The mindreading software that can listen to the ‘voices in your head’...

The mindreading software that can listen to the ‘voices in your head’ - and could let the paralysed speak again

0 935

Scientists believe they have found a way to read our minds. They have created a computer program that can decode brain activity that creates the ‘voice in our head’ and put it into words. The breakthrough could give the ‘locked in’ or paralysed hope that they could one day communicate using the system. ‘If you’re reading text in a newspaper or book, you hear a voice in your own head,’ Brian Pasley told New Scientist .

‘We’re trying to decode the brain activity related to that voice to create a medical prosthesis that can allow someone who is paralysed or locked in to speak.’

The team conducted their first experiments in 2011, and are currently looking at patients who suffer from Aphasia, a condition that robs you of the ability to communicate. Aphasia can affect your ability to express and understand language, both verbal and written, and typically occurs suddenly after a stroke or a head injury. In their first experiments, the team recorded the brain activity of seven people undergoing epilepsy surgery while they looked at a screen displaying the nusery rhyme Humpty Dumpty, the Gettysbury Address or the inaugural speech of President John F Kennedy. Their brain activity was monitored as they read aloud the text and when they read it silently in their heads. From the spoken data the team managed to build a personal ‘decoder’ for each patient which interpreted the information and turned into a visual representation. They then applied the decoder to brain activity during silent reading and found that they could reconstruct several words that were being thought just through neural imaging alone. The reseachers also tested the decode and algorithm with Pink Floyd sons to see which neurons respond to different musical notes. Although, at an early stage, the team is hopeful that eventually it could be used to monitor what people are thinking when they can no longer speak. They say it could offer a lifeline to those whose speech has been affected by stroke or degenerative disease, but many will be concerned about the implications of a technique that can eavesdrop on thoughts and reproduce them.

In the earlier experiment, neuroscientists at the University of California Berkeley put electrodes inside the skulls of brain surgery patients to monitor information from their temporal lobe, which is involved in the processing of speech and images. As the patient listened to someone speaking, a computer program analysed how the brain processed and reproduced the words they had heard. The scientists believe the technique could also be used to read and report what they were thinking of saying next. In the journal PLoS Biology, they wrote that it takes attempts at mind reading to ‘a whole new level’. Robert Knight, professor of psychology and neuroscience, added: ‘This is huge for patients who have damage to their speech mechanisms because of a stroke or Lou Gehrig’s [motor neurone] disease and can’t speak.

HOW THEY DO IT

In their 2011 study, the team recorded the brain activity of seven people undergoing epilepsy surgery while they looked at a screen displaying the nusery rhyme Humpty Dumpty, the Gettysbury Address or the inaugural speech of President John F Kennedy. Their brain activity was monitored as they read aloud the text and when they read it silently in their heads. From the spoken data the team managed to build a personal ‘decoder’ for each patient which interpreted the information and turned into a visual representation. They then applied the decoder to brain activity during silent reading and found that they could reconstruct several words that were being thought just through neural imaging alone.

‘If you could eventually reconstruct imagined conversations from brain activity, thousands could benefit.’ In the first study in 2012, researchers tested 15 people who were already undergoing brain surgery to treat epilepsy or brain tumours. They agreed to have up to 256 electrodes put on to the brain surface, as they listened to men and women saying individual words including nouns, verbs and names. A computer programme analysed the activity from the electrodes, and reproduced the word they had heard or something very similar to it at the first attempt. Co-author Brian Pasley said there is already mounting evidence that ‘perception and imagery may be pretty similar in the brain’. Therefore with more work, brain recordings could allow scientists to ‘synthesise the actual sound a person is thinking, or just write out the words with a type of interface device.’ Their study also shows in sharp relief how the auditory system breaks down sound into its individual frequencies - a range of around 1 to 8,000 Hertz for human speech.