As reported on The Verge.
By Bryan Bishop
Researchers at the University of California, Berkeley have created a map showing how and where the brain categorizes the actions and objects we see every day. The study began by taking five human subjects and placing them in an MRI machine while they each watched two hours of movie trailers. Each of the trailers had been analyzed to track the different objects and types of motion present — a leaf holding still, or a butterfly flying, for example. The MRI then allowed the researchers to track the blood flow in the brain to see what areas were being stimulated during the screenings.
They then used regression analysis — a statistical technique that is able to determine relationships between different events — to determine how the approximately 30,000 locations in the brain they were tracking matched up to the 1,700 types of objects and actions in the movie trailers. There were commonalities across all the subjects, allowing the researchers to create a map of what they call the brain’s “semantic space.” The researchers note that it’s historically been thought that different categories or types of objects were represented by different regions of the visual cortex. This new kind of mapping, however, revealed a much more nuanced network of overlapping associations, which extended out beyond the visual cortex to cover around 20 percent of the brain.
The expanded map could lead to improved diagnosis techniques and better treatment of brain disorders, they say, and may even be able to serve as a model for improving the way computers recognize and process images. If you’d like to take a look at the semantic maps themselves, the researchers have put an interactive version of the data online that you can peruse right here.