fMRI v. the Frye & Daubert Standards of Evidence: Re-searching for the Truth

Although jurors are tasked as fact-finders in deciding criminal cases, the human brain is not inherently proficient in discerning truth-telling behavior. A 1991 study, for instance, revealed that Secret Service personnel were only able to detect liars about sixty-four percent of the time. Laypersons, who primarily comprise juries, fared worse, with accuracy rates no better than chance. [1] Technology has attempted to fill this void by developing lie-detecting tests. However, from Lombroso’s 1895 pulse and blood pressure readings to Larson’s 1921 polygraph test, all existing pieces of technology have faced reliability concerns. [2] Specifically, the scientific field of polygraph research likely will never progress far enough, as the “inherent ambiguity of the physiological measures” stifles any hope for improvement through further experimentation. [3]

To address these criticisms, functional magnetic resonance imaging (fMRI) has taken center. This process measures changes in blood-oxygen-level-dependent activity within specific regions of the brain at a given time associated with cognitive activities like deception. [4] fMRI deception studies reveal “deceit patterns,” or neural areas that regularly activate when formulating a lie. [5] Purportedly, fMRI is more reliable than the polygraph, since it assesses central nervous system activity implicated in deception rather than responses of the peripheral nervous system like blood pressure and heart rate, which are less definitive indicators. [6] However, before jurors gaze upon colorful fMRI scans to resolve criminal cases, there lie two major legal hurdles to fMRI’s admissibility in court: the Frye and Daubert standards. Given current data, fMRI-based lie detection fails to meet these evidentiary standards, unless research in this field progresses significantly.

The first doctrine, primarily applied by state courts, is the Frye standard, which was set forth in the 1923 appeals court case Frye v. United States. In this case, appellant James Alphonzo Frye, convicted of second-degree murder, claimed that the lower court had erroneously dismissed expert testimony pertaining to systolic blood pressure deception test results in deciding his conviction. [7] Rejecting Frye’s arguments, the Court of Appeals of the District of Columbia reaffirmed the lower court’s decision, concluding that for the results of any scientific method to be relied upon as evidence in court, the method must first have “general acceptance” within its respective scientific field. Since the systolic blood pressure deception test had not yet gained such standing, evidence from such test results was appropriately rejected. [8] 

A similar “general acceptance” of fMRI-based lie detection is severely lacking within the scientific community. Few studies correlating deception and localized brain activity have been replicated by other laboratories. Moreover, the determined ‘deceit patterns’ fMRI studies look for to identify a lie vary across research, making it difficult to garner agreement. [9] Thus, while extensive scientific literature suggests that neuroscientists accept fMRI scans as evidence of brain damage, strokes, or lesions, fMRI has not been thoroughly researched for depicting past mental states like intent to commit a crime or bias. [10] This analysis is consistent with the 2010 ruling in Wilson v. Corestaff Servs. L.P., wherein the plaintiff, Cynette Wilson, argued that the trial court erred in dismissing fMRI evidence to bolster the credibility of a key witness for her case. The Kings County Supreme Court affirmed the lower court's decision, ruling that a review of the scientific studies demonstrated the “lack of acceptance of the fMRI test” for determining credibility or past mental states, thus justifying the dismissal of fMRI evidence on the credibility of Wilson’s witness. [11] 

The second doctrine is the Daubert standard. Building on Frye, Daubert is consistently applied by federal courts, emerging from the 1993 Supreme Court case Daubert v. Merrell Dow Pharmaceuticals, Inc. In Daubert, petitioners argued that the district court erred in dismissing expert analysis concluding that the prescription drug Bendectin was a risk factor for human birth defects. [12] In turn, the Supreme Court challenged the lower court’s application of Frye, and developed a new two-pronged test for admissibility: the scientific technique must be both relevant and reliable. On the question of relevance, expert testimony must assist jurors “to understand the evidence” or “determine a fact in issue.” Reliability is based on scientific validity. Judges must consider testability, methodical peer review, the known or potential error rate, if standards or controls govern application of the technique, and if general acceptance by the relevant scientific community exists (i.e. the Frye standard). [13] By providing a stricter analysis of how judges should evaluate the legitimacy of scientific techniques—such as the existence of peer-reviewed articles, standards of control, an n-value, etc.—the Daubert standard standardizes the definition of “general acceptance.” Its specificity provides an element of predictability in similar cases, which is critical to generating court precedent on the validity of evidence from scientific methods such as fMRI.

In failing to satisfy Frye’s “general acceptance” standard, fMRI-based lie detection automatically fails the fifth Daubert plank. Further analysis indicates that this method also fails the first four Daubert planks. First, as to whether fMRI-based lie detection evidence can be tested in some objective sense, the breadth of research available is overwhelmingly biased. The participants tested in the majority of fMRI lie detection studies are not representative of the general population: they are predominantly healthy, right-handed, and often recruited from within academia, a field that largely lacks gender and racial diversity. [14] Second, a lack of efficient study replication hinders the peer review process. If a participant moves slightly or engages in any mental process other than the formulation of a lie during a trial, fMRI scan data may be easily obstructed, making it difficult to replicate results. [15] Third, a true error rate does not exist for fMRI lie detection, since available studies utilize small sample sizes and none involve real-world lying. As to the fourth plank, there is no standard process for developing the neurological deception activation map, meaning that there are no centralized standards governing how fMRI lie detection tests function. [16] Indeed, a failure to satisfy the Daubert planks was seen in United States v. Semrau (2010), where defendant Dr. Semrau underwent three fMRI scans to prove his truthfulness in denying healthcare fraud. The district court ruled the fMRI studies failed under Daubert’s third plank, stating that “[t]here are no known error rates…outside the laboratory setting.” [17] Moreover, they observed that the fMRI studies also lacked uniform testing standards, as they incorporated different “deception-generating paradigms,” meaning that participants were asked to lie about different mock crimes in different ways, varying based on the study. [18]

Despite these grim findings, there may be a path to fMRI evidence’s admissibility if certain conditions are met by the existing body of research. Specifically, researchers should prioritize meeting Daubert standards for two reasons. First, although state courts generally exercise discretion as to applying Daubert or Frye, federal courts only deliberate under the more substantial Daubert standards. [19] Second, by meeting Daubert standards, fMRI evidence would implicitly satisfy the Frye standard through the fifth plank of Daubert, allowing for general acceptance of fMRI lie detection tests as evidence in American courts. 

The fifth Daubert plank—the Frye standard itself—can be met in two ways. First, the technology can undergo more rigorous peer review until reliability concerns wane out. Such was the case in Pettus v. United States (2012). In Pettus, the appellant argued that the trial court had erroneously admitted expert FBI forensics testimony showing the appellant’s handwriting matched handwriting left on a murder victim. The Court of Appeals for the Second Circuit deemed the handwriting match test admissible under Frye (and hence, Daubert’s fifth plank), citing precedent from the 2005 D.C. Circuit Court decision United States v. Jenkins: “[s]cientists significant either in number or experience [must] public[ly] oppose a new technique...as unreliable” before it fails Frye, which was not the case with the research on this handwriting test. [20] If fMRI lie detection technology improves by achieving an extensively peer-reviewed high accuracy rate, it can reach a level where a significant number of scientists no longer deem the technique unreliable—providing a path for “general acceptance.” 

Second, courts can lower the threshold of what constitutes “general acceptance” of a scientific technique, noting that such techniques are almost always in a state of constant development. In United States v. Donney Love, Sr. (2011), the Court of Appeals for the Ninth Circuit determined that fingerprint analysis passed Frye since the report criticizing this method was merely a “call for better documentation, more standards, and more research.” [21] From this precedent, it could be argued that current concerns about fMRI are not sufficient to negate fMRI-based evidence’s credibility, since these are merely vague calls for improvement. Such critiques are infinitely regressive, since scientists can always call for improvements or more peer-reviewed articles, no matter how reliable the technology becomes. If such calls were sufficient to deem a scientific method not generally accepted, then—as the court found in Love—no method would ever pass Frye, let alone Daubert. Thus, once fMRI meets the highly specific research burdens outlined by Daubert, it will have established a sound foundational body of research, and calls to build on pre-existing literature could be rendered inconsequential. Tailoring future research to address current gaps in fMRI data can lower the threshold of Frye from achieving blanket acceptance from the scientific community to obligating opposing parties to present targeted critiques of the fMRI lie detection method.  

The Love case is also insightful because the manner in which fingerprint analysis methods met Daubert’s other four standards in Love reveals how fMRI-based deception analysis methods can meet these in the future as well. To achieve the remaining four Daubert standards, fMRI research must show the following: that fMRI-based lie detection can be tested using a universal paradigm, that research addresses foundational questions, that the error rate is at or below 0.1 percent, and that standards for comparison during deception analysis are in effect. 

In terms of the first plank, in Love, ridge formations—i.e. indentations on the fingertip surface that serve to identify persons—could be clearly and consistently defined across investigators. This ensured that the results of fingerprint analysis could be conclusively verified by multiple investigators, which the Love court used to conclude that fingerprint analysis satisfied the first Daubert plank. [22] Currently, however, fMRI results cannot be conclusively verified in this manner, since blood-oxygen levels are estimates subject to each neuroradiologist’s discretion. However, utilizing machines of higher magnetic field resonance would generate more localized fMRI data, and consequently, aid in determining incontrovertible deceit patterns, which are consistent no matter the researcher. [23] 

As per the second plank, although the peer-reviewed literature presented in Love was limited in quantity, it addressed “theoretical/foundational questions,” causing the court to uphold fingerprint analysis methodology. [24] For fMRI research, accomplishing this could involve new publications discussing how machines distinguish, isolate, and attribute brain activity to the formulation of a lie as opposed to mental countermeasures. Moreover, research should demonstrate that laboratory deception results are consistent with lying to escape real-world liability, by prompting participants to lie about personal experiences—as witnesses might in criminal proceedings—rather than random events. [25] 

Regarding the third plank, there was no evidence that could be found in Love showing that fingerprint analysis misidentified persons any more than 0.1 percent of the time. To meet this standard, fMRI-based lie detection studies should incorporate larger sample sizes to likewise determine a statistically significant error rate. 

Finally, as to the fourth plank, although there were no codified standards of control for the fingerprint analysis method in Love, individual labs followed strict procedural standards, such as the requirement to assess every ridge of a fingerprint. [26] Similarly, fMRI deception research should include a consistent account of the steps taken during a lie detection test, specifying how neuroradiologists formulate a baseline neural image and providing a standard deception-generating paradigm. 

Presently, fMRI-based lie detection evidence is too unreliable and, at times, too conclusory, to be able to meet Frye and Daubert standards of evidence. Considering that the looming criticism of fMRI-based lie detection is scientific skepticism over its accuracy, research should build confidence in the validity of this method. Since trial court outcomes heavily depend on a juror’s ability to assess witness credibility, discerning truthfulness always has been, and will likely continue to be, a pressing concern towards maintaining the greatest level of integrity and trust in the courts. fMRI science can—and should—aid in that process. 

Edited by Samantha Velasquez

Sources:

[1] Paul Ekman and Maureen O'Sullivan, “Who Can Catch a Liar?,” 46 American Psychologist 9, 916 (1991).

[2] Jerome H. Skolnick, “Scientific Theory and Scientific Evidence: An Analysis of Lie-Detection,” 70 The Yale Law Journal 5, 696 (1961). 

[3] Committee to Review the Scientific Evidence on the Polygraph, The Polygraph and Lie Detection 213 (Washington DC: The National Academies Press 2003).

[4] Elizabeth A. Disbrow, Daniel A. Slutsky, Timothy P. L. Roberts, and Leah A. Krubitzer,  "Functional MRI at 1.5 Tesla: A Comparison of the Blood Oxygenation Level-Dependent Signal and Electrophysiology," 97 Proceedings of the National Academy of Sciences of the United 17, 9718 (2000). 

[5] Feroze B. Mohamed, Scott H. Faro, Nathan J. Gordon, Steven M. Platek, Harris Ahmad, and Michael J. Williams, “Brain Mapping of Deception and Truth Telling about an Ecologically Valid Situation: Functional MR Imaging and Polygraph Investigation—Initial Experience,” 238 Radiology 2, 680 (2006). 

[6] Daniel D. Langleben, James W. Loughead, Warren B. Bilker, Kosha Ruparel, Anna Rose Childress, Samantha I. Busch, and Ruben C. Gur, “Telling Truth from Lie in Individual Subjects with Fast Event-Related FMRI,” 26 Human Brain Mapping 4, 262 (2005). 

[7] Frye v. United States, 293 F. 1013, 2-3 (D.C. Cir. 1923). 

[8] Id 

[9] Jesper Ryberg, "When Should Neuroimaging Be Applied in the Criminal Court? On Ideal Comparison and the Shortcomings of Retributivism," 18 The Journal of Ethics 2, 82 (2014). 

[10] Teneille Brown and Emily Murphy, "Through a Scanner Darkly: Functional Neuroimaging as Evidence of a Criminal Defendant's Past Mental States," 62 Stanford Law Review 4, 1178 (2010). 

[11] Daniel D. Langleben and Jane Campbell Moriarty, “Using Brain Imaging for Lie Detection: Where Science, Law, and Policy Collide,” 19 Psychology, Public Policy, and Law 2, 225 (2013). 

[12] Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 585-586 (U.S. Supreme Court 1993). 

[13] William A. Woodruff, “Evidence of Lies and Rules of Evidence: The Admissibility of FMRI-Based Expert Opinion of Witness Truthfulness,” 16 North Carolina Journal of Law & Technology 1, 157 (2014). 

[14] Sean A. Spence, Catherine J. Kaylor-Hughes, Martin L. Brook, Sudheer T. Lankappa, and Iain D. Wilkinson, “‘Munchausen's Syndrome by Proxy’ or a ‘Miscarriage of Justice’? An Initial Application of Functional Neuroimaging to the Question of Guilt versus Innocence,” 23 European Psychiatry 4, 310 (2008). 

[15] Jesper Ryberg, "When Should Neuroimaging Be Applied in the Criminal Court? On Ideal Comparison and the Shortcomings of Retributivism," 83. 

[16] Teneille Brown and Emily Murphy, "Through a Scanner Darkly: Functional Neuroimaging as Evidence of a Criminal Defendant's Past Mental States," 1184. 

[17] Langleben and Moriarty, “Using Brain Imaging for Lie Detection: Where Science, Law, and Policy Collide,” 225. 

[18] Id 

[19] Legal Information Institute, Daubert Standard, Cornell Law School (2021), online at https://www.law.cornell.edu/wex/daubert_standard (visited August 11, 2021). 

[20] Robert E. Pettus v. United States, No. 08-CF-1361, 7 (D.C. 2012).

[21] United States v. Donny Love, Sr., NO. 10cr2418-MMM, 13 (S.D. Cal 2011). 

[22] Id at 5.

[23] Disbrow, Slutsky, Roberts, and Krubitzer, "Functional MRI at 1.5 Tesla: A Comparison of the Blood Oxygenation Level-Dependent Signal and Electrophysiology," 9722.

[24] United States v. Donny Love, Sr., NO. 10cr2418-MMM, 7 (S.D. Cal 2011). 

[25] Anthony D. Wagner, Richard J. Bonnie, BJ Casey, Andre Davis, and David Faigman, “fMRI and Lie Detection,” 17 Macarthur Foundation Research Network on Law & Neuroscience 10, 4 (2016). 

[26] United States v. Donny Love, Sr., NO. 10cr2418-MMM, 10-11 (S.D. Cal 2011).