New Scientist
Nowhere to hide
09 Jun 01
We can tell you if you're guilty or innocent. You can't fool the lie
detector that knows what you are thinking. John McCrone investigates
YOU have just been arrested on suspicion of murder. You're sweating it out
in the interrogation room with a pair of beefy detectives. But your lips are
sealed-you know your rights.
Then with a smirk they slip a thing like a hairnet covered in dozens of tiny
electrodes over your head and sit you down in front of a computer. Pictures
of the crime scene begin to flash up on the screen interspersed with
multiple-choice questions. Flash! A photo of a brick wall. Flash! "What lies
behind this wall?" Flash! "Cement and blacktop?" Flash! "Sand and gravel?"
Flash! "Weeds and grass?"
You said nothing. You were even trying not to think. But sorry buddy, your
brain just gave you away. It couldn't help but show an electrical start of
recognition at the image matching the memory of hurdling a wall and wading
through a backyard of weeds as you fled.
An Orwellian fantasy? No, this technique was actually used in a recent test
case at a County Court in Iowa. The brain reading technology was developed
in university labs with CIA money. And it's not the only way that
researchers are searching for new ways to probe a lying mind. The US
Department of Defense is funding research into the use of multimillion
dollar brain scanners. Other labs are looking at more low-tech methods, such
as a simple reaction time test that can be an astonishingly reliable way of
discovering "guilty knowledge" you might rather conceal.
The field of lie detecting is long overdue for a shake-up. The polygraph is
still hugely controversial, based as it is on emotional responses such as
sweaty palms and changes in blood pressure or breathing patterns. Polygraph
results can be offered as evidence for the defence in US courts, and the
American Civil Liberties Union estimates that more than a million tests are
performed each year. But many people believe the polygraph is unreliable-the
Internet will tell you how to fool the machine by clenching your buttocks or
biting your tongue. However, the test is still widely used by security
forces in the US, Israel and Japan. In the US, the FBI and CIA screen
potential employees, and the US government is even pushing through the
polygraph for scientists working at national research labs.
But what if you could get inside someone's head? Forget about easy-to-fake
emotional responses. Just look for the differences in brain signals that
reveal when someone is lying, or even probe directly for the information
they're trying to conceal. Believe it or not, brain researchers can already
do this with startling accuracy.
It all began in the early 1990s when the CIA gave a little money to Emanuel
Donchin, a psychologist at the University of Illinois, Urbana-Champaign, and
his student, Lawrence Farwell, to see what they could do with an EEG test.
The EEG, or electroencephalograph, uses super-sensitive electrodes to
measure fluctuations in electrical potential caused by patterns of brain
activity. Donchin is an expert on a particular characteristic bump in the
EEG trace called the P300, which happens about a third of a second after you
notice something significant. It's like a mental click of recognition.
Crucially, it's automatic and utterly predictable.
How would the P300 expose a lie? There are two ways of using a polygraph.
The standard way is to first ask a stressful, but general, question like
"Have you ever driven while slightly over the limit?" This creates a
baseline reading before you jump in with the serious questions such as "Have
you had unauthorised contact with a foreign national?" The rationale is that
only guilty people will react strongly to actual accusations. It is of
course the ease with which the knowledgeable can pump up their arousal
during the baseline readings, disguising any later lies, which has brought
the polygraph into such disrepute.
But there is an alternative, little-used form of testing, known as the
"guilty knowledge test". Subjects are probed with pictures or phrases
significant only to them. A suspected KGB agent might have been tested for
an emotional reaction-such as a skip of the heart or ragged breathing-to KGB
code words. A suspected criminal would be tested for knowledge of a
particular crime.
Donchin and Farwell realised that the guilty knowledge test dovetailed
neatly with P300 recording. People with secret knowledge should show a P300
to otherwise innocent-looking pictures or phrases. They set up a lab test in
which subjects had to play-act spy scenarios-fictitious missions like
delivering the "owl file" to a contact in a blue coat in a particular
street. Then they recorded brain responses to lists of words which included
innocuous alternatives like the "fog file" and a contact in a red scarf.
Analysis of P300 responses picked out nearly 90 per cent of the "spies".
More importantly, there were no false positives where "guilty" brain waves
betrayed innocent people.
Although the researchers published their findings in 1991 in the journal
Psychophysiology, nothing much more happened until this year when a hearing
at Pottawattamie County Court in the backwoods of Iowa suddenly grabbed
international headlines. Someone was trying to use P300 evidence to get a
convicted murderer released.
Terry Harrington was jailed for life in 1978 for shooting a security guard
in the street. Harrington was just 17 at the time and claimed he'd been
miles away at a pop concert. But he was convicted on the testimony of
several witnesses, some allegedly his accomplices, and forensic evidence
including gunpowder traces found on his jacket. In a bid to win the right to
appeal, Harrington came to court to show that his brain did not react to any
memories of the crime scene but responded strongly to phrases connected with
events at the concert.
The scientist running the EEG tests was Farwell, who'd set up shop in Iowa
in the hope of turning the P300 research into a business. Farwell had
quietly spent the 1990s working with the CIA and the FBI trying to prove his
technology in the field. If Pottawattamie County Court could be persuaded to
accept his methods in this test case, he expected to revolutionise the whole
field of crime fighting.
"In a criminal act, there may or may not be many kinds of peripheral
evidence, but the brain is always there, planning, executing, and recording
the crime," says Farwell. "The fundamental difference between a perpetrator
and a falsely accused, innocent person is that the perpetrator, having
committed the crime, has the details of the crime stored in his brain, and
the innocent suspect does not."
Farwell's dream is that EEG testing, which he has dubbed Brain
Fingerprinting, will become a painless way of eliminating innocent people as
well as fingering crooks in any investigation. He says if every police
station had the right gear, suspects could volunteer to take the guilty
knowledge test and perhaps clear their name.
He claims his Brain Fingerprinting is foolproof if performed right. People
can disrupt the recording by blinking or refusing to look at the words. But
they cannot cheat to produce a false reading. Farwell says that he tried it
out on a psychopathic serial killer, whose lack of emotions would have been
a disaster for any polygraph test. "This guy never showed much of any kind
of emotion. He certainly wasn't normal. But I got a big Brain Fingerprint
response to facts about a murder," he says. "This method taps straight into
the cognitive processes of the brain and doesn't rely on an emotional
reaction."
When it came to the Harrington case, there were immense difficulties because
the murder happened more than 20 years ago, so Harrington's memories were
hardly fresh. Farwell also had to find details related to the murder which a
court could believe that Harrington had not learned during the original
trial or in the many years since. Poring over old court transcripts and
visiting the crime scene, Farwell felt that they could use the route the
guard's killer must have taken as he ran away which involved jumping a ditch
and crossing a weedy plot. When Farwell ran the tests, Harrington indeed
showed no P300 to these details, and a clear response to details about the
concert which was his alibi.
A cut and dried case? Unfortunately for Farwell and Harrington, it does not
seem so. In court, expert witnesses, including Farwell's old professor
Donchin, said the procedure was still too much of an unknown art, even
though the science was certainly sound. District attorney Rick Crowl scoffed
at Farwell's claims that there was a deep ditch at the time or that weeds
would have been that memorable in the rush to escape. Harrington's positive
response to the alibi details would simply have come from rehearsing his
story for so many years.
Farwell himself came under attack. Fun was made of the fact that he taught
Kung Fu and had said he was on the faculty of Harvard Medical School
(Farwell admitted this consisted of some brief consultancy work).
In March, the judge eventually refused Harrington leave to appeal. Farwell
says he is saddened, but at least he presented his Brain Fingerprinting
evidence before a judge, which sets a precedent for its use in future
hearings. Are we going to see a Brain Fingerprinting technician in every
police station? That's not likely to happen any time soon according to other
people at the trial, including district attorney Crowl.
The difficulty isn't the equipment, which is no more technically demanding
than the polygraph. Instead, it has more to do with the culture of
interrogation which prefers to see someone sweat. Cynics say the polygraph
is used purely to intimidate suspects. The aim is to prove the machine
cannot be fooled, making people think they have no hope of escape and so
confess. Any approach that would extract answers from a subject's mind in a
detached and clinical way wouldn't have the same effect.
But this isn't slowing the researchers down. Studies of what characterises a
lying brain are suddenly abundant. At the high-tech end of the market, the
US Department of Defense is funding Stephen Kosslyn, a psychologist at
Harvard University, to do magnetic resonance brain imaging studies. Kosslyn
says his first results are not that encouraging. People's brain activity
seems to be far from consistent when they are lying-but it is early days.
At least half a dozen other US labs are working on EEG measures. Perhaps the
most successful is Peter Rosenfeld from Northwestern University in Illinois.
Whereas Farwell's technique depends on a guilty knowledge test that shows
whether a person has a memory for a particular fact, Rosenfeld has recently
discovered a detectable distortion in the P300 signal just because you need
to concentrate when telling a lie.
In Rosenfeld's experiment a subject's own year of birth slipped into a
random series of four figure numbers was enough to produce a bump of
recognition in the P300. Some volunteers were instructed to answer "no" when
asked if they had seen it. When they lied there was a distinctive pattern in
the way the strength of P300 signal was distributed across the scalp.
Rosenfeld says he hopes EEG tests will both reveal guilty knowledge and
whether people are trying to lie during an interrogation.
And then taking everyone by surprise was the publication in February of a
low-tech version of the guilty knowledge test which needs no scanner or
electrodes, but just measures reaction times. Travis Seymour and Colleen
Seifert from the University of Michigan, Ann Arbor, repeated exactly the
same spy scenario as Donchin and Farwell, but they simply looked for
hesitations in the subjects' answers.
Seymour says they found that those who were telling the truth about a phrase
being unfamiliar could hit the "no" button in half a second. But people
telling a lie took more like a second. Even if they knew what was giving
them away and given a chance to practise, they could not react any faster.
"This seems a super-cheap and easy way of doing the guilty knowledge test,"
says Seymour. "All you need is an ordinary PC and a keyboard. No
electrodes." However, he adds that it would require much more work to take
such lab demonstrations further.
Rosenfeld agrees, saying researchers have been surprised at what you can do
in the lab but no one is doing the extra work to prove the techniques would
be safe for the interrogation room. Even if the scientists do a good job on
the test protocols, he feels that won't stop brain wave and reaction time
technology being abused just like the polygraph. He says the FBI knows that
the polygraph is unreliable. But they still value it as a prop because
people can easily be frightened into confessing if they believe the machine
is reading their every emotion. How much better a prop would a set of
electrodes and a box of expensive electronics make? The scientific validity
of brain measures would be almost beside the point.
And yet there seems real potential in the recent EEG work. Civil rights
activists take note. Tomorrow we may still enjoy the right to remain silent.
But that might be pointless if the investigator can read your mind.
Brainwave patterns show if you're innocent or guilty
Further reading:
* More information at www.brainwavescience.com
John McCrone is author of Going inside: a tour round a single moment of
consciousness published by Faber and Faber
>From New Scientist magazine, vol 170 issue 2294, 09/06/2001, page 24
