Neuroscientists who use a popular scanning technique to look at brain activity can breathe a sigh of relief, researchers say.

A paper in last week's Science suggests that, as well as identifying which parts of the brain are activated during certain tasks, the scans also provide a measure of actual brain­cell activity (R. Mukamel et al. Science 309, 951–954; 2005). Although researchers have been working on this assumption, it is the first study to show this in humans.

Functional magnetic resonance imaging, or fMRI, has been widely used to investigate which parts of the brain are active during anything from being in love to telling lies. But some researchers have questioned whether the technique reveals anything significant about the brain's workings. This is because fMRI does not measure brain-cell activity directly, but changes in blood flow within the brain in response to various stimuli.

The latest results could put those doubts to rest says the study's lead author, Rafael Malach of the Weizmann Institute of Science in Rehovot, Israel. “It's a great relief to know that these studies in which dozens of labs are now involved are not a waste of time,” he says.

In 2001, researchers in Germany showed that fMRI scans could be correlated with brain-cell activity in anaesthetized monkeys (N. K. Logothetis et al. Nature 412, 150–157; 2001). But until now, no one had done the same thing for conscious human brains.

Models using brain-cell data show good agreement with real scans. Credit: AAAS

Malach joined forces with a group at the University of California, Los Angeles, led by neurosurgeon Itzhak Fried, that works with patients suffering from severe epilepsy. Before they can have surgery, the patients spend several days with electrodes wired into their brains so that the doctors can pinpoint the source of their seizures. Fried's team took advantage of this situation by recording the activity of the brain cells in two epilepsy patients who were shown nine minutes of the Clint Eastwood film The Good, the Bad and the Ugly.

The researchers built a computer model to convert the measured activity into fMRI images. They then compared these predicted images with actual fMRI scans taken of 11 healthy volunteers who watched the same movie clip. In general, the model was accurate half to three-quarters of the time, but for certain points in the movie, agreement reached 90%. The correlation is encouraging, the team says, given the dramatic differences in the patients and their settings.

The results apply to only one area of the brain, and describe the reaction to just one kind of stimulus. But some neuroscientists are pleased that the study is looking at real human behaviour.

“They're actually getting a little bit closer to what the brain is doing, and I think that's great,” says neurophysiologist David Leopold of the US National Institute of Mental Health in Bethesda, Maryland.