The truth, the whole truth, and nothing but the truth is a lofty goal. But, as Pontius Pilate said to Jesus, "What is truth?" Researchers have long sought a scientifically sound technology that can tell, definitively, whether a person is lying. Now, functional MRI, which assesses brain function in a quick, noninvasive way with a high degree of both spatial and temporal accuracy, is being used commercially for deception detection. Two companies are marketing fMRI as lie detector to the public, starting at $4000 a session and boasting accuracy rates of more than 90%. Using images based on blood oxygen- level dependent scans, Los Angelesbased No Lie MRI and Cephos of Tyngsboro, MA, are offering fMRI as "truth verification." No Lie MRI founder Joel Huizenga charges $5000 for a 45-minute session using 3T MRI in two Southern California locations. Huizenga has a degree in molecular biology and has worked for Smith Kline & French Laboratories. His 2003 patent application, based on research by the University of Pennsylvania's Dr. Daniel D. Langleben, an associate professor of psychiatry at the Veteran's Administration Medical Center, is pending. Huizenga calls Langleben "independent" but acknowledges that the company has paid for his research. He estimates that the eventual market demand for fMRI lie detection scans will reach $3.6 billion. His clients, mostly men, usually are highly motivated. "It has to do with sex, power, and money, in that order," Huizenga said. Many of the men he sees have been accused of child molestation. Some are spouses trying to prove their fidelity. "It's a massive issue among people," he said. Women using his services are most often attempting to prove they've been faithful. Other clients have been involved in workers' compensation cases. Huizenga's customers need a lot of hand-holding, he said. "If you get accused of child molestation, there's nothing more important in your life," he said. He declined to say how many people his company has tested in the two years it has offered the service but noted that subjects must be compliant. "Our customers are individuals who want to prove they're telling the truth," Huizenga said. "Liars aren't going to be customers."
Cephos founder Steve Laken, Ph.D., says more than 300 people have been tested in the company's 1.5T scanner in Framingham, MA, since it began offering fMRI in 2004. His clients, who pay $4000 for scans, have included men involved in criminal cases like molestation and violations of restraining orders. Others were involved in employment cases stemming from racial discrimination complaints. Laken's pending patent application is based on research by Dr. Frank Kozel done at the Medical University of South Carolina. Kozel is now an assistant professor of psychiatry at the University of Texas Southwestern Medical Center in Dallas. Cephos paid for some of his research. While Huizenga and Laken both stress that fMRI is more accurate in deception detection than is polygraph testing, with its paltry 60% accuracy rate, neither technique is admissible in court. ADVANTAGES OVER OTHER TESTS Functional MRI has the theoretical advantage over the polygraph of measuring central rather than peripheral nervous system activity and the advantage of spatial resolution over electroencephalography, which measures electrical currents. Polygraphs measure several physiological features that are associated with nervousness or stress, such as systolic blood pressure, heart and breathing rate, and skin sweatiness. Courts in the U.S. have never generally considered the testing sufficiently reliable for its results to be admitted into evidence. In the wake of the Wen Ho Lee spy case at Los Alamos National Laboratory, in which polygraph examinations played a role, the National Research Council was asked to report on the value of polygraph evidence. The agency concluded that polygraph testing was not sufficiently valid to be used regularly in national security screening. Some judges do allow several kinds of brain scans, such as CT, PET, and SPECT, primarily to show structural damage as a mitigating factor for the defense in a criminal trial or to establish damages in a civil suit. The scans are most frequently found in sentencing hearings, where evidentiary rules are the least stringent. As forensic neuroscience consultant Daniel Martell told The New York Times in 2007, they are almost always employed by the defense in death penalty cases. While fMR scans have never been admitted, the U.S. Supreme Court did admit fMRI evidence in 2005 in Roper v Simmons, a case involving a minor on death row, to help establish that an adolescent brain works differently from an adult one. While Justice Anthony Kennedy didn't explicitly cite fMRI in his majority opinion against executing people under 18, many experts think it was an influencing factor. The exoneration of innocent criminal defendants or inmates would be among the most valuable goals of reliable lie detection technology. Both the biggest hurdle and the brightest future selling point, however, for using fMRI in deception detection lie in establishing its scientific validity. For most scientists, including radiology experts, the jury is still out. Henry T. "Hank" Greely, J.D., Deane F. and Kate Edelman Johnson Professor of Law at Stanford University, directs both the law school's Center for Law and the Biosciences and the neuroethics program at the Stanford Center for Biomedical Ethics. He says the idea has great potential but cautions that it has a long way to go to establish reliability.
Although several studies have established brain activity with deception, none of the results have been replicated by an independent group, and many findings point to different areas of the brain as being responsive, Greely said. Shortage of research is not the problem: At least a dozen peerreviewed articles have been published regarding fMRI-based lie detection, three by Langleben, whose technology has been licensed to No Lie MRI, and another three by Kozel, whose research is being developed by Cephos. In Langleben's study, described in 2002 in NeuroImage, 18 Penn students were given a playing card, then told to lie or answer honestly to questions about what card they held. In the averaged results, researchers found that two brain regions showed statistically significant increased activation when the subjects lied. The first ran from the left anterior cingulate cortex to the medial aspect of the right superior frontal gyrus. The second region was a 91-voxel cluster, Ushaped along the craniocaudal axis, that extended from the border of the prefrontal cortex to the dorsal premotor cortex. This also involved the anterior parietal cortex from the central sulcus to the lower bank of the interparietal sulcus. Part of the first region is known to be activated when someone stops him- or herself from responding in the easiest way, the authors said. They speculated that the second region may be involved in providing additional help in overcoming the first response to answer truthfully. In a second similar test session, Langleben's results showed activity in some, but not most, of the areas in the earlier study, and the researchers themselves noted that the results contradicted some of the assumptions of the earlier study. Based on the studies, Langleben's researchers created one model showing fMRI to be 78% accurate at showing when a subject is lying. Using the same brain scans, they subsequently performed another analysis using a high-dimensional nonlinear pattern classification method, in which they reported being able to show when subjects were lying about 90% of the time. Kozel did two studies, a pilot and a replication study. The pilot study, published in 2004 in the Journal of Neuropsychiatry and Clinical Neurosciences, used eight young men who were questioned about where a $50 bill was hidden. His researchers predicted activation in three regions: the orbitofrontal cortex, anterior cingulate, and amygdala. They found statistically increased activation when the subject was lying in the first two, as well as in three other regions (the superior temporal gyrus, cerebellum, and frontal gyrus). They did not, however, see increased activation in the amygdala. The researchers found variable patterns of activation when they looked at individual results: One of the eight subjects showed no regions of differential activation when lying, and the other seven showed diverse activation patterns. The researchers did find a correlation between the patterns of electrodermal activity and regions of brain activation. In the second similar test using 10 subjects, Kozel's researchers found increased activation in five areas when the subjects were lying and no areas of increased activation when subjects were telling the truth. Of 11 areas with significantly increased activation during lying, five were identified in the pilot study. The investigators again found quite varied patterns of activation among individual subjects. When they defined regions broadly, however, they found that seven of the 10 subjects had increased activation in the right prefrontal cortex.
In a third study, Kozel used 30 subjects who were instructed to "steal" either a ring or a watch from a drawer. They were then told they would receive an extra $50 if their lie was not detected. Results showed significant increased activation with lying in seven clusters of brain regions, including five the researchers had seen before. Analysis showed they were able to detect lying about 90% of the time. TIME WILL TELL Sean A. Spence, a professor of psychiatry at the University of Sheffield in the U.K., published the first fMRIbased lie detection paper in Brain Imaging: NeuroReport in 2001, based on his experiment with 30 people who were questioned about their activities that day (for example, did they make their beds). Analysis showed the subjects took longer (up to 12%) to lie than to tell the truth. Writing in the American Society of Law, Medicine & Ethics, Greely points out that the experiments showed activation in many different areas of the brain during deception but no consistency, except when the brain regions were very broadly defined (2007; 33[2,3]:377-431). The cortical areas activated in the tests included parahippocampal, anterior cingulate, left posterior cingulate, temporal and subcortical caudate, right precuneous, left cerebellum, insula, putamen, caudate, and thalamus; the anterior, ventromedial, and dorsolateral prefrontal regions; and regions of the temporal cortex. These regions are known to correlate with a wide range of cognitive behaviors, including memory, self-monitoring, conscious self-awareness, planning and executive function, and emotion. Since so many different brain areas showed activity during in the tests, however, the findings cast significant doubt on any particular method of lie detection, Greely said. He also pointed out that the artificial nature of the questioning involved in the experiments does not reflect real-world situations. "Instructing a research subject to lie about whether he's seeing the seven of spades is very different from questioning a criminal defendant who's saying, 'No, I never tried to buy cocaine from that officer,'" Greely said. Then there are human frailties like long-rehearsed lies that people half convince themselves are true, which would probably be hard to detect. In addition to being compliant, subjects must also remain motionless during testing; simple movements of the tongue or jaw will make fMRI unreadable. Greely was also concerned about subjects using countermeasures, both detectable and undetectable. Simply thinking about other things may activate other brain regions in ways that interfere with the lie-detection paradigm, he said. A subject could try doing mental arithmetic or memory tests, for example, while giving true answers, thus making true answers harder to distinguish from false ones. In fact, the people we most want to catch may well be the ones best trained-by criminal gangs, foreign intelligence agencies, or terrorist cells-in countermeasures, according to Greely. Noting that reliable fMRI lie detection requires on accurately evaluating some fairly sophisticated brain patterns, he describes himself as evenly split over whether it is possible at all and genuinely agnostic about the tests. "We find very little in the peerreviewed literature that even suggests that fMRI-based lie detection might be useful in real-world situations and absolutely nothing that proves that it is," Greely said. If and when fMRI lie detection tests are proven scientifically reliable, federal oversight will be necessary, requiring FDA-like regulation of all lie-detection drugs and devices, according to Greely (Am J Bioethics 2005;5:50-52).
"We should prohibit the sale, marketing, distribution, or use of lie detectors until they have been proven, by rigorous trials, to be safe and effective," he said. NO PINOCCHIO Functional MRI represents a quantum leap forward over any previous technology for imaging the brain, according to Joy Hirsch, Ph.D., director of Columbia University's Functional Magnetic Resonance Imaging Research Center and the Program for Imaging and Cognitive Sciences. But she says using fMRI for lie detection still poses significant legal and credibility problems. "Is there a specific pattern of brain activity that's associated with false representation?" Hirsch said. "There simply is no evidence of that." She points out that if an individual is lying, he or she would still have to cooperate to get a good functional scan. "It's a matter of interpretation of the brain images and of the individual's intent," Hirsch said. She notes that cognitive control can overcome the prepotent response to tell the truth. "Lying is a volitional intent to deceive, and it's complex," she said. The American College of Radiology has not taken a formal position on using fMRI for lie detection. "The potential is great, especially for exonerating falsely accused people, to be able to have something in addition to a polygraph or supplement it," said Dr. Peter Kalina, chair of the ACR Ethics Committee. He concurs that more work is necessary for fMRI to be accepted by the scientific community, the courts, and the public. "The technology should be embraced and studied, and we should keep searching and researching to determine the ultimate answer whether or not this will become the gold standard in truth detection," Kalina said. "There needs to be more testing to assure the public that there won't be false positives or false negatives." So what does Hirsch think about the potential value of using fMRI for lie detection? "It's not a Pinocchio detector," she said. "But if people are willing to spend $5000 for a brain scan, they can. And if they've got a tumor, fMRI will detect it, so they're better off."