Mind power: Scientist turns thoughts into actions, literally

The task was simple: Move a computer cursor through the sort of small maze that might challenge a 6-year-old. True, it took three adult volunteers about a half-hour each to complete it, but they had a good excuse: They had no mouse, just the power of their minds.

The experiment, led by Seung-Schik Yoo of Brigham and Women's Hospital, used a super-powerful MRI scanner to analyze the activity in the volunteers' brains and translate it into moves in the maze.

The procedure rests on the concept that different types of thoughts cause activity in different regions of the brain, creating distinct patterns that can be seen by the scanner and linked to different cursor movements.

If the subjects recited ''Mary had a little lamb" in their minds, for example, the cursor would move down. Calculating ''50-3=47" would move the cursor up. If they imagined tapping their right-hand fingers, it would go right; the same for the left.

The process was cumbersome -- it required a multiton, multimillion-dollar machine, and some test subjects had a hard time maintaining the necessary mental focus to move the cursor. ''It's not practical; it's just to show the possibility," Yoo acknowledged.

But the experiment impressed other researchers who are working on ways to hook the brain into a computer that can interpret thoughts and commands.

''This is a new use of a very new technology that is a baby step toward mind-reading, toward being able to read out the cognitive functions of human beings in real time, noninvasively," said Christopher deCharms, a brain imaging researcher at Stanford University.

Several brain researchers noted that Yoo's experiment already could be used by ''locked-in," paralyzed patients whose mental functioning remains intact.

''It would be very crude and rudimentary but it would work, and the hope is that it's a first step toward a future where that kind of technology would be cheaper, better, and faster," deCharms said.

Yoo's mind-reading tool, a ''functional MRI," can track the inner workings of the brain by picking up changes in blood oxygen levels in various parts of the brain.

Yoo believes the experiment, to be published in the July 19 issue of NeuroReport, represents the first time that those changes have been turned into action.

''What we do is translate thought into distinct categories and, amazingly, we can do that," he said. ''We can detect classes of thought."

And that is only the beginning, he said. He said he believes it is possible to detect 20 or more classes of thought, enough to allow a patient to move a cursor on a virtual typewriter.

Yoo and his colleagues began the experiment by doing a bit of brain-mapping on the subjects, determining which areas most ''lit up" during the different kinds of thought.

They then linked the pattern for each type of thought to computer control of the cursor.

Each cursor move the subjects made involved crunching a hefty 12 megabytes of data, Yoo said.

Though the computer processing is likely to keep getting faster, the MRI method's speed is limited by the time it takes to detect changes in blood flow. That likely means it will always take a few seconds for each cursor move, slower than other brain-computer interfaces.

The lag may be acceptable if you have no choice, or just want to play chess, said James T. Voyvodic, a brain imaging researcher at Duke University. But ''given the slowness, expense and inconvenience of functional MRI, it is unlikely to present a serious challenge" to other methods, he stated in an e-mail.

Yoo agrees, at this point. But he believes that his method may complement other brain-computer interface methods. In particular, he said, it could help researchers determine the best places for inserting or attaching electrodes that pick up the brain's electrical signals.

Other brain-computer experiments pose difficulties of their own. A Foxboro company, CyberKinetics Inc., has just begun clinical trials in paralyzed patients of a device that, like Yoo's experiment, is meant to allow direct brain control of a cursor. But it requires hours of surgery to implant electrodes directly into the brain, and such major surgery poses a small risk of hemorrhage or infection.

As for real mind-reading, deCharms said, it is decades away. With something like Yoo's method, he said, ''it's not going to be possible to tell the difference between thinking of your mother and thinking of your car." But, he said, ''I think 50 years from now that will be possible."

Carey Goldberg can be reached at goldberg@globe.com.