The scans watch for giveaway brain patterns when people are under interrogation. Five years ago, fMRI researchers at Temple University, Philadelphia, found that we use different parts of our brains when lying: our cingulate gyrus, in the middle of the brain, lights up if we are being honest, but lying stimulates the limbic lobes, towards the forebrain. Lying also requires us to use more brain power. “Virtually all our clients want to show that they are telling the truth,” Laken says. “One third are in some private family or work matter. About a third are civil or criminal. The other third are people who have been jailed and want to prove they are speaking the truth about their innocence.

“There are ethical things that we have to face,” he acknowledges. “We would refuse anyone who wanted to bring their spouse in for a ‘surprise’ MRI scan while they were being asked about fidelity.” But what if a paying client tests as guilty about some awful misdeed? “Not everyone who has come to us has proved their innocence. When we communicate the results to a client, they can say, ‘Get rid of the data,’ and we do. It’s their data; it’s up to them what happens to it.”

With the help of American technology, Professor Sean Spence of Sheffield University has pioneered the use of brain-scan lie detection in Britain, employing the test on Susan Hamilton, 43. The Edinburgh mother was sentenced to four years in prison in 2003 for poisoning a child in her care with salt overdoses, but she maintains her innocence and is campaigning for the sentence to be repealed. Spence’s study, published last year in European Psychiatry, reported that the scan was consistent with someone telling the truth. The results have not, however, been submitted as evidence to a British court.

Steven Laken claims that the technology can be developed far further – into a new discipline of “forensic brain-scanning” that examines people’s intentions, goals and feelings. “Does someone understand that what they did was wrong, or did they intend to do it? This makes the difference between murder and manslaughter. We may also be able to tell if someone has been in a terrorist camp, or had certain motivations. For example, if you show someone a place they recognise, their brain reacts differently under fMRI than if they are seeing a picture of a place they never visited. With eyewitnesses, false memories light up different parts of the brain than true memories, which could be very useful for asking witnesses to identify criminals.”

How soon can all this happen? “It depends on whether people want to commit money and resources,” says Laken. “It could be done in 12 to 16 months if there were the government will. After all, we developed the atom bomb in less than four years.”

The technology is constantly developing. A new superconducting material, doped rare earth iron oxyarsenides, may enable scientists to boost the magnetic fields that create an fMRI scan’s sharpness and shrink the giant scanning machines to a portable size. Another imaging system under development, near-infrared spectroscopy, uses lasers to measure blood flow in the brain. It is non-invasive and may be carried in a suitcase, enabling investigators to interrogate people at home.

Joel Huizenga, chief executive of the boldly named No Lie MRI, says he has already applied for patents in America and Britain to use both fMRI and near-infrared scanning. He is negotiating to open facilities with UK companies that use scanners for medical applications. Huizenga’s product does not stop at lie detection: “People want clarity over three topics: sex, power and money. We are using our systems to do ‘Are you going to buy?’ detection for advertisers, ‘Do you recognise a face?’ for police line-ups, and ‘Are you in pain?’ for compensation claims.”

Despite all the bullish optimism, sceptics argue that much commercial brain scanning may prove more akin to brain scamming. Dr Lewis has been working in the field since the Eighties, when he says there was “commercially no interest in it at all”. Now that neuroscience’s tempting potential is becoming commonly understood, he says, entrepreneurs are offering products that race ahead of what is currently possible. “There are companies doing well-substantiated stuff with this technology, but there are also cowboys. There is speculation, there is speculation squared, and then there is neuroscience. The people who buy it are often too eager for the promises to be true.”

Indeed, a research review published last December in Perspectives on Psychological Science condemned the scientific basis of half of the 54 peer-reviewed fMRI brain-scanning studies it covered. The review, conducted by the University of California at San Diego, concluded that 27 of the studies’ statistical measures of brain activity were so poorly analysed that the findings were worthless.

Lewis adds: “The amount of light we have shone on the brain merely serves to show how much more darkness there is. You can scan the brain while it’s experiencing something and see certain areas light up, but correlation is not causation. When you scan someone’s brain while they watch a Pepsi or Coke commercial, the conclusions you draw from the data can only be extraordinarily speculative. In terms of technology, we are just beyond the Wright brothers, flying around in a few flimsy biplanes. But often people are selling and buying the products as though they were jumbo jets.”

John-Dylan Haynes is in the vanguard of scientific breakthroughs using fMRI scans to predict thoughts and actions. But he shares Lewis’s fears and wants to see a vigorous debate on the technology’s applications. Real advances continue to be made, he says, but a fundamental question remains: can they ever cut it in the brain-swirling world outside the laboratory?

“The science is so exciting that it is being picked up by commercial people who are vastly overselling its power,” says Haynes. “There are companies offering lie detection, but I don’t believe these techniques are operating at a level fit enough for a commercial service or a court of law. Even with sufficient funding, it will take ten to twenty years possibly to develop techniques that are close to a reliable universal lie detector.

Explore Health

• Expert Advice

• Health Features

• Mental Health

• Alternative Medicine

• Child Health

• Health Club