By making two changes to how brain scans are conducted and read, Stanford researchers have considerably broadened their usefulness. Not only could they tell what type of mental activity their subjects were engaged in – essentially what they were thinking – the changes should now allow scanning of a much wider group of individuals than was possible in the past.

Human thought doesn't seem to arise from activity in single brain regions. Coordinated activity between different regions – networking – plays a vital role in mental processes.

The researchers eliminated the need to know exactly when a subject begins concentrating on a specific mental task. And when reading scans, they looked at the activity of networks in the brain instead of small, individual brain regions.

Standard fMRI (functional magnetic resonance imaging) scans show when parts of the brain become more active by recording changes in blood flow and chemistry. Typically, scan interpretations focus on the one or few brain regions that are most active when subjects are engaged in a particular mental task. But human thought doesn't seem to arise from activity in single brain regions. Coordinated activity between different regions – networking – plays a vital role in mental processes. Networks consisting of 2-12 different brain regions working in concert have been identified as necessary for language, vision, hearing, emotion and other cognitive processes.

So the researchers didn't focus on what was occurring in individual brain regions of their subjects. They looked at how active combinations (networks) of 90 different brain regions were.

Standard fMRI scans also require on versus off comparisons. To interpret them, researchers have always needed to know exactly when a subject began a mental task, such as solving a math problem. It's not always possible to find this out. And human brains don't seem to operate in an on/off manner. That's how computers work.

So the Stanford researchers allowed their subjects to think in a more normal, free-flowing manner while they were scanned. Subjects were told to engage in one of four different mental activities, at their own pace, during scans that lasted anywhere from 30 seconds to 10 minutes. The activities were recalling the day's events, counting backwards from 5,000 by threes, singing song lyrics (silently) or just relaxing.

By reviewing four 10-minute scans (one for each activity) of the 14 volunteers, the researchers were able to see what networking patterns were created by each mental activity. The researchers then gave 10 new subjects similar scans.

They were able to pinpoint with 85% accuracy which of the four activities the new subjects were engaged in by looking at a scan. And when the subjects were given a fifth task, imagining walking through the rooms of their home, it was readily apparent from a scan that the subjects were thinking of something completely different than the other four tasks.

Scans of one minute or longer all could be interpreted with 80% or higher accuracy.

The four mental tasks that were studied are just the start. The researchers believe that their approach will allow decoding of a host of other mental states from short brain scans.

The new method should also help in the diagnosis and treatment of people with cognitive disorders. It's very hard to have such people follow the precise instructions required for typical on/off fMRI scans to be useful. Now, they can simply be scanned, with the scan results offering information on which networks in their brains are functioning abnormally.

The research was published online in the journal Cerebral Cortex on May 26, 2011, ahead of print.