:05 PM
Larry J said...: It's old news that it's easier to spot a lie when listening to a lier then when also looking at a lier. However, it's the first time I hear about listening device for detecting lies.

This isn't the first time I've heard of this. I first heard of it back in the 1970s. Allegedly, someone used it when Nixon was speaking (IIRC, his "I am not a crook" speech) and the test indicated he was lying. I'm not saying the technology is accurate or not, just that the idea has been around for almost 40 years.

For what it's worth, here's a Wikipedia article on Voice Stress Analysis. As always, take it with skepticism. Here's a link to freeware VSA software if someone wants to play with it.; 11/11/11, 5:16 PM
Darrell said...: "Microexpressions"[pushed by Dr. Paul Ekman who was the basis for the character of Cal Lightman on Lie To Me], voice stress analysis, body language, etc. are all methods of lie detection that turn out to be slightly less reliable than flipping a coin when the results are subjected to post-hoc scientific scrutiny when the actual facts were known. The US Government has spent hundreds of millions of dollars on this bullshit since WWII because everyone is looking for an edge when knowing the truth quickly is critical. Trained, experienced detectives and interrogators usually check out around 70% accuracy when subjected to the same sort of analysis, but nobody wants to hear that.; 11/11/11, 5:30 PM
Peter V. Bella said...: If it was Bill Clinton, Barack Obama, John Kerry, Joe Biden, or Al Gore, would we be having this conversation?cothahui; 11/11/11, 5:38 PM
Anonymous said...: Darrell -- The only problem is that you don't get to interrogate anyone at the visa line. There is no time and no money.; 11/11/11, 5:47 PM
Darrell said...: Well, Seven, flipping that coin would have given you a slight edge over microexpressions scientifically. I'm sure the Franklin Mint has a Hillary coin that the State Department can use in the field today.; 11/11/11, 6:08 PM
Anonymous said...: I agree that microexpressions is useless. But interrogation is cost prohibitive for visas.; 11/11/11, 6:12 PM
MisterBuddwing said...: OK, here's my .03 - supposedly, if you rehearse your lies out loud, over and over again, your bodily responses won't betray you when you undergo a polygraph examination.

Supposedly.

WV: scrac; 11/11/11, 10:27 PM
Nora said...: Larry J said...

For what it's worth, here's a Wikipedia article on Voice Stress Analysis. As always, take it with skepticism. Here's a link to freeware VSA software if someone wants to play with it.

Thanks. I thought that even if technology exist, it would not be able to work without initial callibration on responses data for tested individual, like a set of 'true' value questions at the start of the lie detector tests to analize the test results against.; 11/12/11, 2:03 AM
Darrell said...: Here's a better reason why these
mathods of lie detection will never go away--You only make yourself a target when you pull away from the herd. When you are wrong making a personal judgement, your career suffers. By using the "tricks" that everyone else is using, you have something to point to in the morbidity and mortality inquisitions that follow the explosions. That is why bankers are always looking to issue loan in the "hot" areas, whatever they may be--energy conservation, "dotcoms", etc. When they blow up, no one ever
assigns any individual blame. Heck, everyone was doing it.

And if you are ordered to use it, as you were, Seven, there is even more reason to do so. Mine was just a comment, not a criticism.; 11/12/11, 5:08 AM