Having your voxels and timing them too

The two major noninvasive functional human brain imaging modalities developed during the last part of the twentieth century, high-density scalp EEG (electroencephalogram) and fMRI (functional magnetic resonance imaging), appear from a logical viewpoint to be largely complementary. BOLD (blood oxygen level difference) signals can have a spatial resolution of less than 1 cm3, meaning time series of blood oxygenation level differences can be recorded from many more than 10,000 spatially and structurally identifiable brain regions (voxels). However, changes in blood oxygenation are inherently slow, much slower than the firing of individual neurons (circa 1 ms) or tile frequencies at which neural networks tend to synchronize (1-200 Hz or more). EEG signals, on the other hand, can be recorded at sampling rates of a kHz or more per channel, and can thus easily and accurately record cortical potentials throughout their frequency range, if they reflect a sufficient amount or density of synchronous activity within cortex that the summed local fields reach the scalp without canceling one another. It therefore seems easy to argue, as many researchers have, that by concurrently recording EEG and fMRI BOLD signals, researchers could acquire functional brain activity data with both high spatial and high temporal resolution.