Asked by: BBC Focus magazine
In 2008, an Indian court sentenced 24-year-old student Aditi Sharma to life imprisonment for murdering her ex-fiance. An EEG-based lie detection test was accepted as key evidence.
Brain-based lie detection has often been put before courts, but in most countries it is rejected as unreliable. In fact, India has since declared it inadmissible. One issue is whether the brain activity seen in lying is indicative of deception.
A major neural ‘marker’ of deception is high activity in the frontal brain areas that concoct stories. The assumption is that lying involves making things up while telling the truth does not. In standard lab tests the marker is pretty reliable, but a similar pattern of activity can occur in other situations giving false results. In one study, volunteers were shown a series of cards and asked to select one without saying which. When they saw the card again and were asked if it was the one they had noted, their brains reacted similarly whether or not they were lying about having selected it. This suggests that merely investing something with meaning may cause your brain to react to it in a different way to other things.
A second question is whether the lie detector itself can be deceived, and a study at the Memory Lab at Stanford University showed how the system can be beaten. Twenty-four volunteers were shown 200 novel faces on one day. The next day, they were shown the faces again, along with 200 new faces mixed in. They then had to hide their memories – when they saw a familiar face, they had to say they didn’t recognise it, and also had to claim they recognised unfamiliar faces (see image). Their brain activity convinced the machine that they were telling the truth when they were lying.
Until lie detection cracks these problems, it is unlikely to appear in criminal courts. Given current progress, though, this may not take long. Meanwhile, Sharma has been released from prison pending appeal.