Brain–Computer Interfaces - Index of

190 F. Nijboer and U. Broermann
that the chance an organism will perform a particular kind of behaviour increases
when it is rewarded for it, and decreases when it is punished or not rewarded for
it. A good example of operant conditioning in our own life can be observed when
our email program produces its typical sound for incoming email, we rush to the
inbox, because we have been rewarded with the sight of new emails in the past
and we expect to be rewarded again. Our behaviour, the checking of the inbox, will
vanish after we’ve heard the typical sound many times without having the reward
of a new email. When behaviour vanishes, because no reward for the behaviour is
given, psychologists call this extinction. The same principle can now be applied to
Brain–computer interface use. When the user produces brain signals that result in
a successful manipulation of the environment (reward), the likelihood that the user
will produce the brain signal again in future increases.
In 2006, Birbaumer [21] suggested why learning through operant conditioning
might be very difficult, if not impossible, for complete locked-in patients
(see also chapter “Brain–Computer Interface in Neurorehabilitation” in this book).
Birbaumer remembered a series of studies on rats [40] paralyzed with curare. The
rats were artificially ventilated and fed and could not affect their environment at all,
which made it impossible to them to establish a link between their behaviour and
the reward from outside. It was not possible for these rats to learn through operant
conditioning. Birbaumer hypothesized that when CLIS patients can not affect
their environment for extended periods of time, the lack of reward from the environment
might cause extinction of the ability to voluntarily produce brain signals. If
this hypothesis is correct, it supports the idea that BCI training should begin before
loss of muscle control, or as soon as possible afterwards [38]. It should be noted
that no patients have proceeded with BCI training from the locked-in state into the
complete locked-in state. Maybe a LIS patient who is able to control a BCI is very
capable of transferring this ability in the complete locked-in state.
Other ideas have been put forward as to why CLIS patients seem unable to
learn BCI control. Hill and colleagues [38] proposed that it might be very difficult
for a long term paralyzed patient to imagine movements as is required for SMR
self-regulation or that it is impossible to self-regulate SMR, because of physiological
changes due to the disease. Also, general cognitive deficits, lack of attention,
fatigue, lack of motivation or depression have been mentioned as possible factors
that hamper BCI learning [38, 41].
Neumann [42, 43] and Neumann and Kübler [44] describe how attention, motivation
and mood can affect BCI performance of ALS patients. Nijboer and colleagues
also found that mood and motivation influences BCI performance in healthy subjects
[45] and in ALS patients (data is being prepared for publication). It is impossible
to ask CLIS patients how motivated they are to train with the BCI, but data from
communicating ALS patients show that ALS patients are generally more motivated
to perform well with BCI than healthy study participants. Personal experience with
ALS patients and healthy subjects is compatible with this conclusion. ALS patients
who participated in our studies during the past few years were highly intrinsically
motivated, because they see the BCI as a final option when communication with the
muscles is no longer possible. ALS patients were also very willing to participate in
our studies, despite extensive training demands, discomfort of putting the electrode