Page 137 - Handbook of Biomechatronics

134 Domen Novak

generated a P300), the procedure can be repeated until the system is suffi-
ciently certain of the correct command.
The P300 requires no training to utilize, but has a lower information
transfer rate than SSVEPs in state-of-the-art BCIs, 20–25bits/min com-
pared to 60–100bits/min with SSVEPs (Nicolas-Alonso and Gomez-Gil,
2012). Again, false positives are problematic, as P300 responses also occur
naturally in the absence of visual stimuli. Furthermore, the P300 suffers from
the same disadvantage as the SSVEP: a screen must be used to present the
stimuli.

Motor Imagery
Unlike SSVEPs and the P300, motor imagery has the advantage that no
devices or other external stimuli are required for it. Its principle is simple:
the user thinks of making a motion, and the activity of the motor cortex
changes as a result of the imagined motion even if no movement is actually
performed. This activity can be measured and used to control
biomechatronic devices. For example, imagined left-arm movement could
be used to move the left arm of a full-body exoskeleton. However, effective
use of motor imagery requires special user training, and only a small number
of motor images can be distinguished using EEG (Nicolas-Alonso and
Gomez-Gil, 2012). For example, the user may be able to select whether
to move the left or right arm of an exoskeleton, but would not be able to
choose the specific movement that should be performed with that arm.

Mental Imagery
Mental imagery is similar to motor imagery, but instead of imagining
motions, the user performs different types of cognitive activities: mental sub-
traction, auditory imagery, spatial navigation, etc. (Friedrich et al., 2012)As
the frequency distribution of the EEG changes depending on the user’s
mental workload (Herrmann et al., 2004; Antonenko et al., 2010), BCIs
can use this information to determine whether or not the user is performing
a certain cognitive activity. Furthermore, since different cognitive activities
are connected with different regions of the brain (e.g., frontal regions for
mental subtraction), it is possible to differentiate between them using
EEG recorded from different regions. By programming the BCI to perform
specific commands in response to specific mental imagery (e.g., start moving
a wheelchair if mental subtraction is detected), we can thus allow users to
control biomechatronic devices through different cognitive activities.