Who needs pens? The computer that writes using brain waves

Just last year, the software that gives famous physicist Stephen Hawking a voice received a major upgrade. The scientist, who suffers from ALS (a degeneration of the motor nervous system), used to control his computer with a single cheek muscle whose movements were registered by an infrared sensor on his glasses. In this way, he painstakingly constructed each sentence letter by letter, word by word, typing one or two words per minute. And that’s only because he perfected the technique over the years. An incredible test of patience.

In late 2014, a new program was developed using data from years of analyzing Hawking’s speech. Suddenly, the physicist was able to express himself ten times faster. The reason: The new software recognizes complex correlations and can predict Hawking’s language by analyzing countless data points. A true technological revolution – not only for him, but potentially also for many other ALS patients.l

Brain waves converted to text

But this cutting-edge software could soon be replaced by another groundbreaking invention that scientists have been discussing for a long time: direct communication between humans and machines. Brain-to-Text is the name of a new software developed by German and American scientists that will soon make it possible to write using only the power of thought.

What we can normally do with ease through our vocal cords or our hands is now done by brain waves – electrical signals from the brain that are sent out, for example, when processing sensory impressions and also when speaking and thinking. Modern technology can read these streams, decode them and then reconstruct them into sounds, words and whole sentences.

Epileptics are valuable subjects

The software was tested in a study of seven people. The number of participants was small as the study required them to have special electrodes implanted in their brains. This is because the caps used to measure electrical currents are not yet sensitive enough to detect them correctly through the skull. As a result, implanted amplifying electrodes were needed – and seven participants, all of whom had epilepsy, possessed them. They were originally designed to indicate where seizures were occurring and where surgery might be needed.

During the study, participants were asked to read various passages aloud while the computer simultaneously processed the neurodata. The software’s task was to associate each sound with the corresponding brain waves, and the program’s sophisticated, self-learning algorithms became even more reliable over time. In the study, Brain-to-Text worked only on audible, spoken language – but the results are an important first step toward recognizing thought language.

Promising results

Given that the study was based on only seven participants but still provided reliable data, the results are very promising, said Peter Brunner of Albany Medical College. Speech recognition programs typically rely on several thousand hours of audio data to deliver reliable results.

So will pens become obsolete in the future? Will authors write their works in a few days instead of many weeks? Will brainwave-to-text software write down all thoughts categorically or will it be possible to set a start and end point? Will people like Stephen Hawking soon be able to communicate almost instantaneously? And if so, will the software required to do so be available at affordable prices? On the one hand, the fact that so many people are asking these questions suggests that the project is still in its infancy. On the other hand, it points to the great potential that lies within.