Information Technology Reference
In-Depth Information
13.3 Mental States Used in Passive BCI
Brain activity measurements, speci
cally EEG techniques, provide signals that can
be used to read out user mental state. These signals and related mental states could
be categorized into signals related to user emotional state, signals related to user
'
s
task, signals related to error potentials, and signal related to user
s mental workload.
In the following section, the EEG markers that could be used to read these different
mental states are described.
'
13.3.1 Emotional State
Some EEG patterns have been shown to be correlated with user
s emotional state
(Onton and Makeig
2009
; Heraz and Frasson
2007
; Molina et al.
2009
). These can
be patterns over time (e.g., EEG rhythms) but emotional state can also in
'
uence
punctual patterns (e.g., P300). In the following, we mainly refer to the works
reported in Molina et al. (
2009
).
Lang et al. show that some rhythmic EEG patterns can be correlated to emotional
state; it seems possible to detect dominance, arousal, and pleasure
1
using EEG
signals (Heraz and Frasson
2007
). Lang et al. (
1997
) use the International Affective
Picture System which is a set of pictures that are known to cause speci
c emotions
and measure EEG activity when presenting the pictures to the user. A correlation
between different EEG rhythms and the three emotional states has been observed:
pleasure: negative correlation with delta, theta, and beta rhythms
2
(highly for beta);
arousal: positive correlation with theta and beta rhythms (highly for theta) and
positive correlation with alpha rhythms;
dominance: negative correlation with delta, theta, and beta (highly negative for
beta) and a positive correlation with alpha rhythms.
Numerous works also seem to reveal a relation between frontal activity and
emotion. A difference of symmetry has, for example, been shown between the left
and the right frontal hemisphere in the alpha frequency band during emotion (Coan
and Allen
2004
).
Gamma band and emotion have also been found to be related to emotional
process (Gemignani et al.
2000
, Onton and Makeig
2009
). Studies revealed an
increase in gamma band over left frontal hemisphere during negative emotional
stimuli (Gemignani et al.
2000
). Onton and Makeig (
2009
) also describe a negative
correlation between pleasure of imagined emotion and gamma power in occipital
region.
1
Emotional state can be represented into a three-dimensional space where the axes are pleasure
(from unpleasant to pleasant), arousal (from calm to excited), and dominance (i.e., control) (Lang
et al.
1997
).
2
Beta is considered to be frequency above 12 Hz in this study.
