Teams at University College London and University of California in LA could tell what images people were looking at or what sounds they were listening to.

The US team say their study proves brain scans do relate to brain cell electrical activity.

The UK team say such research might help paralysed people communicate, using a "thought-reading" computer.

In their Current Biology study, funded by the Wellcome Trust, people were shown two different images at the same time - a red stripy pattern in front of the right eye and a blue stripy pattern in front of the left.

The volunteers wore special goggles which meant each eye saw only what was put in front of it.

In that situation, the brain then switches awareness between both images, sometimes seeing one image and sometimes the other.

While people's attention switched between the two images, the researchers used fMRI (functional Magnetic Resonance Imaging) brain scanning to monitor activity in the visual cortex.

It was found that focusing on the red or the blue patterns led to specific, and noticeably different, patterns of brain activity.

The fMRI scans could reliably be used to predict which of the images the volunteer was looking at, the researchers found.

Thought-provoking?

The US study, published in Science, took the same theory and applied it to a more everyday example.

They used electrodes placed inside the skull to monitor the responses of brain cells in the auditory cortex of two surgical patients as they watched a clip of "The Good, the Bad and the Ugly".

They used this data to accurately predict the fMRI signals from the brains of another 11 healthy patients who watched the clip while lying in a scanner.

Professor Itzhak Fried, the neurosurgeon who led the research, said: "We were able to tell one part of a scene from another, and we could tell one type of sound from another."

Dr John-Dylan Haynes of the UCL Institute of Neurology, who led the research, told the BBC News website: "What we need to do now is create something like speech-recognition software, and look at which parts of the brain are specifically active in a person."

He said the study's findings proved the principle that fMRI scans could "read thoughts", but he said it was a very long way from creating a machine which could read anyone's mind.

But Dr Haynes said: "We could tell from a very limited subset of possible things the person is possibly seeing."

"One day, someone will come up with a machine in a baseball cap.

"Then it really could be helpful in everyday applications."

He added: "Our study represents an important but very early stage step towards eventually building a machine that can track a person's consciousness on a second-by-second basis.

"These findings could be used to help develop or improve devices that help paralyzed people communicate through measurements of their brain activity.

But he stressed: "We are still a long way off from developing a universal mind-reading machine."

Dr Fried said: "It has been known that different areas of the temporal lobe are activated by faces, or houses.

"This UCL finding means it is not necessary to use strikingly different stimuli to tell what is activating areas of the brain."