Much of what science knows about the human brain has come through deduction. If a stroke or trauma has destroyed a particular are, researchers can look at what that person can no longer do - talk, move the left pinky, do math - and infer that the affected region is linked to that behavior. In animal models, researchers often produce lesions artificially, or they inject a drug to inhibit or excite neural activity in a specific area. Yet as important as this approach has been, there are many things it can't accomplish. Chief among those is pin-pointing which of the many kinds of cells in a given brain region are the ones that matter.
As a result, when it comes to autism, Alzheimer's disease and a long list of mental illnesses, what we do understand is dwarfed by all that we can only imagine. Treatments, too, are often a matter of trial and error. To try to prevent intractable epileptic seizures, for example, surgeons may destroy a part of the brain they believe is implicated. Often it works; sometimes it doesn't. Understanding aberrant behavior, obsessive thoughts, learning disabilities, depression, anxiety, aggression - for all of those, the learning curve remains very steep.
Much of the problem stems from the brain's sheer complexity. "It's made of 100 billion interconnected cels, which fall into many distinct classes - differing by shape, molecular composition and function - and which change in different ways in different brain disorders," says Edward Boyden, a neuroscientist and biological engineer in the MIT Media Lab, whose goal is to develop technologies for "fixing the broken brain."