Yesterday, Magistrate Judge Tu M. Pham in the Western District of Tennessee recommended that testimony about an fMRI lie detection test should not be admitted as evidence. Judge Pham reasoned that the fMRI lie detection test conducted by CEPHOS lacks sufficient scientific reliability (under Federal Rule of Evidence 702) and is more prejudicial than it is probative (under Federal Rule of Evidence 403). Judge Pham’s report and recommendation (available here) will be considered by the trial court judge, Judge Jon McCalla, who will make a final decision on whether these fMRI lie detection test results are admissible as evidence in this trial. All quotations in this post are taken from Judge Tu M. Pham’s report.
Background of US v. Semrau
The defendant in this case, Dr. Lorne Semrau, is a psychologist and the owner, president, and CEO of two corporations that contract with psychiatrists who provide mental health services to patients in nursing homes, including monthly medication management. Dr. Semrau “allegedly engaged in a scheme to defraud Medicare, Medicaid, and other health care benefit programs by submitting false and fraudulent claims for payment” for these services. According to the prosecution, Dr. Semrau instructed his billing personnel both to use improper CPT codes, codes which are used by medical providers to describe services, that paid a higher rate of reimbursement and to file for separate reimbursement for a test that should have been included with the monthly services. By manipulating the CPT codes, Dr. Semrau’s companies were allegedly overpaid $3,000,000 in fraudulent reimbursement claims. The U.S. Attorney’s Office charged Semrau with 60 counts of health care fraud, 12 counts of money laundering, and one count seeking forfeiture of property.
Dr. Semrau pleaded not guilty. To prove he was telling the truth when he denied the charges, Dr. Semrau sought out the services of CEPHOS, a company that performs fMRI lie detection tests. While in the fMRI scanner, Dr. Semrau answered questions about whether he intended to use the wrong billing codes and whether he knew better.
CEPHOS planned to scan Dr. Semrau twice. During each scan he was told to lie when answering the questions for half of that scan and then to tell the truth during the other half of the scan. CEPHOS compared the activity in three brain regions when he claimed to be lying with the activity in the same regions when he claimed he was telling the truth. According to CEPHOS, lying requires greater activity in these three brain regions. Therefore, if there was less activity during the half of the scan when Semrau denied the accusations, then he was telling the truth about his innocence; if there was greater activity when Semrau denied the accusations, then his denials were lies.
The fMRI scans were conducted on December 30, 2009, one in the morning and one in the afternoon, and were analyzed five days later by Dr. Steven Laken, the CEO of CEPHOS. After Semrau’s first scan, Laken concluded that Semrau was telling the truth about his innocence. But Laken found the second scan, done later the same day, indicated Semrau was lying. CEPHOS decided to conduct a third scan; at the hearing, Laken said this was appropriate because Semrau was tired during the second scan, which can lower the accuracy of the test. The third scan, conducted on January 12, 2010 and analyzed by Laken on January 18, indicated Semrau was telling the truth.
The defense attempted to submit the results from CEPHOS as evidence that Dr. Semrau was telling the truth about being innocent of the alleged fraud. The prosecution moved to exclude the evidence. On May 13-14, the court heard argument on the prosecution’s motion. Although pretrial motions in federal court are normally heard by the judge who will conduct the trial, due to a scheduling conflict, Judge McCalla was unavailable for the hearing. U.S. Magistrate Judge Tu M. Pham presided instead. As a magistrate judge, he was limited to making a report and recommendation to the trial judge, who will make the ultimate decision on the motion.
The hearing explored the reliability of CEPHOS’s fMRI lie detection tests to determine whether they could be a valid basis for expert testimony. Arguing for admission, Dr. Steven Laken, CEO of Cephos, testified on behalf of the defense. Arguing against admission, Drs. Marc Raichle and Peter Imrey testified for the prosecution. Their testimony and the attorneys’ arguments lasted into the evening of Thursday, May 13 and most of the following morning. Seventeen days later, on May 31, 2010, Judge Pham issued a report recommending that the fMRI lie detection results should not be admitted. This report is summarized below.
Judge Pham’s Report: Motion to Exclude Under 702
In federal court, in order for expert testimony to be considered sufficiently reliable (under rule 702), the underlying reasoning or methodology must be scientifically valid. Federal courts apply the Daubert standard to determine whether the underlying method is valid, which requires the judge to consider a non-exclusive list of factors including: (1) whether the theory or technique can be tested and has been tested; (2) whether the theory or technique has been subjected to peer review and publication; (3) the known or potential rate of error of the method used and the existence and maintenance of standards controlling the technique’s operation; and (4) whether the theory or method has been generally accepted by the scientific community.
First, the court made the preliminary decision of whether Dr. Laken was qualified as an expert to offer an opinion on fMRI-based lie detection in this case. The court determined he was. The court then examined whether fMRI-based lie detection satisfied the requirements of Rule 702 under the Daubert standard, such that Dr. Laken’s expert testimony would be reliable. The court determined that it did not for the reasons explained below.
Judge Pham concluded that the CEPHOS fMRI lie detection methodology satisfied the first two criteria: it is testable, and has been the subject of peer reviewed publications. In footnote 16, Judge Pham acknowledges that many of the peer reviewed publications submitted by the defense are of limited value because they do not attempt to differentiate between truthfulness and lies within an individual, citing Hank Greely and Judy Illes’s review of the published research from 2001-2008. But Judge Pham was satisfied that “since 2008, several additional studies have been published on this subject.”
The report goes on to explain that fMRI lie detection does not satisfy the last two criteria listed above: known error rate and standards for operation, and general acceptance.
The report’s assessment of whether error rates are known focuses on the lack of proven validity in real life settings. Although error rates in laboratory settings are impressive, the Judge Pham concludes that “the error rate of real life fMRI-based lie detection is unknown.” He supports this assertion by quoting CEPHOS CEO Dr. Laken’s own co-authored publications. Quoting the article Functional MRI Detection of Deception After Committing a Mock Sabotage Crime, co-authored by Dr. Laken, Judge Pham points out that the laboratory conditions in publised research do
“not equal the level of jeopardy that exists in real-world testing…In addition, this study only involved healthy adults who were not taking any medications. Thus, whether fMRI deception testing would work is unknown for participants who are taking a medication, who have a significant psychiatric or medical condition, or who are outside the 18-50 year age range.”
The judge noted that Dr. Semrau is 63 years old and hence fell outside of the tested range. Judge Pham also quoted another article by the same CEPHOS-affiliated research group, which further explains why their technique is not ready to be applied outside of a laboratory setting,
“Furthermore, different types of lies may produce different brain patterns. For instance, differences have been reported in false confessions versus false denials,… Finally, and importantly, deceptive behavior is controlled in the laboratory setting. In all studies to date, the research team directs participants to deceive about a certain condition or choice of conditions.”
Standards for Operation
The report explains that because fMRI lie detection is so new, there are not yet standards controlling how it is to be conducted. The judge found “without such standards, a court cannot adequately evaluate the reliability of a particular lie detection examination.” Furthermore, to the extent that Dr. Laken’s testimony about CEPHOS’ protocol represents a set of standards, CEPHOS violated those standards when it re-scanned Dr. Semrau after the second scan yielded unfavorable results.
“His decision to conduct a third test begs the question whether a fourth scan would have revealed Dr. Semrau to be deceptive again. The absence of real-life error rates, lack of controlling standards in the industry for real-life exams, and Dr. Laken’s apparent deviation from his own protocols are negative factors in the analysis of whether fMRI-based lie detection is scientifically valid.”
Judge Pham concluded that fMRI-based lie detection has not yet been accepted by the scientific community, after citing several publications. In particular, Judge Pham again quotes published research co-authored by CEPHOS CEO Steven Laken. In the 2005 article, Detecting Deception Using Functional Magnetic Resonance Imaging, the authors, several of whom are affiliated with CEPHOS, explain that experts in the field are of the opinion that fMRI “is currently not ready to be used in real-world lie detection.”
Judge Pham’s Report: Motion to Exclude Under 403
The court also ruled to exclude the fMRI lie detection evidence under Rule 403. Under Rule 403, if the risk of unfair prejudice substantially outweighs the probative value of the evidence, the evidence is inadmissible.
The report gives two reasons for why this evidence is especially prejudicial to the jury. First, the test was conducted “without the government’s knowledge and without an opportunity for the government to formulate, submit, or approve the questions asked of Dr. Semrau during the examination. Dr. Semrau risked nothing in having the testing performed, and Dr. Laken himself testified that had the results not been favorable to Dr. Semrau, they would have never been released.” Secondly, citing the plurality in United States v. Scheffer, admitting lie detection evidence to bolster Semrau’s credibility on issues that are central to the case would diminish “the jury’s role in making credibility determinations” and collateral litigation over the lie detection evidence that “threatens to distract the jury from its central function of determining guilt or innocence.” This second reason was also the basis for excluding fMRI lie detection in a recent case in New York, Wilson v. Corestaff Systems (opinion available here).
Furthermore, the probative value of these results is reduced because CEPHOS cannot make any conclusions about whether Semrau told the truth in response to any specific question. The most CEPHOS claims to do is determine whether Semrau was lying during an entire block of questions. “The court fails to see how his testimony can assist the jury in deciding whether Dr. Semrau’s testimony is credible. Therefore, the danger of unfair prejudice associated with admitting Dr. Laken’s fMRI-based lie detection opinions substantially outweighs any probative value attributable to them.”
Judge Pham’s recommendation to exclude the fMRI-based lie detection under rules for lack of reliability (under rule 702) and for risk of unfair prejudice (under 403) will be considered by the trial court judge. The trial itself started today. We will keep you posted as the trial proceeds.
The Center for Law and Biosciences director Hank Greely, former fellows Emily Murphy and Teneille Brown, and I (fellow Kelly Lowenberg) provided information to prosecutor Assistant US Attorney Stuart Canale as he prepared for the Daubert hearing. Of note, both a paper by Hank Greely and the Stanford Lawyer magazine article about the Center for Law and Bioscience’s work were cited by Judge Pham’s report, as was a law review article written by Emily Murphy and Teneille Brown. For more information about what this case demonstrated about the current state of forensic fMRI lie detection, watch this space for my upcoming paper on the topic.
– Kelly Lowenberg