Information Technology Reference
In-Depth Information
(
2007
) studied emotional reactions to pleasant and unpleasant music induced by
dissimilar piano excerpts. They employed Principal Component Analysis on fMRI
and EEG, and they reported that a left cortical network involved with pleasant
feelings. In contrast, unpleasant emotions involved the activation of the right frontal
and limbic brain areas.
The frontal activation emotion hypothesis has been also tested across different
modalities, age groups, and measures. Davidson and Fox (
1989
) found that
asymmetrical frontal brain activity discriminated sweet and sour tastes in newborns.
Schmidt et al. (
2003
), tested 3-month-old to 12-month-old infants using musical
excerpts of varying valence and arousal (happy, sad, and fear). The authors suggest
that taken together their
findings, which showed an emerging asymmetry of acti-
vation in the presence of an overall decrease of EEG power, indicate maturation of
cortical music processing as well as a
“
calming
”
in
uence of music by the end
of the
first year of life. Baumgartner et al. (
2006
) investigated neural correlates of
sadness, fear, and joy. They observed that auditory information interacts with visual
information in several limbic and paralimbic structures. Activity changes in these
structures were stronger during combined presentation of fearful and joy photo-
graphs with fearful and joy music, compared to when only visual information was
present.
The research on asymmetries of EEG activity mainly focused on the analysis of
alpha band power. Relatively few studies have examined frequency bands other
than alpha including theta, beta, and gamma (Aftanas and Golocheikine
2001
;
Sammler et al.
2007
; Pizzagalli et al.
2002
,
2003
; Flores-Gutierez et al.
2007
). The
results of these studies provide evidence that theta band plays a more important role
in emotion processing that previously believed. Therefore, it is important to
examine other frequency bands than alpha carefully as these may provide additional
information not re
ected in alpha.
6.7.1 EEG and Emotion Recognition
In addition to peripheral physiological responses to music, EEGs from the brain
gained recently great attention for emotion recognition. Estimating the emotion
from EEG is important because brain waves are generated by the brain and are
deeply related to cognition processes. Furthermore, the ongoing brain activity
provides noninvasive measurement with high resolution. It appears that EEGs
provide more insight into emotional processes compared to peripheral CAN signals.
There are an increasing number of studies on EEG-based emotion recognition. In
these studies, different approaches with respect to both feature extraction and
classification algorithms were investigated. Most of the early studies on EEG-based
emotion recognition focused on spectral power changes in few bands and speci
c
brain areas. Power spectra of EEG signals in different frequency bands were used to
examine the relationship between brain activity and emotional states. A common
indicator of musical emotion is the alpha-power asymmetry at the anterior region of
