The debate over localization long predates the introduction of fMRI. In the early 19th century, Franz Joseph Gall, the founder of the now-discredited discipline known as phrenology, proposed that the brain comprised distinct functional units whose usage was reflected by bumps in the skull. But when experimental physiologists of that era tried to confirm this notion by studying brain damage in birds, they failed to find specific functional losses.

Later, studies of human strokes, the discovery of neurons, and the beginnings of a distinction between localized symptoms and localized functions continued to fuel arguments about whether brain functions are discrete and easily mapped.

Some researchers think that the truth lies between the two extremes: simpler functions are localized in modules, or specific areas, while more complex ones are distributed. “You have some modularity,” said Yale’s Douglas Rothman, Ph.D. ’87, “but the modularity itself is supported in networks, not in discrete regions completely responsible for complex function.” (Or, as Sally Satel, M.D., and her co-author, Scott O. Lilienfeld, Ph.D., put it, “most neural real estate is zoned for mixed-use development.”)