Most of the time, we don’t really want other people to know what we’re thinking. When a friend starts spouting conspiracy theories or a relative asks what we think of her new tattoo, it’s just as well that only our soothing platitudes are heard, while the words running through our heads remain unspoken
Outside of science fiction and fantasy, no one has ever been able to reach into another person’s mind and extract those unspoken words. But that may change in the future, because modern technology is making it possible to see what happens in the brain when we hear someone talking—and because this activity is thought to be pretty much the same whether we hear someone say a sentence, or think that sentence ourselves, it may not be long before we are able to turn hidden thoughts into spoken words via computer.
(Sound familiar? Back in September, I wrote about similar work that used brain activity to recreate images people saw.)
New Scientist recently ran an article, written by Helen Thomson, on the work of a research team at the University of California, Berkeley. Led by Brian Pasley, the team presented spoken words and sentences to 15 people undergoing surgery for epilepsy or a brain tumor, while recording neural activity from the surface of a portion of the brain near the ear that’s involved in processing sound. Then they tried to associate different aspects of speech to different kinds of brain activity in the recordings.
The brain breaks down speech in terms of frequency (pitch), frequency fluctuation, rhythm and more. And sure enough, the team was able to correlate many of these aspects of speech to the neural activity they recorded in their subjects’ brains.
Next, they trained a computer program to interpret the neural activity and turn it into a spectrogram, a graphical representation of sound that shows how much of what frequency is occurring over a period of time. To test the spectrograms, they compared the ones they created from neural activity with spectrograms they created from the original sounds.
A second computer program converted the reconstructed spectrogram into audible speech. The result? “Coarse similarities” between the real words and the reconstructed words, says Pasley, that human listeners can kind of pick up on, but which computers were able to analyze more accurately.
Recording brain activity and turning it into spoken language via a computer would have one very obvious and very exciting application: helping those who have lost the ability to speak through paralysis or some other physical problem to once more communicate with the outside world.
Nor is Pasley’s team the only one working toward this goal. At Boston University in Massachusetts, Frank Guenther interprets the brain signals that control the shape of the mouth, lips and larynx to try to figure out what a person is trying to say. So far, all they’ve managed to produce are a few vowel sounds; nothing more complex. But it’s a start.
Steven Laureys at the University of Liege, Belgium, is seeking ways to distinguish brain activity corresponding with “yes” and “no” to help those who cannot speak.
Pasley is anxious to try to develop technology to make the thought-patterns-to-speech translation happen. He’d like to develop safe, wireless, implantable devices suitable for long-term use.
Of course, having anything implanted in your brain is going to be a tough sell, and to begin with, at least, only people already undergoing essential brain surgery would be likely candidates. And the words-from-brain-activity software is still in its infancy, anyway.
But the concept certainly seems viable, and exciting…although I don’t think it’s something most of us would want installed, no matter how safe, cheap or effective it might become.
“It is better to remain silent and be thought a fool than to speak and remove all doubt,” goes an old saying
If the day ever comes when our every passing thought is revealed to the world, the number of people revealed to be fools will surely astound.
Although, now that I think about it, hasn’t that already happened with Twitter?