Lie detection through fMRI February 2, 2007Posted by Johan in Neuroscience.
I came across a clip about this on Youtube:
Curiously, they say that this technique “may one day be better than the polygraph.” In fact, a few papers by Langleben et al (2002, 2005) indicate that this may already be the case. Of course, beating the polygraph isn’t saying a lot.
Langleben uses a paradigm where he contrasts activation when participants lie about having seen a certain playing card to when they tell the truth. With this, he claims predictive accuracy in the range of 85 percent. Granted, the ecological validity of this paradigm is limited compared to that used in the video clip (unless you’re in Vegas).
So what is this activation that appears in the “lie” condition, but not in the “truth” condition? Langleben suggests that it represents the participant having to inhibit the truthful response.
For people who are scared of government surveillance through CCTV and such, the notion of an infallible lie detector may not sound like progress, exactly. However, keep in mind that fMRI relies on the participant remaining dead still for the duration of the scan, and I imagine Langleben’s paradigm requires that the participant pay close attention and provides responses, too. In other words, there would be no practical way of running this test on anyone who did not consent to do it. So in court, it would most likely only offer a new way to prove someone’s innocence.
As an aside, why is it that MDs are doing this line of research? The guys in the video are physicians, and so is Langleben.