MRI may be better than polygraph for revealing lies

Aug 9 2004

A new study involving people telling lies while undergoing functional magnetic resonance imaging (fMRI) provides early indications that understanding the brain basis of deception may lead to a better lie detection method than polygraph tests.

Researchers have found that for lying, compared with telling the truth, there is more activation in five brain regions. Results of the study will be presented at the 112th Annual Convention of the American Psychological Association (APA) in Honolulu and also published in the August issue of Behavioral Neuroscience, a journal published by the APA.

Despite widespread use, there is mounting evidence that polygraph tests are not an accurate lie detection tool. The National Science Foundation recently highlighted the need for new methods to identify deception after concluding in a report that the polygraph lacked evidence to support its use in detecting deception. Researchers say the problem with the polygraph is that it measures peripheral arousal, not deception itself.

Blood oxygen level-dependent (BOLD) fMRI, however, may be able to measure brain activity associated with deception. In a study designed to replicate their previous pilot study, Frank Andrew Kozel, M.D., and colleagues at the Medical University of South Carolina and the Ralph H. Johnson Veterans Affairs Medical Center in Charleston, South Carolina, studied 10 healthy adults who were told to tell the truth or to lie while being imaged in a magnetic resonance imaging scanner. Before the scanning began, the participants visited a room in which they were told to locate and leave in place fifty-dollars bills that were placed under two of six objects in the room. During the scanning procedure, the subjects viewed pictures of the six objects from the room just visited and had to indicate by pushing a “yes” or “no” button whether money was hidden below the object. Participants were instructed to give truthful and deceptive answers. Subjects were given an incentive of an additional fifty dollars if one of the researchers could not detect when they were lying.

Results ­ which were consistent with the pilot study ­ revealed significant activation of five brain regions during lying compared with truth telling. These areas included the right inferior frontal, right orbitofrontal, right middle frontal, left middle temporal and right anterior cingulated areas. Several of these regions have been identified in other neuroimaging studies as being important in cognitive tasks that could be related to deception. These regions could inhibit an overlearned response (the truth), especially in order to gain a reward, according to the researchers. More research is needed to determine what each of these regions does during lying, such as arousal or response inhibition.

While significant group effects were found, the results do not currently support the use of fMRI to detect deception in real world individual cases, according to the researchers. “The study was designed to identify brain regions associated with deceptive answers versus truthful answers, not to formally test the method as a means of lie detection,” Dr. Kozel cautions. “Subsequent work will be needed to determine whether this technology can be used to distinguish deceptive responses from truthful responses with individuals.”

Presentation: “Using Functional Magnetic Resonance Imaging (fMRI) to Detect Deception," Frank Andrew Kozel, M.D., Medical University of South Carolina; Session 5140, 8:00 - 8:50 AM, Wednesday, July 28, Hawaii Convention Center, Level 3 ­ Meeting Rooms, Meeting Room 316B.

Journal Article: “A Replication Study of the Neural Correlates of Deception,” Frank Andrew Kozel, Medical University of South Carolina and Ralph H. Johnson Veterans Affairs Medical Center, Charleston, South Carolina, Tamara M. Padgett, Medical University of South Carolina and Mark S. George, Medical University of South Carolina and Ralph H. Johnson Veterans Affairs Medical center, Charleston, South Carolina, Behavioral Neuroscience, Vol. 118, No. 4.

http://www.musc.edu