杏吧原创

DONNA insists that she met friends for lunch on the afternoon of 25 January 2008 and did not violate a restraining order against Marie. But Marie told police that Donna broke the terms of the order by driving up to her car while it was stopped at a traffic light, yelling and cursing, and then driving off.

A polygraph test proved unsatisfactory: every time Marie鈥檚 name was mentioned Donna鈥檚 responses went sky-high. But when Donna approached of Tyngsboro, Massachusetts, for an fMRI scan, which picks up changes in blood flow and oxygenation in the brain, it was a different story.

鈥淗er results indicated that she was telling the truth about the January 25 incident,鈥� says Steven Laken of Cephos, who maintains that when people are lying, more areas of the brain 鈥渓ight up鈥� than when they are telling the truth (see the scans of Donna鈥檚 brain).

Laken claims that when the police saw Donna鈥檚 fMRI results, they dropped the investigation against her 鈥� although Donna鈥檚 cellphone records also corroborate her story.

Cephos, which started offering commercial testing in the US earlier this year, is one of several companies that claim they can show whether someone is lying using fMRI. Until recently, these firms claimed they could detect lies up to 90 per cent of the time. However, last week, at at the University of Akron in Ohio, Laken announced that Cephos had achieved an accuracy of 97 per cent. The secret, he said, was to ask questions in small chunks rather than as a continuous stream, a tweak that allows the brain to 鈥渃ome back to a resting or normal state before you ask another series of questions鈥�, leading to clearer results. 鈥淎s far as I know that鈥檚 the highest degree of accuracy anyone has managed.鈥�

Laken hopes to pass another, more controversial milestone by early next year: getting what is essentially crude mind-reading admitted as evidence into the US courtroom.

Will he succeed? The decision will be made on a trial-by-trial basis by judges who will likely turn to a set of criteria known as the Daubert standard, which says that scientific evidence accepted in a US court must be testable and produce consistent results, be peer reviewed and published, have known error rates and have gained acceptance by the scientific community.

Laken argues that fMRI meets these criteria and is a widely accepted technology cited in more than 10,000 scientific publications. 鈥淲hen we look at what the courts are requiring, I believe we have met those bars,鈥� he says. Leo Kittay of in New York agrees that Laken is close, although he says judges may also be swayed by other neuroscientists鈥� disapproval of the technology.

Even if fMRI meets the Daubert standard, it is not clear that it is appropriate to the courtroom 鈥� at least right now. 鈥淭his explosion in neuroscience is something that courts are going to have to consider, and we鈥檙e not really ready for it yet,鈥� says of the University of Akron School of Law, who organised last week鈥檚 meeting in part to educate judges and lawyers about neuroimaging.

鈥淭his explosion in neuroscience is something that courts are going to have to consider and we鈥檙e not really ready for it yet鈥�

One criticism is that the vast majority of fMRI studies do not focus on its use in lie detection specifically. Of those that do, barely any involve real-world situations. 鈥淢ost of the studies have been of students telling fairly innocuous lies,鈥� says , a neuropsychiatrist at the University of Sheffield, UK.

Cephos achieved its 97 per cent accuracy result using men and women in their 30s instructed either to tell a lie or the truth. 鈥淭his is probably very different from a criminal trying to get off a criminal charge by lying to the police,鈥� says of Stanford Law School in California.

He believes that until many more studies are carried out on a broader range of people, including those who have a personal stake in telling a lie, there should be a ban on non-research use of fMRI-based lie detection. But such studies are extremely hard to conduct, as you can鈥檛 expect someone accused of a criminal act to reveal whether they are telling the truth.

People may also be able to trick the test. It is well known that people can defeat the polygraph by controlling their breathing when they are lying, and by trying to artificially increase their heart rate during control questions. Laken argues that fMRI is harder to fool than the polygraph because it monitors changes in the brain during the formulation of a lie, rather than the stress responses associated with lying.

Greely counters this by pointing out there may be other ways to trick fMRI. At least one study suggests that fMRI scans of people who are telling spontaneous lies look quite different to those of people telling a planned lie. 鈥淢aybe if you鈥檝e rehearsed a lie and memorised it, it doesn鈥檛 look like a lie,鈥� he says.

What鈥檚 more, because fMRI is motion-sensitive, you need a cooperative subject. Someone who doesn鈥檛 want to be scanned just has to move their head or even their tongue inside their mouth to scupper the results. 鈥淚t鈥檚 very hard to figure out how you鈥檙e going to fix that,鈥� says Greely.

Even if independent studies confirm Cephos鈥檚 results, there will still be questions about the ethics of using fMRI in court. 鈥淚t would be an invasion of what we see as a typical jury function: judging credibility,鈥� says Moriarty. 鈥淚 think people also see it as invasive of people鈥檚 privacy on some level that courts find discomforting.鈥�

Greely says he鈥檇 be surprised if fMRI was admitted in a courtroom by 2009, but doesn鈥檛 rule it out. 鈥淚 think there鈥檚 a chance that there are judges that will admit this,鈥� he says. 鈥淭here are a lot of judges out there, and sometimes they surprise me.鈥� While the decision of one trial court judge doesn鈥檛 bind other judges, it may encourage others to follow suit.

鈥淭here鈥檚 a chance that judges will admit this as evidence 鈥� there are a lot of judges out there, and sometimes they surprise me鈥�

Concerns also loom large that the US military is using fMRI to screen people it suspects of committing terrorist acts to decide who to interrogate, something the Department of Defense has denied. One thing is clear: Cephos鈥檚 attempt is just the beginning of fMRI鈥檚 influence outside the research world.

See Editorial: Latter-day lie detectors have yet to prove their worth

The Human Brain 鈥� With one hundred billion nerve cells, the complexity is mind-boggling. Learn more in our cutting edge special report.