Brain–Computer Interfaces - Index of

Dynamics of Sensorimotor Oscillations in a Motor Task 57
initial, short-lasting peak was followed by a broad-banded peak within the next
seconds. This later peak is very likely the result of the conscious executed motor
imagery task. An initial recognition peak after visual cue presentation was already
reported in a memorized delay movement experiment with left/right finger and foot
movements [71] and after right/left hand movement imagination [72].
The initial, short-lasting separability peak suggests that the EEG signals display
different spatio-temporal patterns in the investigated imagery tasks in a small time
window of about 500–800 ms length. So, for instance, short-lasting mu and/or beta
ERD patterns at somatotopically specific electrode locations are responsible for the
high classification accuracy of the visually cued motor imagery tasks. The right
side of Fig. 7 presents examples of such ERD maps obtained in one subject, and
the left side shows examples of separability curves for the discrimination between
left vs. right hand, left hand vs. foot and right hand vs. foot motor imagery from
the same subject. The great inter-subject stability and reproducibility of this early
separability peak shows that, in nearly every subject, such a somatotopically-specific
activation (ERD) pattern can be induced by the cue stimulus. In other words, the cue
can induce a short-lived brain state as early as about 300 ms after cue onset. This
process is automatic and probably unconscious, may a type of priming effect, and
could be relevant to a motor imagery-based BCI.
We hypothesize, therefore, that the short-lived brain states probably reflect central
input concerning the motor command for the type of the upcoming MI task.
This may be the result of a certain “motor memory”, triggered by the visual cue. It
has been suggested that such “motor memories” are stored in cortical motor areas
and the cerebellum motor systems [73], and play a role when memories of previous
experiences are retrieved during the MI process [74].
8 Observation of Movement and Sensorimotor Rhythms
There is increasing evidence that observing movements reduces the mu rhythm and
beta oscillations recorded from scalp locations C3 and C4. Gastaut and Bert [75]
and later Cochin [76] and Muthukumavaswauny [77] reported an attenuation of the
central mu rhythm by observation of experimental hand grasp. Altschuler et al. [78]
found that the mu rhythm desynchronized when subjects observed a moving person,
but not when they viewed equivalent non-biological motion such as bouncing balls.
Also, Cochin [76] reported a larger mu and beta power decrease during observation
of a moving person than during observation of flowing water.
Consistent with the reported findings, a previous study in our laboratory [79]
showed that the processing of moving visual stimuli depends on the type of the
moving object. Viewing a moving virtual hand resulted in a stronger desynchronization
of the central beta rhythm than viewing a moving cube (Fig. 8). Moreover,
the presence of an object, indicating a goal-directed action, increases the mu rhythm
suppression compared to meaningless actions [77].
Modulation of sensorimotor brain rhythms in the mu and beta frequency band
during observation of movement has been linked to the activity of the human mirror