Brain–Computer Interfaces - Index of

158 N. Birbaumer and P. Sauseng
3 Brain–Computer Interfaces for Communication in Complete
Paralysis
The author’s laboratory was the first to apply an EEG–brain–computer interface to
communication in the completely paralyzed [4]. This research was based on earlier
work from our laboratory showing amazing and consistent and long-lasting control
of brain activity in intractable epilepsy [16]. Some of these patients learned
to increase and decrease the amplitude of their slow cortical potentials (SCP) with
a very high success rate (from 90 to 100%) and they were able to keep that skill
stable over more than a year without any intermediate training. If voluntary control
over one’s own brain activity is possible after relatively short training periods
ranging from 20 to 50 sessions as demonstrated in these studies (see [4]), selection
of letters from a computer menu with slow cortical potential control seems
to be possible. The first two patients reported in [5] learned over several weeks
and months sufficient control of their slow brain potentials and were able to select
letters and write words with an average speed of one letter per minute by using
their brain potentials only. These patients suffered from amyotrophic lateral sclerosis
(ALS) in an advanced stage, having only unreliable motor twitches of eyes
or mouth muscles left for communication. Both patients were artificially ventilated
and fed but had some limited and unreliable motor control. After this first
report in the last 10 years, 37 patients with amyotrophic lateral sclerosis at different
stage of their disease were trained in our laboratory to use the BCI to select
letters and words from a spelling menu in a PC. Comparing the performance of
patients with a different degree of physical restrictions from moderate paralysis to
the locked-in state, we showed that there is no significant difference in BCI performance
in the different stages of the disease, indicating that even patients with
locked-in syndrome are able to learn an EEG-based brain–computer interface and
use it for communication. These studies, however, also showed that the remaining
7 patients suffering from a complete locked-in state without any remaining muscle
twitch or any other motor control who started BCI training after entering the
completely locked-in state were unable to learn voluntary brain control or voluntary
control over any other bodily function: limited success was reported for a Ph-value
based communication device using the Ph-value of saliva during mental imagery
as a basis for “yes” and “no” communication [37]. Saliva, skin conductance, EMG
and cardiovascular functions are kept constant or are severely disturbed in end-stage
amyotrophic lateral sclerosis and therefore provide no reliable basis for communication
devices. Thus, brain–computer interfaces need an already established learned
voluntary control of the brain activity before the patient enters the complete lockedin
state where no interaction with the outside environment is possible any more.
On the basis of these data, the first author developed the hypothesis which was
later termed “goal directed thought extinction hypothesis”: all directed, outputoriented
thoughts and imagery extinguishes in the complete locked-in state because
no reliable contingencies (response reward sequences) exist in the environment of
a completely locked-in patient. Any particular thoughts related to a particular outcome
(“I would like to be turned around”, “I would like my saliva to be sucked out of