Tuesday, April 23, 2013

fMRI lie detection and the Semrau case

Semrau is a psychologist accused of committing fraud to Medicare and Medicaid. The case became mostly famous, because he asked that fMRI lie detection would be a evidence in court. The judge had to decide if fMRI was admissible and after hearing scientists advocating for both sides, he has decided not to admit such evidence. However, the question is: Will it be possible to use fMRI lie detection one day?, because the reason for not admitting it has been based on the error rates and acceptance by scientific community and that can change any day...

Image from here

So how does fMRI lie detection work at the moment?
- A deception task is presented to the volunteers: they have to lie about the object they have taken from a box (or similar, such as a card from envelope).
- The volunteer goes inside the scanner and structural MRI is performed and a motor task can also be performed to make the volunteers more familiar with the MRI itself.
- The deception task starts and the volunteer is asked questions about the stolen object among other questions. The volunteer has to lie about stealing the object. During this time, EPI (Echo Planar Imaging) images are acquired.
- Processing of data starts, which includes reorientation and motion correction. Brain patterns are analyzed to detect lying.

Findings have shown that there are specific activated areas (anterior cingulate and the prefrontal cortex) in subjects in the task of deception when a group study is performed. This is a group study, but for fMRI to become a lie detector, it has to stand in individual studies. This has been difficult, because fMRI is a technique with a low signal-to-noise ratio, but some studies have been done. Moreover, deception tasks in these studies are still simple ones, while more complex ones (like the Semrau case) have not been performed.

One of the studies which presented results on individual basis (the one referenced here at the bottom) has led that a company has been formed to sell this type of service (CEPHOS). This was the company involved in the Semrau case and the CEO of this company is the scientist advocating for the fMRI lie detection. The two scientists which advocated against the fMRI lie detector were Marc Raichle, PhD (Wash. U. St. Louis, Neuroscience) and Peter Imrey, PhD (Cleveland Clinic, Statistics).

Anyway, my personal belief is that fMRI should be on the service of health and not of law...

Other Links:

Kozel, F., Johnson, K., Mu, Q., Grenesko, E., Laken, S., & George, M. (2005). Detecting Deception Using Functional Magnetic Resonance Imaging Biological Psychiatry, 58 (8), 605-613 DOI: 10.1016/j.biopsych.2005.07.040


  1. When a polygraph test is conducted properly, by a skilled and experienced examiner, accuracy will be extremely high, in the 95-98% range.

    Polygraph technology is relatively inexpensive and 50 years of high quality research demonstrate its accuracy and efficacy.

    Given the expense of fMRI technology, and the paucity of research regarding its accuracy, there is little reason to spend the time and money on such a test.

    Louis Rovner, Ph.D.

  2. Dear Dr. Louis Rovner,

    Thank you for your comment. Indeed, you make a good point about the different expenses between the two technologies. However, that is usually not a reason to convince researchers to stop going on. I guess many researchers will try to show in the next years that fMRI can be useful, specially those who have started companies based on such services...

    Nevertheless, desperate people trying to prove their innocence might pay for these services...

    Just like I stated at the end of the post, I agree that MRI in general should be on the service of health rather than law!