Marketing managers spend millions every year on focus groups in an attempt to probe consumers' decision-making processes.

Now a technique known as neuromarketing promises to provide snapshot images of brain activity at crucial moments of retail choice. Scientists have been putting volunteers into MRI (magnetic resonance imaging) scanners to find out what goes on in their brains when they look at pictures of consumer goods.

Their findings appear to offer new opportunities for manipulating consumers. But the idea that scientists can equip companies with sinister powers to influence the public has been a recurring fear and an unfulfilled promise since the 1950s.

Until recently MRI has been used only in clinics - for diagnosing strokes or discovering tumours - or for pure research, such as identifying brain regions linked with movement or emotion. Now laboratory insights are being matched to the needs of marketing managers. Two neuromarketing centres, Brighthouse Institute and the Mind Marketing Laboratory, have recently opened in the US.

MRI scanners were used this year for an investigation of a modern marketing conundrum: why Coca-Cola outsells Pepsi, even though blind tasting frequently shows more people prefer the taste of Pepsi. When Read Montague of Baylor College of Medicine in Houston, Texas, re-created the Pepsi Challenge blind tasting campaign of years ago, he found those who preferred Pepsi showed a five times stronger response in one of the brain's reward centres (the ventral putamen) than those who liked Coke. Then he ran the scans again but this time the volunteers knew which drink they were tasting.

"The result was remarkable," says Dr Montague. Not only did the subjects nearly all say they preferred Coke but another area at the front of the brain, the medial prefrontal cortex, which is linked with thinking and judging, lit up as well as the ventral putamen. "This showed that subjects were allowing their memories and other impressions of the drink - in other words its brand image - to shape their preferences." A strong brand, it seems, can override our taste buds.

The conclusion was that if you find what stimulates the medial prefrontal cortex you may have the basis of a successful advertising campaign.

Dr Montague's work was picked up by the Bright- house Institute of Thought Sciences, based in the neuroscience wing at Emory University Hospital in Atlanta, Georgia.

"This [the medial prefrontal cortex] is the area that is linked to our sense of self," says Clint Kilts, scientific director of Brighthouse. It is the area that used to be knocked out in a lobotomy: damage here can cause drastic personality change. "If it fires when you see a particular product," he says, "you are more likely to buy because that product clicks with your self-image." The Brighthouse team has found that when volunteers say they "truly love" something, the medial prefrontal cortex lights up on the scans.

The scans can also give other indications of what is going on in your mind. When the team see an area called the somatosensory cortex fire up, for instance, they know the subject is imagining using an object. This region, which controls our physical movements, is also active when we are thinking about a movement. But until the medial prefrontal cortex lights up, we have not fully identified with the product.

Such findings could open the way for companies to discover whether their products are triggering this vital response. "They show that preferences have measurable correlates in the brain; you can see it," says Justine Meaux, a neuroscientist at Brighthouse. "We can use this difference to guide our decisions about how we market to people."

Peering into the consumer's brain in this way is not cheap. Use of an MRI scanner that generates moving pictures costs about $1,000 an hour, while a single experiment involving 12 subjects can cost $50,000. The fee for a recent study by Brighthouse, with 30 subjects viewing images ranging from rock climbing and President George W. Bush to BMWs and the National Enquirer, was $250,000.

Against the cost is the argument that insights lead to an advertising advantage that could be worth millions.

The frontal cortex is the area of the brain that is the last to develop fully and the first to decline. It is at its peak between about 12 and 50 years of age. This means very young children and old people are more susceptible to urges that come from the emotional amygdala.

Although neuromarketing is so far largely a US phenomenon, scientists at the University of Ulm in Germany recently reported on men asked to rate the attractiveness of cars. When they looked at pictures of glamour cars such as Porsche and Ferrari, an MRI scan showed activity in a brain region used for processing faces. This area has strong links with the main emotional centres.

Other studies, using the more more basic electroencephalogram, which records electrical activity from the scalp, claim to have found that activity in the left prefrontal cortex indicates an "approach" response, while one on the right indicates "revulsion". The EEG has also been used to explore how well a message is remembered. Richard Silberstein, a neuroscientist at Swinburne University of Technology in Australia, argues that "the better you remember a commercial message the more likely you are to buy the product".

Before anyone gets excited or anxious about manipulation and control, some neuroscientists are expressing scepticism. Chris Frith of the Institute of Neurology in London, says: "Just because you can see and measure an increased level of activity in the brain, people feel that is more authoritative than someone saying what they are thinking or feeling. But we don't really know enough about how the brain works as a system to be able to apply this research. It is too early to say what findings like these mean."

The uncertainty Prof Frith outlines is highlighted by work under way at the Max Planck Institute in Tubingen in Germany. Researchers there have found that while MRI scans give accurate information about how much information is coming into a brain area, they are not so informative about what is being passed on elsewhere. They may tell only half the story.

But the promise of neuromarketing is alluring and, rightly or wrongly, we may yet see political focus groups swept away as volunteers are herded into scanners to see which politicians tickle their medial prefrontal cortexes.