As fMRI permits us to witness brain activity, the barrier between subjective and objective knowledge is beginning to shift.

"There are elements of human experience that we all know are there, but we haven't been able to measure," says Dr. John Gabrieli, a Stanford neuroscientist. "But that doesn't mean they're not real. For every psychological thing you can think of, I believe you can find a correlative function in the brain if you measure it right."

In one novel experiment carried out in Kanwisher's lab, graduate student Frank Tong asked subjects to wear glasses that showed a different image to each eye. One eye saw a picture of a face, and the other a picture of a place. Because of a widely studied phenomenon called "binocular rivalry," the brain will only process one of these pictures at a time, but it randomly switches back and forth from one to another. Anyone who's looked at ambiguous figures such as "Old Woman or Young Woman?" may have experienced a similar phenomenon.

While the MRI machine recorded brain activity, the subjects in Tong's study pressed a button to indicate when the image switched for them. The active regions of the brain switched at the same time, physiologically verifying a phenomenon only the subjects could experience.

The benefits of viewing the brain this way—of translating the subjective into the objective—are potentially enormous to many fields. In education, for example, most decisions about methodology are based on behavioral outcomes. Reading instruction is a case in point. According to standardized tests, some children do better when they are taught phonics, learning to sound out letters and letter groups. Other students do better when they are taught to recognize individual words, seeing "dog" as a word, instead of sounding out "duh-aw-guh." This makes sense, because one person's brain can be very different from another's, stronger and weaker in different areas. Currently, the only way to tell how a child learns best is to wait a year or more to see if he or she falls behind at learning something one way, and then try to teach it another way.

What if we had a better way than standardized tests (which are really just behavioral output) to measure learning? Gabrieli and his colleagues have begun using fMRI technology to study how we learn reading skills. During the experiment, adult subjects are given mirror-reversed text to read. They read it very slowly, laboriously, often confusing "b" and "d," or "p" and "q." At first, their right parietal cortex, the brain region associated with spatial processing, is very active. After they've had a couple of weeks to practice, they return to the machine and can read better, though still laboriously. But now, their left temporal cortex, the classic language cortex, is active, just as it would be if they were reading normally. With further experimentation, researchers may get a good idea of what learning "looks" like, enabling them to measure it in children far earlier than standardized tests allow.