It will become freaky now, folks. Apart from using an EEG to record brain activity in order to control a BCI, one can, in principle, record electrical activity directly from nerve cells, using implantable electrodes as sensors to target individual cells. This approach has certain advantages (more) and some rather obvious problems, which have to do with the necessity of an implantation.
First experiments with microelectrodes in the brain were performed as early as in the 1960s, but only in recent years microelectrodes became sophisticated enough to produce a stable interface with the brain without moving or causing scar tissue to grow (more).
As far as BCIs are concerned, by now, using a systems with implantable electrodes, monkeys have managed to move cursors on a screen and even operate a robot arm via the internet. In some individual cases, severely handicapped humans with lock-in-syndrome (more) have also received electrode-based systems, generally with encouraging results.

eeg / why is it a language / further nerve cell chatter
firing neurons / implantation / where the future lies
further chatter > neurons


Microelectrodes have improved considerably in recent years. What is commonly used today are so called conic electrodes, which first popped up towards the end of the 1980s.

The idea behind conic electrodes is to provide an interface as stable as possible between the metal and the tissue, i.e. the nerve cells. This is realised by having a hollow structure which contains nerve growth factor, a chemical substance used by organisms naturally to encourage the growth of nerve cells. Boosted by this stimulus, nerve cells grow into the electrodes and form a hybrid structure, resulting in a remarkably stable interface.

In Wessberg’s experiments, the electrodes were implanted for 24 months and still did their job in the end.

Some words about the interface...

Some background science here: Learn about eeg in medicine and why it can de used for BCI-control.