fMRI reveals brain’s parallel processing structure

MIT Technology Review : The brain is often referred to as a parallel computer because it can run many different processes simultaneously. Harris Georgiou of the National Kapodistrian University of Athens in Greece and his colleagues have now used functional magnetic resonance imaging (fMRI) to determine how many independent processes are running and on what scale. In Georgiou’s experiment, the fMRI scanner represented the brain as a 60 × 60 × 30 grid of 3D voxels, with each voxel embodying roughly 3 million neurons and each neuron having tens of thousands of connections to its neighbors. In comparison, current attempts to model brain functionality use computer chips containing only about 1 million neurons, each with just 256 connections. Using oxygenation in each voxel as a measure of neuronal activity, Georgiou’s group studied the brains of test subjects performing visuo-motor and reasoning activities of differing difficulties. The fMRI revealed that complex visuo-motor tasks activated roughly 50 sections of the brain at a level of structure above that of individual neurons.