« Taking another turn ... | Main | Stealth and the Laws of War »

Friday, January 06, 2006

New Lie Detectors on the Horizon?

A student in my Criminal Law class this Fall, Dominic Lindauer, sent me this interesting email over the break. You quickly may gather that he comes out of a tech background prior to entering law school:

“Over the break, I've been reading up on technologies that combine some clever engineering with breakthroughs in neuropsychology … and there are several interesting technologies being deployed to determine when people are lying. Given the impetus provided by homeland security and corresponding funding, I would guess practical equipment and protocols (easy to use and not too expensive) will be deployed in under five years. The interesting thing is that the initial results now are pretty definitive (as to whether you're lying), and getting better. Apparently there are even techniques which may allow discrimination between a person who really believes what they are saying, and an intentional lie or evasion.

“Which begged this question in my mind: What is the purpose of the jury? We’ve been taught that their job is to find fact, while the judge applies the law. But what about the up and coming technologies that appear to have promise to determine when a person is telling the truth?

“As a defense attorney, how would you use a magic box that can tell if someone is telling a lie? How as a prosecutor would this affect your approach? Sure would put an interesting spin on civil matters where the burden of proof is lower...”

I do not know much about these developing technologies described by my student. My student promised to share some of his literature, but I wondered whether any readers know something about this technology. Does it test truth-telling more accurately than old-school lie-detectors? If so, how? And, is this technology really on the horizon?

If anything like what my student describes, this technology soon may raise a host of tricky questions for our current, very human method of jury fact-finding and decision-making. Would such machines eliminate the need for juries altogether? If not, would juries retain any fact-finding role, or would they assume some sort of “voice of the community” role, where they take machine-driven data about witness credibility and filter it into a verdict? According to my student’s email, this technology even may differentiate between witnesses telling the truth and witnesses “who really believe[] what they are saying” – the George Costanzas of the world, I suppose. But, would this technology further distinguish between lies and firmly-believed mistakes – which many people mischaracterize as “lies?”

Even if this technology some day can fill all these fact-finding roles, do we want it to? Would it spare us from the inherently flawed subjective fact-finding of human juries? Or, would we lose something more important to us if we replace the inherently flawed fact-finding of juries with the fact-finding of high-tech “truth-telling” devices? I would worry about losing a place for human values and conscience in our system of justice. And, I suspect that many people would feel uncomfortable with allowing machines to return verdicts instead of identifying witnesses simply as “truthful” or “not truthful.” But, do juries separate their fact-finding function from their decision-making role, or do they go hand-in-hand?

Posted by Brooks Holland on January 6, 2006 at 05:25 PM | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference New Lie Detectors on the Horizon?:

» Are juries just lie detectors? from Novus Diem
The best analogy is really DNA and other technological evidence; it brings the same worries about expert witnesses replacing the jury and privacy invasion. My guess is that it would be another tool to provide better justice, making the jury more accu... [Read More]

Tracked on Jan 7, 2006 3:04:39 PM


Such a device would need several critical features:

1. Accuracy. The device would need to detect intentional falsehood at an extremely high rate. I'm talking 99.99% or something like that.

2. Reliability. The device would need to deliver the same level of accuracy and repeatability over time.

3. Transparency. The source code for the software and the design specifications for the hardware would need to be completely and totally open source. That doesn't mean that the device should be readily available -- simply that everyone should have access to knowledge about how it works. Governments could restrict the sale of some key component in order to prevent mass production of the device (the societal effects of the widespread availability of such a device would be far-reaching and not all for the good). And the administration of the testing itself would need to be done with transparency as well in order to prevent skewing of the results.

Of course, even a device designed within these parameters would not end the need for a criminal justice system. A device of this kind would not be able to discriminate between a lie and a mistake of memory or perception, and such mistakes are often just as critical in the criminal justice system as outright falsehoods.

Having said that, I would have to say that a "lie detector" which embodied accuracy, reliability, and transparency would have a profound effect on the criminal justice system. The results from this device would almost certainly be admissable as evidence in criminal trials. If a criminal couldn't credibly deny his or her guilt, he or she would almost always plead guilty, since otherwise the results of his or her test would go in front of a jury. Conversely, if a criminal could state his or her innocence with total credibility, what prosecutor would try to convict him or her? What prosecutor should be allowed to try?

I would suspect that the real barrier here will be accuracy. There will always be freakish people who have the ability to lie with utter credibility, and there will always be some subset of them who can even beat the machine. The article's content and the prior comment notwithstanding, the way to defeat any "lie detector" is to control the body's response to stress for a discrete period of time while you convince yourself that the lie is actually the truth.

Posted by: anonymous | Jan 8, 2006 1:34:14 AM

"Apparently there are even techniques which may allow discrimination between a person who really believes what they are saying, and an intentional lie or evasion."

I thought that's the only thing lie detectors already and have always been employed to do.

Posted by: rodgerlodger | Jan 7, 2006 3:17:07 PM

I think the article is question is here:


Posted by: Rodney | Jan 6, 2006 7:51:52 PM

The comments to this entry are closed.