Problem for Brain-Based Lie Detectors
Developing a reliable technology by which we can read an individual’s brain activity and determine, with high accuracy, whether or not they are lying, would be extremely useful. Such technology might have medical and research applications, but most obviously for law enforcement.
Current techniques use functional MRI scanning (fMRI) to look at brain activity while recalling an event. The structures in the brain that support recalling a legitimate memory are different than those involved in a novel stimulus.
Studies using this technique report 70-90% accuracy in differentiating a genuine memory from a feigned one. This still may not be accurate enough for the courtroom, but a recent study shows the problems for real-world application are even more severe.
Uncapher and colleagues publishing in The Journal of Neuroscience found that it is possible to fool the test. They studied subjects with fMRI scans while they were involved in a task including either studied faces or novel faces. The subjects were instructed to consciously conceal whether or not they were familiar with the faces they were viewing. The authors found that:
“…the ability to decode memory mostly failed, and even slightly reversed, when participants used simple cognitive strategies to thwart classifiers trained on truthful memory responses.”
In other words, when subjects tried to conceal their familiarity with the images they were viewing the evaluators could no longer tell, and in some cases were fooled to the opposite conclusion.
This poses a significant barrier to the real world application of fMRI-based lie detection. It may not be a fatal problem, however, if researchers learn how to detect the countermeasures, but it certainly makes the task much more difficult.