Esquire magazine (of all places) has an excellent neuroscience article that discusses the case of Erik Ramsey, a young man with locked-in syndrome whose only hope for communicating with the outside world is a prototype brain computer interface that needs to be implanted directly into his cortex.
Locked-in syndrome is a condition that can occur after certain forms of brain stem stroke. The brain stem acts as the relay station to the peripheral nerves of the body, and hence the control of muscles.
The syndrome is where the person is mentally fine but are physically unable to move any muscle in the body, usually except muscles associated with eyes.
Current methods of communicating typically involve having someone hold up a board with the letters of the alphabet on it. The assistant starts reading off the letters and the locked-in person moves their eye when they arrive at the right letter, and through this method, they slowly spell out sentences.
Famously, Jean-Dominique Bauby created one of the most incredible books ever written, The Diving Bell and the Butterfly, using this method after becoming locked-in.
There is currently a great hope for ‘brain computer interfaces’ that, with a bit of training, might allow locked-in people to communicate through a computer system that translates specific types of neural activity into letters or words.
This is usually described as translating ‘thoughts into words’ but most systems simply link specific patterns of brain activation to specific computer outputs, so as long as they can reliably distinguish between different types of brain activity and reliably produce specific responses the job is done.
In other words, if thinking of sea lions reliably produces an ‘A’ and thinking of a scratch-my-nose action reliably produces a ‘B’ (and so on) this is enough, but the leap between the content of thoughts and the output is not at the level of meaning (where thinking of sea lions would produce ‘sea lions’ as an output).
Interestingly, the training method most of these systems use largely relies on operant conditioning (a type of trial and error learning). We know that we can be conditioned to have certain responses unconsciously, so it may be the case that people using the system don’t ‘feel’ like their thinking about something specific for any particular response. Eventually it just ‘happens’, like driving a car.
The researcher behind the system described in the Esquire article is neuroscientist Phillip Kennedy who was recently featured in an excellent article from the Dana Foundation on his work.