Multi Voxel Pattern Recognition

Here's an interesting article from Discover about Ken Norman and a bit about Frank Tong's fMRI pattern recognition research. Ken has an article coming out in TICS about this - looks interesting.

"MVPA is asking a different question," said Kenneth Norman, lead author of the paper. "Instead of asking, ‘What does this little piece of brain do?’ MVPA looks at lots of voxels at once and asks, ‘What information is represented in the pattern of activity distributed across the brain?’"

Norman, an assistant professor of psychology at Princeton University, explained to Discovery News that pattern classification and data mining techniques used for handwriting recognition are now being applied to functional magnetic resonance imaging of the brain.

For example, a researcher may ask someone to think about certain faces, locations and objects. The researcher then records the brain patterns associated with each thought. When subjects are asked to repeat the exercise, computer software can match the prerecorded brain patterns to the current ones, which may tell the classifier what the person is thinking.

Researchers can also see certain thoughts developing, even before they become full-blown, conscious ideas. Around 5.4 seconds before a person recalls something seen, heard, or experienced earlier in the study, category-specific patterns of brain activity start to emerge, which help investigators predict what the person is going to remember.