Will we ever have a true lie detector?
And will any of them be admissible in court?
The quest for a method to detect deception has been ongoing for several hundred years. Our tradition of oath-taking in court prior to giving testimony comes from a medieval practice of requiring witnesses to swear before God, placing their immortal souls at risk if they were to lie. The notion was that no earthly consequence of misconduct, even being drawn and quartered (look at the closing scenes in Braveheart if you don’t know what this is), could trump spending eternity in torment. That method didn’t work any better then than it does now.
The polygraph is still the gold standard for truth verification, although there are as many people who condemn the polygraph as are willing to rely on it. Computerized voice stress analysis (CVSA) is popular in some circles, possibly because it’s neither as expensive as a polygraph nor does it require as much training to learn to use. Unbiased tests of the CVSA place its reliability at about the level of a coin toss, which in most cases will be cheaper than a CVSA machine and the training course.
The latest contender for a deception indicator lies (no pun intended) with research at Columbia University and elsewhere on a technique labeled as voice biometrics. Voice biometrics uses a computer to analyze variations in speaking volume, pitch, tempo and word usage to determine if the statement under the microscope is the truth or a lie. The technology is also useful for determining whether the speaker is angry, stressed or drunk. That might not have as much value in truth assessment, but it could be useful in interviews conducted for other purposes.
There has been a lot of research and training in kinesic interviewing techniques, where interviewers are taught to look for indicators of deception. What qualifies as an indicator of deception depends on who you ask. I’ve heard one speaker say that someone who looks up and to the left while speaking is formulating a lie, while a glance in the opposite direction indicates truthfulness. Then I heard another “expert” say the opposite. If there really is any science to this, it probably requires that the interviewer first establish a baseline for the subject’s behaviors. I know people who never break eye contact while talking, and others who usually look away, glancing back occasionally to see if they still have your attention. I don’t think either type is necessarily deceptive as a matter of habit. It’s more just the way they behave.
Functional magnetic resonance imaging (fMRI) has been tried as a method of detecting deception, and there is evidence this may be the closest thing we have to a true lie detector right now. Similar to the MRI scans that detect torn rotator cuffs and soft tissue tumors, fMRI indicates portions of the brain or other anatomy that are experiencing momentary changes in blood flow or electrical activity. Neurophysiologists know that certain areas of the brain activate when recalling memories, while others are stimulated when we’re being creative, e.g. lying. The idea is that a lie requires some creative thought, while answering a question truthfully relies only on memory. Similarly, a suspect can be shown pictures of a crime scene and their brain activity analyzed to see if the image is registering as something never seen before (because they had not been present at the scene of the crime), or whether this is old stuff to them.
Don’t look for your detective division to be adding an fMRI machine anytime soon. They cost a million or two, require metal-shielded rooms and huge power supplies, and require the services of some expensive technicians and other people with lots of letters after their names to run.
The best lie detector is probably a very good interviewer, one who can penetrate the defenses of the subject, gain their trust, and then abuse it. You can learn the basics of that in a short training course, but it’s a skill that even the best admit they can’t perfect. For every great interviewer, there will always be several more terrific liars.