In observing traces of actual thought processes, we're learning that the brain is much more complicated than any computer we've ever encountered, even those designed to perform many tasks at the same time.

"There's a level on which the brain must have algorithms to figure things out, like computers do," says Gabrieli. "And it must have input and output. But the brain is capable of massive parallel processing on a very high level. Some computers can do that, but not nearly on the same scale as the brain does." And the computer analogy, while useful, ultimately breaks down: almost every brain cell is its own processor, yet no cell can operate independently as a computer chip can, relying instead on neural networks. To an engineer, this may seem like an odd design, but there can be no question that the brain gets the job done very efficiently.

By mapping out functions within the brain and finding out how they overlap, we may begin to understand what mechanisms it uses to break apart tasks. We start to see how our brain transforms us from a metabolic, mobile "machine" into a person, complete with consciousness, creativity, and the ability to learn.

This kind of technology could also help us diagnose mental disorders and abnormalities, paving the way for better medical treatments. Gabrieli's lab has shown that children with attention-deficit/hyperactivity disorder (ADHD), a condition in which people are physically restless, have difficulty controlling their impulses, or have difficulty paying attention, show different fMRI activation patterns from children without the disorder. In one of the tests, young boys were asked to press a button whenever any letter appeared on the screen in front of them except the letter X. ADHD kids had trouble suppressing the impulse to press the button and didn't show increased activity in the regions of the brain that help regulate how we pay attention. Ritalin, the drug commonly prescribed for these children, increased ADHD kids' performance in the test and increased activation in those regions.

"It doesn't matter whether you're looking at a drug's effects or a learning process. By looking at the 'before' and 'after' conditions, we can watch the brain adapt its organization based on new functions," Hirsch says. She and her colleagues are using fMRI to uncover imaging criteria for what makes a good drug for the treatment of chronic pain. By mapping the brains of patients experiencing pain before and after the administration of various drugs, they learn better what to aim for in the development of new pain medication.

Neurosurgeons are also turning to fMRI to help remove previously inoperable brain tumors. Since tumors can often push other structures out of place in the brain, surgeons often will not risk removal for fear of damaging a nearby motor or sensory cortex because they don't know its new boundaries. Surgery might create a worse problem than the risk presented by the tumor.

But by using fMRI, Hirsch's lab has vastly decreased that risk. A patient can be scanned before surgery, mapping every standard function to create an individually-tailored image of the functional organization of that patient's brain. "Neurosurgical planning gives the surgeon a high-resolution roadmap so the operation can be tailored to the individual tumor," she says. The technique has facilitated the successful removal of tumors from over two hundred people.

Over the next decade, as the images become sharper and the black box more transparent, many more people will benefit. Likewise, our understanding of what makes us more than a collection of atoms will improve. Scientists are finally analyzing the brain, not by taking it apart, but by watching it at work.