Libbey 1

Madeline Libbey
Dr. Russell Poldrack
Reading the Brain
1 March 2017
Brief Overview of First Draft Revisions:

1) Dr. Poldracks Comments:

a) On page 4, I reworded the quote from the Bizzi et al. so instead of a long quotation, it is a shorter

explanatory sentence.
b) I removed Cephos from my examples of commercialized fMRI after it was brought to my

attention they no longer provide that service

c) On page 5, beginning of page 6 I distinguished between the two types of motion I discuss in

those paragraphs: indistinguishable finger movements and head movements that can create

imaging artifacts.
d) In the Rule 702 section and US v Semrau section of the essay (pages 7 and 8), I added points and

analysis about the Daubert Standard to further discuss the judiciarys interaction with scientific

evidence in the past and how that might affect the future.
e) On page 9 I changed the wording ruin lives to something a little less dramatic, and hopefully

more accurate
f) The biggest change I made was on page 10 with the discussion of the regulation committee to

clarify just what policy I want to implement and how. I added a few sentences about the specifics

of the committee, what it might look like and what the regulations might look like.

2) Peer Review Comments

a) Grammatical/Wording Changes: For the majority of grammar and wording changes throughout

the paper, I made the fixes. Most of them had to do with clarity and specificity which I wouldnt

have noticed, so I really appreciated those. However, I intentionally kept the we throughout the

paper, because I feel it is powerful to my persuasive writing style.

b) Question/Answer Writing Format: I combed through the paragraphs looking for question/answer

style and varied the syntax for some. I did keep a few though, because, along with the we I feel

the questions add to the oratorical persuasive style of my paper.

 Libbey 2

c) Show dont tell: I removed all of the language similar to This paper will show etc. but I

usually kept the rest of those passages other than that specific phrase because I see them as more

prefaces/outlines for what is to come in the paper.

d) Cut down quotations: For each quote more than a sentence long, I paraphrased the quote using

brackets. I did keep the block quotes for Rule 403 and 702 because I think the logical

progression from the rule followed by my analysis is as clear as I can make it.
 Libbey 3

The Futile Quest for Truth: A Policy Proposal for the Regulation of fMRI Lie Detection

Its generally understood that a memoir is not a complete and accurate account of ones

life. Instead, it is the recollections of ones past, a record of memories and thoughts about

experiences preceding the now. Unlike an autobiography, not every fact has been checked for

objectivity. In fact, a memoir often relates in prose what cannot be fact-checked. But, sometimes

it does. Many a memoir has been found to contain complete falsehoods, even when they are

unbeknownst to the writer. An authors truthful recounting of their story, the way they remember

it, may in fact be entirely different from the real, objective, truth. How reliable are our personal

truths? And, maybe more importantly, our lies? While this phenomena of false remembrance in

memoirs may have some consequences, these pale in comparison to when this ambiguous truth

arises in court. This is the work that we have laid out for the triers of fact: discern the truth from

the lies, the false remembrances, and the noise. To investigate these truths, we gather evidence,

we cross-examine, we call upon experts, and then finally, we decide. Its a system that at its best,

balances due process, truth, and the rights of the accused.

However, today this balance is being tipped. With the introduction of neuroscience-based

lie detection into a court of law an fMRI becomes our trier of fact, and false remembrances and

scanning image noise can be mistaken for truths. The issues with fMRI-based lie detection

today are two-fold: First, we must address the preliminary question of efficacy; and only after we

do, can we address the secondary question of ethicality. To date, the scientific community is

divided on the causes and effects of lies, the neural markers of lies, and the scientific definition

of a lie itself. fMRI lie detection cannot be reliable until the nature of a lie (versus a false belief,

or omission) is defined, and cannot be replicable until the neural markers of deceit are known
 Libbey 4

and consistent. Therefore, as it is now, fMRI lie detection is not replicable, nor reliable. Some

critics argue that it never can be. There is much work to be done before this method will be

accepted in the scientific community as effective, and arguably even more work to be done

before it is accepted in the court system.

Not only is fMRI lie detection technologically flawed, un-replicated, and limited, it is

also fundamentally incompatible with the judiciary. Even if all of these technological obstacles

were surmounted the issue of introducing fMRI lie detection data as evidence into court poses

additional challenges to be resolved. The Federal Rules of Evidence, Fifth Amendment and right

to privacy, along with past judicial opinions, all suggest that fMRI lie detection would be

considered more prejudicial than probative in a court of law. To understand this we must delve

deeper into the limits and challenges of an effective fMRI and its implications on the rights of the

accused, due process, and the integrity of our courts. Ultimately this evidence reveals an urgent

need for legislation creating a regulatory system for pre-market approval of lie detection

techniques, which would play a crucial role in preserving the authenticity of the judiciary.

To evaluate the legitimacy of using fMRI lie detection, it is critical to first understand the

basic mechanisms behind the technology. Functional Magnetic Resonance Imaging has been

utilized in the scientific and medical community for more than a decade (Wagner et al.). In

analyzing this data we can isolate the changes in the brain correlated to specific presented stimuli

or the performance of a task. This is done by subtracting the noise brain activity from the

image, which is found by comparing a control image to the experimental image and subtracting

the difference. Considering the complexity of brain function and cooperation among systems, the

key to obtaining significant results is choosing an adequate control image. For example, if a

researcher is looking for what areas of the brain are triggered by reading aloud, the control might
 Libbey 5

be viewing words without speaking (Bizzi et al.). With this method it is possible to isolate and

correlate brain activity with behavior. fMRI-based lie detection is attempting to correlate brain

activity with a mental activity, which introduces additional challenges.

The preliminary issue is fMRI technology, as it is now, is simply not advanced enough to

detect lies in real world situations. The inability to identify an accurate control image, the

inability to mimic a real-world lie in a laboratory, the misinterpretation of data, and the

unchecked potential for countermeasures all contribute to the unreliability of this method. All of

these issues ultimately beg the question: how do we keep this unreliable technology out of the

courts? The answer is in regulatory legislation.

Establishing a sufficient control image is the most basic challenge that fMRI lie detection

faces. The primary nature of deceit within the brain is still mysterious to researchers.

Considering this, it is nearly impossible to find a control from which to make inference. For

instance, a control used in a laboratory setting may intrinsically leave out crucial convoluting

brain activity that would be present in a real world situation. Dr. Elizabeth Phelps, explains that

in her findings, the neural signatures of lying may be heavily intertwined with emotion and

imagery(Bizzi et al.). The imagery and emotion that would likely be present when a lie is

personally relevant and important might interfere with the use of fMRI signals to detect

familiarity(Bizzi et al.). In other words, in the laboratory researchers find an isolation of activity

when a subject tells what would be considered a lie. However, there are significant differences in

brain activity from a subject telling an instructed untruth, to a suspect or witness of a crime

answering highly emotional and anxiety inducing interrogations. Therefore, correctly identifying

a control for a real-world applicable situation would be as impossible as disentangling the

 Libbey 6

synapses and neurons in the parahippocampus (the emotion/memory region as hypothesized by

Phelps).

Provided that a lab setting differs drastically from real-world application, laboratory

studies are testing a different set of brain functions than those that would be ignited by a true lie.

This not only poses problems with identifying a control image, but also causes issue in the way a

stimulus is presented. In a meta-analysis by Martha J. Farah, she found that the types of

instructed lies utilized by fMRI lie detection studies are cognitively complex: the subject must

remember the stimuli, suppress the truth, decide to lie, and choose a response (Bizzi). Dr.

Phelpss research tells us that while it is quite likely that a real lie may follow a similar sequence

of processes, the involvement of emotion and personal attachment would alter the brain activity

drastically. Real-world versus laboratory applications aside, according to Bizzi et al., for any

process that is cognitively complex it is impossible to jump from observing brain activity to

inferring mental processes due to the sheer complexity and interaction between regions of the

brain. It is for this reason that fMRI is widely understood as a purely correlative tool. If we

cannot establish neither a baseline for a controlled brain image from which to present a stimulus,

nor an acceptable stimulus to present; it is obvious why fMRI lie detection has been met by so

much skepticism.

Obtaining a reliable control image and stimulus is just the first of many issues

surrounding the research into fMRI lie detection. Companies like No Lie MRI, an investigative

services that offers fMRI lie detection, claims their current accuracy is over 90% and is

estimated to be 99% once product development is complete (Greely). Dr. Nancy Kanwisher

explains in her essay, The Use of fMRI Lie Detection: What Has Been Shown and What Has

Not, that the statistics included on these sites come from an analysis of averaging group data as
 Libbey 7

opposed to analyzing individual subject data. On a fundamental level, group studies are not

useful for determining whether a particular subject is lying, rendering the rationale behind these

companies groundless(Wagner et al.). Yet even when we examine the research analyzing

individual subject data, the results, while scientifically probative, are inconclusive and unreliable

for real-world application. For example, in a study conducted by Kozel and colleagues, the

research question posed was not Can you tell if the subject is telling a lie or a truth? but

instead, it investigated the question Which of these responses are lies?(Kozel). In fact, most

studies of fMRI lie detection are investigating this secondary question instead of true lie

detection. It is misleading to equate these laboratory conditions and group data generalizations

with a real-world application, especially when they have the potential to be applied in a context

as weighty as criminal culpability.

Supposing we do take this context out of the lab and use it within a real world situation;

an additional technological concern would be the inability of fMRI to detect countermeasures. A

research subject getting paid to perform a fMRI study is much more likely to cooperate than a

criminal being tried for murder. And still, even when we have cooperative subjects trying their

best to help us and give us good data, we still throw out one of every five, maybe ten, subjects

because they move too much (Wagner et al.). Head movement that causes artifacts and ruins

scan data can be detected by a researched, but other movements, for example the movement of a

finger, are indistinguishable by the fMRI and cause interference with the tested task for cognitive

reasons. The accuracy rate of lie prediction drops nearly 70 percent when we introduce these

indistinguishable movements (Wagner et al.). Additionally, Dr. Phelps hypothesizes that studies

on the repetition of the Stroop Task indicate that the lying neural signatures could be suppressed

with practice and repetition. Furthermore, Dr. Kanwisher also conjectures that many of the same
 Libbey 8

hypothesized lie regions could be artificially activated by performing mental arithmetic.

Certainly, these findings are concerning when considering their possible implications within a

court of law.

It is clear that these basic technological obstacles impose serious credibility issues on

fMRI lie detection data. But even if we do apply these faulty techniques to real world situations

and present them as evidence in court, the introduction of fMRI lie detection into a trial as

evidence would most certainly fail. The judiciary imposes safeguards against such misleading

data as this, first and foremost with the Federal Rules of Evidence (Greely). Rule 702 is the

original rule of evidence concerning admissibility of expert witness testimony and scientific

methodology. Established in 1976 this rule was later expounded upon and specified with the

Daubert Standard set forth in the Daubert v Merrell Dow Pharmaceuticals Case. The most recent

stipulation of Rule 702 here reading:

If scientific, technical, or other specialized knowledge will assist the

trier of fact to understand the evidence or to determine a fact in issue, a
witness qualified as an expert by knowledge, skill, experience, training
or education may testify thereto in the form of an opinion for otherwise,
if (1) the testimony is based upon sufficient facts or data, (2) the
testimony is the product of reliable principles and methods, and (3) the
witness has applied the principles and methods reliably to the facts of
the case.

The Daubert Standard further specifies these requirements with the following: the method must

be testable and tested, peer reviewed, the rate of error must be known and controlled, and the

method must be generally accepted into the scientific community. If it were brought into court

today, lie detection evidence provided by fMRI would never pass these stipulations, nor should

it. To address the first requirement, the evidence given earlier in this brief explains the gross

insufficiency of data surrounding this technology. Nearly no two studies can agree on even the

basic biological mechanisms behind lying. If we are to understand sufficient to mean that
 Libbey 9

many data agree, fMRI would not come close. It has also been made clear that the methods of

obtaining lies and truths through fMRI are themselves unreliable until we understand much more

about how the processes of deceit within the brain work. Until then, not only are the methods

unreliable, but they also yield very unreliable results.

It is important to remember that for purposes of real-world application, based on what we

know (and the scientific community corroborates) we may never be able to disentangle the

cognitive processes to the extent that is demanded by the law. But even if the confounds present

suddenly disappeared, Rule 403 in the Federal Rules of Evidence would likely bar the data from

seeing its day in court. Rule 403 reads:

The court may exclude relevant evidence if its probative value is

substantially outweighed by a danger of one or more of the following: unfair
prejudice, confusing the issues, misleading the jury, undue delay, wasting
time, or needlessly presenting cumulative evidence.

Certainly, the prejudicial value of unreliable lie detection evidence would radically outweigh the

benefits. In fact, there arent any. Introducing this data in trial, even if it were accurate, would

surely incite prejudice and bias among a jury. Our system requires that if presented with any

incriminating evidence, the jury must still remain impartial, assume the innocence of the

defendant, and weigh all facts together for their verdict. This type of cumulative evidence would

mislead the jury, essentially replacing a trial of peers run by fact and circumstance with statistics

and weighing probabilities of error. Which, as found in People v. Collins, threatens the very

purpose of the jury itself. While assisting the trier of fact in the search of truth, [evidence] must

not cast a spell over him.(Greely).

This argument is corroborated by the decision in the case of US. v. Semrau, where the

court held to exclude fMRI lie detection evidence presented to bolster defendant testimony

credibility. The scans done on Dr. Semrau were kept out of court because the technology had
 Libbey 10

not been fully examined in real world settings [and] was not consistent with tests done in

research studies(Greely). This finding is consistent with what we would expect when analyzing

the data from previous studies: when applied to a real-world setting, this form of lie detection

will not provide predictable results. Additionally, it was found that this evidence violated Rule

403 of evidence, because the data were found to be essentially inconclusive, as well as the scans

themselves posing constitutional concerns.

All of this raises the issue: the law understands an effective scientific technology as

nearly always an effective judicial technique, but the evidence presented here would support an

argument against this assumption. If most researchers agree that fMRI lie detection is flawed and

for now should stay in the laboratory, and if four U.S. Supreme Court justices have preemptively

argued that the introduction of this type of evidence--reliable or not--would unduly infringe the

province of the jury, who is arguing for fMRI lie detection use in the judiciary? The obvious

answer is those who are profiting from it (Greely).

Companies like No Lie MRI that market fMRI lie detection services to attorneys have

motive to lobby for fMRI-based lie detection in the courts. The typical fMRI lie test, run just

once, would cost a customer over $5,000. That is about twice the amount it usually costs to

compensate an expert witness to testify on DNA findings, and five times the amount for a DNA

test. But of course, the marketed motive behind this technology isnt just the money, it is the

ability to make something as messy as truth and criminal justice into something as simple as a

yes or a no. In a court system where individuals are given advantage for what they look like and

where they come from, this idea is enticing. If we strip the courts of subjectivity, we strip them

of bias and prejudice. It isnt that simple. The bottom line is that unreliable fMRI lie detection is

a liability. False, ineffective, and uncorroborated science such as this being marketed as fact has
 Libbey 11

the potential to affect momentous change in our judiciary. While research is one sphere, the legal

sphere is quite another. It should be made clear that while we do have a precedent in place, our

constitution, which qualifies scientific methods in court when theyre reliable and effective, even

an effective scientific method can fail to be an effective technique of justice. The bottom line:

regulation is desperately needed.

As demonstrated by the evidence in this article, safeguards against misleading and

prejudicial science is clearly needed. The legislation proposed is not a radical measure by any

means. It is only imposing the same regulations we expect from medical research to

neuroimaging research. The FDA was established for many of the very same reasons that fMRI

is raising concern within the legal and scientific communities (Greely). Today, we require that

drugs be proven to be both safe and effective before they can legally leave the laboratory and

reach pharmacy shelves. This legislative change suggests a similar pre-market approval for

fMRI, with the establishment of a commission within the FDA dedicated to this issue and other

issues in the scientific sphere of neuroimaging. Though fMRI lie detection is currently too

unreliable to pose serious threats to the judicial system, the potential harms of this technology as

it stands now, and how it could evolve, are too great to wait. The framework of this regulation

commission is laid out as such: the subject of the regulation will include all neuro lie detection

techniques, including polygraph, EEG, fMRI etc. The committee actually doing the regulation

will accredit methods of lie detection following a similar structure to FDA drug approval, an

initial standardized set of rules dictating the regulations for trials, and research will be

established and revised as technology changes. This will certainly not be easy but with help from

committees from the Department of Defenses Academy for Credibility Assessment, MacArthur

Law and Neuroscience Foundation and others, it is possible to establish some sort of baseline
 Libbey 12

from which to begin assessment. A regulatory structure would entirely resolve the current issues

facing fMRI lie detection credibility without stifling attempts to improve upon them. The future

of fMRI lie detection is uncertain. Many argue that it will never reach a level of sufficient

reliability and replicability to be accepted into the scientific community, let alone the legal one.

Nevertheless, this regulation system will establish the precedent that until it improves, whether

that day comes or not, fMRI lie detection cannot breach the barriers of theory or research while

disguising itself as fact.

Humans like patterns. We like black and white, simplicity. We like to think we are simple,

but we are not. It truly would be revolutionary if we could discern truth from falsehood by

identifying voxels of electrical impulse in our hippocampus. Unfortunately, telling a lie is much

more complicated than igniting a lying region of the brain. It is intertwined with cognitive

complexity: emotion, imagery, suppression, control. Neuroscience can begin to understand this,

fMRI can scratch the surface of deceit, but it cannot detect lies. At any relevant level of

sophistication or reliability. The research conducted on fMRI-based lie detection has revealed

worlds of complexity and possibility within the brain, but it has not revealed a reliable method of

lie detection. Even if it did prove itself as a reliable and replicable method in the real world, it

still is not compatible with the integrity and function of our judiciary. The possibility of fMRI

passing the standards of evidence set out by our constitution is close to none. Nevertheless, if the

day comes when fMRI-based lie detection passes the regulations in our rules of evidence, the

rules of evidence must adapt to better preserve the power of the jury, and the rights and privacy

of the accused. For now, regulation will do. The passing of this legislation will keep fMRI in the

laboratory, where it--for now--belongs, and in doing so preserve the authenticity and purpose of

our legal system.

 Libbey 13

Works Cited

Anthony Wagner, Richard J. Bonnie, BJ Casey, Andre Davis, David L. Faigman, Morris B.

Hoffman, Owen D. Jones, Read Montague, Stephen J. Morse, Marcus E. Raichle,

Jennifer A. Richeson, Elizabeth S. Scott, Laurence Steinberg, Kim Taylor-Thompson,

Gideon Yaffe, fMRI and Lie Detection: A Knowledge Brief of the MacArthur Foundation

Research Network on Law and Neuroscience (2016).

Emilio Bizzi, Steven E. Hyman, Marcus E. Raichle, Nancy Kanwisher, Elizabeth A. Phelps,

Stephen J. Morse, Walter Sinnott-Armstrong, Jed S. Rakoff, and Henry T. Greely .

(2009). Using Imaging to Identify Deceit: Scientific and Ethical Questions. American

Academy of Arts and Sciences.

Farah, M. J., Hutchinson, J. B., Phelps, E. A., & Wagner, A. D. (2014) Functional MRI-based lie

detection: Scientific and societal challenges. Nature Reviews Neuroscience, 15, 123-31.

Ganis, G., Rosenfeld, J. P., Meixner, J., Kievit, R. A., Schendan, H. E. (2011) Lying in the

scanner: Covert countermeasures disrupt deception detection by functional magnetic

resonance imaging. Neuroimage, 55, 31219.

Greely, H. (2009). Law and the Revolution in Neuroscience: An Early Look at the Field.

University of Akron School of Laws Neuroscience, Law & Government

Symposium.

J. Peter Rosenfeld, Bradley Cantwell, Victoria Tepe Nasman, Valerie Wojdac, Suzana Ivanov &

Lisa Mazzeri (1988) A Modified, Event-Related Potential-Based Guilty Knowledge Test,

International Journal of Neuroscience, 42:1-2, 157-161, DOI:

10.3109/00207458808985770

Kozel, F. A., Johnson, K. A., Mu, Q., Grenesko, E. L., Laken, S. J., & George, M.S. (2005)
 Libbey 14

Detecting deception using functional magnetic resonance imaging. Biological Psychiatry,

58:8, 605613.

Langleben, Daniel. Using Brain Imaging for Lie Detection: Where Science, Law and Research

Policy Collide. Psychol Public Policy Law. 2013 May 1; 19(2): 222234.

doi:10.1037/a0028841.

Ronald J. Allen, M. Kristin Mace, The Self-Incrimination Clause Explained and Its Future

Predicted, 94 J. Crim. L. & Criminology 243 (2003-2004)

Stoller, S. E., & Wolpe, P. R. (2007). Emerging Neurotechnologies for Lie Detection and the

Fifth Amendment. Retrieved from http://repository.upenn.edu/neuroethics_pubs/32