a day ago
Groundbreaking brain chip decodes people's inner monologues in real time
A new brain-computer interface technology can decode a user's inner monologue, an advance that scientists say can help patients with severe speech paralysis communicate their thoughts.
The research, published in Cell, found the new brain-computer interface, or BCI, could successfully decode a user's inner speech on command with up to 74 per cent accuracy.
Scientists hope the new technology can help people who are unable to speak audibly to communicate more easily.
'This is the first time we've managed to understand what brain activity looks like when you just think about speaking,' Erin Kunz, a co-author of the study from Stanford University, said. 'For people with severe speech and motor impairments, BCIs capable of decoding inner speech could help them communicate much more easily and more naturally.'
A BCI uses sensors implanted in brain regions that control movement to decode neural signals and translate them into actions such as operating a prosthetic hand.
Recent research shows that BCIs can even decode attempted speech among people with paralysis.
When users attempt to speak by activating the muscles involved in producing sounds, BCIs can interpret the corresponding brain activity and convert it into text, even if the spoken words are unintelligible.
'If you just have to think about speech instead of actually trying to speak, it's potentially easier and faster for people,' Benyamin Meschede-Krasa, another author of the study, said.
In the latest study, researchers implanted microelectrodes in the motor cortex – the region of brain responsible for speaking – of four participants with severe paralysis from either amyotrophic lateral sclerosis or a brainstem stroke.
Participants were asked to either attempt to speak or imagine saying a set of words. While both attempted speech and inner speech activated overlapping regions in the brain and evoked similar patterns of neural activity, they were also different enough to be reliably distinguished from each other, researchers found.
The latter tended to show a weaker magnitude of activation overall, scientists observed. They then used artificial intelligence to interpret their imagined words.
They demonstrated that their BCI could decode imagined sentences from a vocabulary of up to 125,000 words with an accuracy rate as high as 74 per cent.
The technology could even pick up some inner speech the participants were never told to make, such as numbers when they were asked to tally the pink circles on a screen.
Researchers developed a password-controlled mechanism to prevent the BCI from decoding inner speech unless temporarily unlocked with a chosen keyword.
In an experiment, users could think of the phrase 'chitty chitty bang bang' to begin inner-speech decoding.
The system could recognise this password with more than 98 per cent accuracy, the study found. 'This work gives real hope that speech BCIs can one day restore communication that is as fluent, natural, and comfortable as conversational speech,' Frank Willett, another author of the study, said.
