Truth and Consequences
Scientists are scanning the brain for traces of guilty knowledge
On July 22, 1977, the body of retired police Captain John Schweer lay bleeding on a railroad track in Council Bluffs, Iowa. Blame for the murder fell on teenager Terry Harrington, who has spent over two decades in a state penitentiary protesting his innocence. Now Harrington says he finally has proof that exonerates him. A new technology called "brain fingerprinting" suggests that Harrington's brain doesn't contain memories of the crime scene, but it does recognize specific details about the concert he says he attended on that fateful night.
Reading someone's memories through a helmet of electrodes sounds like science fiction. But surprisingly, scientists say they can do just that. The Harrington case is the first time lawyers have ever submitted such brain measurements to a U.S. court, and within days a judge will either reject them as "junk science" or accept them as evidence that could help reopen Harrington's case.
Until now, detectives have only had polygraph machines, or "lie detectors," to search for physiological measures of guilt and lies. But critics say polygraphs are hopelessly flawed because they measure emotional responses like blood pressure, breathing, and sweating. Nervous truth-tellers can appear guilty while cool liars look innocent, and people can beat the machine. Although polygraph tests have been the mainstay of lie detection since the 1920s, they're controversial enough that most states don't allow them as evidence. This month, the National Academy of Sciences will start a major review of polygraph tests' scientific validity at the request of the Department of Energy, which uses them to screen employees.
Many researchers hope they can do better than polygraphs by targeting the source of the lie: the brain. The "brain fingerprinting" technique used on Harrington was developed by Lawrence Farwell, a scientist and businessman at Brain Wave Science in Fairfield, Iowa. Farwell's technique exploits a signal that the brain emits when it perceives something familiar.
If a person watches random numbers flash on a screen, for example, the brain will suddenly show a distinctive electrical response called the P300 if personal information like a home phone number pops up. Scientists have known about the P300 response for decades, but no one has tried to apply it in a real-world criminal situation. Farwell believes it is valid in the Harrington case because he found details about the crime and Harrington's alibi that weren't in court transcripts or newspapers. For example, the murderer escaped by running behind a building through waist-high weeds and grass. According to Farwell, the real murderer would remember tearing through this thick underbrush. But when Farwell showed Harrington a photo of the building, the convict said he didn't know what lay behind it. And when presented with words such as "cement and blacktop," "sand and gravel," and "weeds and grass," Harrington had no special brain response to the correct answer.
Lacking knowledge. The test isn't a lie detector and doesn't prove Harrington's innocence, Farwell says. But Harrington "does not have certain critical details about the crime stored in his brain. A judge or jury must decide how they're going to interpret that."
The idea that knowledge stored in the brain can be used to determine guilt is an old one, and it even has a name: the Guilty Knowledge Test. In 1959, David Lykken of the University of Minnesota first proposed that detectives could present suspects with multiple-choice questions to see if they would react to specifics that only the perpetrator would know. The challenge is to find biological measures of that reaction that aren't affected by emotional responses. Brain scanning might be one way, but researchers also have others.
Someday, investigators could check to see if a suspect knows other criminals or victims by tracking eye movements. That's because the brain scans familiar faces differently than it does strangers' faces. Another way of getting into the brain might be to look at response times. Travis Seymour, a cognitive psychologist at the University of Michigan-Ann Arbor, has studied how quickly people can press buttons while being shown information on a computer screen. When presented with words and images that they either know or don't know, he has found it takes people longer to press a "don't recognize" button when they're lying--nearly twice as long as when they truly do or don't know.
One problem with such Guilty Knowledge tests is that they won't work for screening exams, which by definition involve questioning people about unknown crimes. But scientists are making progress on that front as well. "I don't know about you, but when I'm lying, I'm aware of it. There's got to be some sort of substrate for that awareness in the brain. It's just a matter of finding the right tools to unmask that substrate," says Peter Rosenfeld of Northwestern University. Rosenfeld is currently trying to use electrical readings other than P300 to see if he can find a telltale profile for lying.
Researchers already have preliminary evidence that brain-imaging scans can be used to distinguish lying from truth telling, according to Stephen Kosslyn of Harvard University. It's not as simple as a Pinocchio lobe lighting up on scans when a person lies. Different types of lies, such as spontaneous or premeditated fibs, seem to depend on different brain areas. Kosslyn's been able to distinguish between lies and truth on brain scans by looking at ratios of activity in various brain regions and time lags in those regions' activation.
Scanning for lies. This brain imaging work is a real breakthrough in lie detection, agrees Terrence Sejnowski of the Salk Institute for Biological Studies in La Jolla, Calif. Sejnowski's currently working on a computer program that can detect lies by analyzing fleeting facial expressions. It's an offshoot of work pioneered by Paul Ekman, a psychologist at the University of California-San Francisco, who can watch people on videotape and catch liars by analyzing almost imperceptible expressions like eyelid flutters and strained smiles. But it takes experts like Ekman an hour to catalog all the microexpressions on one minute of videotape. The goal is to train a computer that can do the analysis in real time, so that a computer with a camera could screen people for lying without them even knowing it.
Is the world really ready for a foolproof lie detector? A world without lying has been explored in popular movies like Liar Liar, and the prospect generally seems to make people nervous. Yet those who feel they've been wrongly imprisoned will embrace any new technology that might set them free. "I have many, many requests from people on death row and people in law enforcement," says Farwell. "I think in many of these cases, we'll be able to shed some light on what they do or do not know about the crime." Eventually, brain fingerprinting and other new methods will likely find their way into courtrooms, where it will then be up to judges and jurors to decide where the truth lies.
Innocent?
Lawrence Farwell (right), creator of the "brain fingerprinting" test that measures the brain's response to information flashed on a computer screen, showed Terry Harrington (left) information that was familiar, unfamiliar, or related to the scene of a 1977 murder for which Harrington had been convicted. Harrington's responses to the crime-scene information closely matched his responses to unfamiliar information--suggesting that he lack knowledge of the crime scene.
Harrington's response to information that was:
[Chart data are not available.]
[Key to chart] Familiar; Unfamiliar; Related to crime scene
Source and Photo: Lawrence Farwell, Brain Wave Science
This story appears in the January 15, 2001 print edition of U.S. News & World Report.