IP: Medical Detection Of False Witness

[David Farber <dave () farber net>]

Insight on the News - National
Issue: 02/04/02


Medical Detection Of False Witness
By Brandon Spun

Try this scenario: Zacarias Moussaoui is led out of a dark cell into a 
silent room at an undisclosed location. An electrode headset is fitted to 
his skull while his lawyer watches disapprovingly. After nearly an hour of 
flicking switches and flashing lights, the procedure concludes. An 
investigator reviews the results and determines exactly what role Moussaoui 
played in the Sept. 11 terrorist attacks. You cannot hide your memories 
from the machine.

Some say this no longer is science fiction. In the months following Sept. 
11, articles about techno-security have appeared in the popular press and 
in professional journals promising just such a result from high-tech lie 
detectors. One of these machines was featured in the New York Times 
Magazine's "2001 Encyclopedia of Innovations, Conceptual Leaps and 
Harebrained Schemes." An Insight review of new polygraph technologies 
suggests that most of them fit all three categories.

"When one uses any kind of lie detector, one is saying 'ask the body, not 
the person,'" says Mike Gazzaniga, director of the Center for Cognitive 
Neuroscience based at Dartmouth College. "One assumes with autonomic 
studies [polygraphs] that the body can't lie," Gazzaniga says. The 
assumption is that the body's autonomic reactions — such as blood pressure, 
breathing and heart rate — cannot be manipulated to support deception. 
"They assume that lies are mental constructs," Gazzaniga continues. "Among 
those things that separate man from animal is his capacity for deception. 
The question is how to make a science and not folk psychology out of this."

One investigative method established for autonomic research is the 
"guilty-knowledge test," which consists of confronting a suspect with a 
series of items, some relevant to a crime and others irrelevant. It is 
believed that a subject who has "guilty knowledge" will react most strongly 
to relevant or target items, while an innocent person will not.

Such tests have been disputed since the first polygraph was invented by 
William Marston in 1917. A polygraph can be an effective tool, but 
high-profile cases such as those of Aldrich Ames and Robert Hanssen, spies 
who beat the machine, frequently are cited to show its fallibility. Critics 
point out that data gathered from these tests are indicative of 
physiological response rather than veracity, with the latter being 
distinguished from the former. They say there is no necessary connection 
between a subject's verbal and physical report.

Polygraph expert Drew Richardson, a former FBI special agent at Quantico 
Laboratories, says "the test is generally not admissible in court because 
it is largely not accepted as a valid scientific technique."

But what if a better mousetrap were to become available? What if there were 
a machine that could not be fooled? What if an autonomic response to lying 
were to be identified that could not be suppressed? Recently, some have 
made such claims. Here is a look at those claims and what is being said 
about them.

Brain fingerprinting: According to Larry Farwell, an independent 
psychophysiologist at Human Brain Research Laboratory in Fairfield, Iowa, 
the fundamental difference between a guilty and innocent person is a record 
of the crime stored in the brain. "The difference between a terrorist who 
has been through a training camp and an Afghan student is the memory," 
Farwell says. Brain fingerprinting identifies memories, he claims.

This examination consists of a modified guilty-knowledge test that uses 
targets and irrelevants, but adds probes that are unique details of a crime 
that only the culprit could know. All subjects are familiarized with the 
targets in order actively to identify them during the test. A guilty 
subject is expected to have the same response to the probes as to the targets.

When one is exposed to something that already is stored in memory, the 
brain emits an electrical response called a p300 wave. This phenomenon 
occurs approximately 300 milliseconds after a meaningful stimulus. The "p" 
stands for positive electrical voltage (certain speech processes emit a 
negative 400). Electrodes on the parietal zero (the top of the back of the 
head) record this activity.

Brain fingerprinting has been tested on FBI agents and in field situations. 
Farwell reports a 99 percent success rate, though he estimates the process 
is applicable for only 70 percent of investigations. What has attracted 
attention is his claim that Sept. 11 may be a good fit. He proposes sorting 
suspects into three groups through his method for lie detecting: those 
involved in the planning of the attacks, terrorists from different cells 
and the innocent. "We were able to detect whether someone was an FBI agent 
by using information from their training manuals," Farwell says. "We could 
use the same method with al-Qaeda's manual for terrorists."

One of the first things other scientists familiar with Farwell's proposal 
for lie detection point out is that the last peer-reviewed paper he 
published on this was in 1991. Emanuel Donchin, a professor of psychology 
at the University of South Florida, contributed to the federal evaluation 
of Farwell's techniques. He also was Farwell's teacher and 1991 research 
partner. "Larry does a nice demonstration, but it needs much more 
research," Donchin says. "It is not ready as a practical tool." Donchin 
raises three major problems with the Farwell approach: stimuli, the oddball 
paradigm and interpretation.

The stimuli problem is that a probe is chosen by subjective analysis. "An 
investigator, not science, makes the decision," Donchin says.

The oddball paradigm involves the order of stimuli. Apparently probes 
change the environment of the brain. "A p300 is enormously sensitive to the 
order of events," Donchin says. It can occur because something is 
infrequent as well as meaningful. So detecting a p300 is equivalent to 
detecting that a memory has been distinguished by the brain. Thus, as 
Donchin succinctly puts it, "response to a probe ensures nothing."

Interpretation is a standard problem for lie detection. Donchin is not 
convinced that Farwell has overcome it. "Larry's interpretation of what 
must be remembered as significant is invalid and subjective," he says. 
"Larry is an entrepreneur, a businessman advertising a product. If you get 
people gullible enough to buy your product, that's all that is required."

A pre-eminent figure in psychophysiology had more to say on the subject. 
J.P. Rosenfeld, a professor at the University of Utah, says evaluators 
should look at the distribution of several p300s recorded from an 
individual subject while Farwell uses only one or two. "Farwell arbitrarily 
selected a wave," Rosenfeld says. "I am sure his data would be discounted 
by impartial psychophysiologists."

Rosenfeld also is concerned about Farwell's choice of subjects. "It turns 
out he has done extremely well using highly motivated, paid subjects from 
his labs," Rosenfeld says.

Conflict resolution: What first interested Daniel Langleben, assistant 
professor of psychiatry at the University of Pennsylvania, in lie detection 
was not terrorism but the behavior of drug addicts. "A substance abuser 
lies with no protective value," Langleben says. "For instance, you are 
practically suicidal if you don't tell an emergency doctor you take 
insulin," and yet they often don't.

Langleben, a psychiatrist as well as a self-taught anthropologist, says his 
mother, a linguistics professor, is an enthusiast for the works of St. 
Augustine and Immanuel Kant. He believes all this has been good preparation 
for studying the willful lie as opposed to philosophical truth. "That is, I 
am not interested in what philosophy would call objective truth," Langleben 
says. "My notion of a lie, what I explore, is remarkably similar to 
Augustine's."

Langleben has performed magnetic resonance imaging (MRI) on the brains of 
subjects as they underwent guilty-knowledge tests. While p300s provided 
information regarding quantity of activity, Langleben looked for blood flow 
and spatial resolution. He found what he says is the part of the brain 
active during a lie.

A normal or bold MRI is used to locate unusual activity, such as tumors, in 
the brain. It works by placing the body within a strong magnetic field 
while a weak, perpendicular magnetic field is turned on and off. This 
causes magnetic resonance, like flicking a car antenna. An fMRI, a bit less 
accurate, reads changes in blood flow up to 4 millimeters. However, by 
overlaying scans from fMRIs and bold MRIs, Langleben located blood flow at 
6 to 8 millimeters even on the worst day.

When a person thinks, parts of the brain need more oxygen. Having performed 
fMRIs on people engaged in guilty-knowledge tests, Langleben claims to have 
isolated the location in the brain active during willful deception. He 
believes it is the anterior cingulate cortex, a part of the brain often 
associated with conflict. Though a subject might control supposed autonomic 
responses, this discovery is promising because we could pinpoint the very 
act of deception.

Langleben has not contacted the Department of Justice and was reluctant to 
comment on what should have been done to check the veracity of those 
arrested and interrogated for terrorism. However, he says, any tool to 
speed up the process of identifying lies in such circumstances is 
important. "Unlike most endless scientific questions, this one is 
solvable," he says.

But admittedly there are problems with this lie detector, says Langleben 
assistant Ruben Gur. "For one thing, you need a cooperative individual, as 
one must remain very still during these tests." Any movement invites error, 
and a subject also must remain relatively relaxed as emotional arousal can 
alter results.

Gur suggests the next step in research should be to perform fMRIs in 
conjunction with other bioindicators. And Langleben is aware of the need 
for more study. "If this is the Manhattan Project we are at 1941, not 
1944," he says.

Other experts note problems. Rosenfeld suggests the theory may be flawed 
because it involves averaging multiple results, preventing case-by-case 
examination. He also points out that fMRIs are extremely expensive.

And, though the anterior cingulate cortex may be related to conflict and 
lying, not everyone is certain. This is a frontier subject. "Some say it is 
responsible for almost anything," says P.J. Casey, professor of psychology 
at Cornell University in New York. Others see it as a homunculus, or a 
little brain running the bigger show. "But most do say it has to do with 
conflict or conflict resolution," Casey says.

The foremost expert in this area, psychiatrist John Cohen of Princeton 
University, is of the opinion that activity in the anterior cingulate 
cortex is indicative of conflict but not resolution. If so, Langleben's 
method may not always tell us what we want to know.

Casey explains this in a paper in which she says "paradigms of affective 
processing often require the subject to induce an affective state or think 
of emotional information that is contrary or in conflict with the subject's 
current state of affairs." This is crucial because, in other words, a 
person can consider an option without actually choosing that option, 
thereby inducing conflict but not producing it. This would leave 
investigators wondering whether a subject lied or merely considered lying.

The eye detector: Lying eyes are the latest addition to the list of science 
wizardry. Fox News and Science Daily are among those that have claimed that 
thermal-imaging lie detectors soon may be in place at airport-security 
checkpoints. They are referring to camera studies by James Levine, an 
endocrinologist at the Mayo Clinic's Honeywell Laboratories.

Levine and his collaborators theorized that subjects blush just before 
lying. They tested this theory using high-definition thermal-imaging 
cameras originally designed to see through heavy makeup and disguises.

Tests were conducted at Fort Jackson, S.C., as new recruits tried to fool 
the camera during an interrogation after some of them stole $20 from a 
mannequin they stabbed. Levine reported that in these tests the camera 
identified a lie with 80 percent accuracy. Advantages of the technology 
include quick investigations, real-time results, noninvasive screening and 
unskilled operation.

Levine says more testing is necessary and that the equipment must be 
refined for airport use, but he claims his lie detectors soon will be in 
commercial use. Other scientists tell Insight that the "lying eye" needs 
more than refinement. In fact, some call the Levine system a mere blush 
detector. "Big Brother? More like special brother," commented one.

While Levine talks about his work as both "new" and "potentially accurate," 
critics say it is neither. Here are excerpts from a letter to the 
prestigious journal Nature that John Furedy, a professor of psychology at 
the University of Toronto, is preparing in response to Levine's 
lie-detector research:

"The procedure was only 'validated' against polygraph examination by 
experts (at the Department of Defense Polygraph Institute). There was no 
mention of a considerable body of scientific, psychophysiological 
literature that casts grave doubt on the scientific basis of this purported 
application of psychophysiology.

"The fundamental problem with the polygraph, even when administered by 
'experts,' is that the measures it uses, such as electrodermal response, 
are virtually useless for differentiating the anxious but innocent person 
from the anxious and guilty one. Why should we think that thermal-imaging 
measures will be any more discriminating?

"It is disturbing not only for Americans, but also the world, that the 
national security of the world's only remaining superpower appears to 
depend on this modern flight of superstitious technological fancy, the only 
effect of which is to spread distrust within those organizations that 
employ it."

Or, as Rosenfeld says of the currently excited state of lie detection, 
"people were excited after 9/11, but that didn't advance knowledge at all, 
just funding."

Brandon Spun is a reporter for Insight.





------------
FAIR USE NOTICE: This contains copyrighted material, which is reproduced
under the Fair Use Provision of Title 17, U.S.C. Section 107, and is posted
for purposes such as criticism, comment, news reporting, teaching,
scholarship, or research. This material is posted without profit for the
benefit of those who, by accessing this site, are expressing a prior
interest in this information for research and educational purposes.

For archives see:
http://www.interesting-people.org/archives/interesting-people/