“John, did you murder your wife?”



     “But the fMRI lie detector indicates that you are lying.

Have you ever imagined a scene like this? Science fiction novels, like The Truth Machine and Midshipman’s Hope, always portray lie detectors as magic mind-reading tools. In real life, many believe that the modern brain imaging technology —  functional magnetic resonance imaging, or fMRI  has the potential to detect lies and be used as evidence in court. However, research indicates that neither science nor law is ready for that stage.

The search for a valid lie detection method dates back to ancient Rome, when people believed in the powers of a marble mask called “The Mouth of Truth.” Legend had it, that if someone put their hand in the mouth of the mask and told a lie, the “mouth” would bite off their hand. The first scientific approach to lie detection was in 250 B.C., when the Greek physician Erasistratus noticed that the heart rate of his patients increased when they lied. In the 1920s, Harvard psychologist William Marston developed the first polygraph that detected subjects’ blood pressure, heart rate, respiration rate, and several other physiological reactions. Today, however, the accuracy of the polygraph, which ranges from 81% to 91%, is not satisfactory for use as courtroom evidence. Thus, polygraph use in the United States is illegal for any non-government application.

Recently, increasing attention has been shifted to brain imaging as a means of lie detection. Among modern imaging techniques, fMRI is thought to be the most promising approach because of its high spatial resolution and broad applications in scientific research.


A series of fMRI images.

Photo by: American Academy of Neurology

fMRIs might sound familiar, how exactly do they work? An fMRI detects neural activity by tracking the oxygen levels in the brain. Oxygen is carried and distributed around the body by a protein called hemoglobin. Oxygenated and deoxygenated hemoglobins behave differently under a strong magnetic field. The magnetic field combined with radio wave can cause molecules to orient towards a specific direction. Oxygenated and deoxygenated hemoglobins send out different signals when they revert to their original orientation. By measuring this signal difference, the fMRI can reflect the oxygen level in certain brain regions. In addition, when a specific region of the brain is activated, more oxygenated blood flows to that area. By reading fMRI results (presented as 3D models that reflect brain activation level with different colors) researchers can know which brain region is more activated.

Research suggests fMRI can be used in lie detection because certain brain areas are more activated during deception. According to a recent meta-analysis, four brain areas— the prefrontal cortex, anterior cingulate cortex, insula, and inferior parietal lobule— usually show increased activity when the subjects in an experiment lie to the researcher. Those four regions are responsible for executive control and attention, since deception is a “high stake” task that involves more control over memory and behavior. Based off this knowledge, analyzing patterns of brain activation during truth and deception trials seems to be a promising approach to lie detection.

It is difficult to differentiate whether brain activation is due to deception or merely false belief.

However, no brain area is consistently more activated during deception. Instead, there are many confounding factors that complicate the brain activation pattern. One factor is memory. A 2012 study showed that, instead of deception, it is memory that triggers the activation of the inferior frontal gyrus, which is the region of prefrontal gyrus that is observed to be more active in most studies on fMRI lie detection.

Imagination and emotion are other confounding factors. A study at Columbia University revealed that what we imagine can actually change brain activation patterns. When subjects were presented with a novel picture, those who believed that they had seen it before exhibited a different brain activation pattern from those who correctly categorized the picture as novel. These findings suggest that it is difficult to differentiate whether brain activation is due to deception or merely false belief.

Emotions can also make lie detection fMRI results less reliable. In a New York University study, the subjects were asked to perform two different versions of the classic Stroop test, and one of them involved emotions. In the regular Stroop test, subjects are asked to name the color that words are printed in, where the words themselves were also names of colors. In the modified Stroop test, the words were not colors but were names of different emotions. The only difference from the original Stroop test was the meaning of the words, and the subjects were only asked to name the color of the words. However, the brain regions activated in each task was different. In fact, the regions activated in the anterior cingulate were different in the task involving emotions. In other words, even though the meaning of the words was not central to the experiment, emotional phrases still had an effect on brain activation. When lie detection is actually used, in a criminal investigation, for example, emotion fluctuations may be quite common.


An example of words and colors used in the classic Stroop test. The recent variant of the test replaced the text with emotion words and asked study participants to simply name the color of the text.

Photo by: OjoHaven

Current fMRI and lie detection research also fails to mimic real-life conditions. Participants are typically offered a small sum of money if they can successfully deceive the experimenter. If stakes were to be higher, as in a real life situation, brain activation might be different as more factors, such as stress and emotion, are involved.

Additionally, fMRIs are also mostly tested on healthy young college students, when in reality, people from various demographics, such as criminals, are subject to lie detection methods. A 2009 study showed that criminals’ brain activation pattern during deception was different from those research subjects.

fMRIs are, unsurprisingly, not ready to be implemented in court. In the hallmark 2012 case United States v. Semrau, fMRIs were rejected as lie detection evidence. According to Federal Rule of Evidence 702, scientific evidence must have a known error rate and obtain general acceptance in the scientific community— both criteria of which are not satisfied by fMRI lie detection. In addition, sensitivity, the correct identification of lies, of fMRIs range from 69 percent to 100 percent—  a huge variation. The specificity, the correct identification of truth, can even be as low as 33 percent.

We may have high expectations for fMRI and lie detection, but for now, it appears that lie detection devices in courtrooms are still the stuff of science fiction.

About The Author

Yechen Hu