SIViP (2017) 11:1347–1355
Discrimination between different emotional states based
on the chaotic behavior of galvanic skin responses
· Ataollah Abbasi
· Ateke Goshvarpour
Received: 4 December 2016 / Revised: 25 March 2017 / Accepted: 31 March 2017 / Published online: 18 April 2017
© Springer-Verlag London 2017
Abstract The purpose of the current study was to exam-
ine the effectiveness of galvanic skin responses (GSRs) in
emotion recognition using nonlinear approaches. GSR of 35
healthy students was recorded while subjects were listening
to emotional music clips. The signals were comprehensively
characterized by nonlinear features. Applying three dimen-
sionality reduction methods, including sequential forward
selection (SFS), sequential ﬂoating forward selection, and
random subset feature selection (RSFS) in combination with
four classiﬁcation approaches, including K-nearest neigh-
bor, least-square support vector machine, Fisher discriminant
analysis, and quadratic analysis, discrimination between
emotional classes was evaluated. In addition, two classiﬁ-
cation strategies were examined, including binary (BIC) and
one vs. rest. The results showed that higher recognition rates
were achieved for Fisher. In this case, the BIC accuracy rates
were higher than 99% in all emotional states and all feature
selection methodologies. The maximum classiﬁcation rate
of 99.98% was obtained using RSFS and Fisher in sadness.
Among all emotion categories, better recognition rates were
Computational Neuroscience Laboratory, Department of
Biomedical Engineering, Faculty of Electrical Engineering,
Sahand University of Technology, Tabriz, Iran
Faculty of Electrical and Computer Engineering, University
of Tabriz, Tabriz, Iran
achieved for peacefulness and fear. This study demonstrates
that nonlinear GSR characteristics can provide an informa-
tive measure to investigate the physiological ﬂuctuations in
different emotional states during music.
Keywords Emotion · Galvanic skin responses · Nonlinear
indices · Feature selection · Classiﬁcation
Emotions play an important role in human life and in the
perception, cognition, memory, attention, reasoning, and
decision-making. Accurate emotion recognition is important
for human–computer interaction (HCI), emotional intelli-
gence [1,2], and many other related areas.
Physiological data analysis is one of the most encouraging
approaches for emotion recognition. Galvanic skin response
(GSR) is an electrical activity generates from sweat glands.
GSR is able to show the sympathetic nervoussystem changes.
It also responds to an enhanced arousal level  and gives
a consistent and stable quantity of the involvement of the
participant during music .
Automatic emotion recognition has become an interesting
challenge in the medical ﬁeld [5,6]. Among them, several
efforts have been done based on physiological signals. Time
and frequency features [7,8], as well as wavelet analysis 
are well-known approaches. Due to the chaotic and the non-
stationary behavior of physiological signals, the nonlinear
techniques have been introduced. Recently, many scientists
have also focused on nonlinear biomedical signal analysis in
other medical conditions [10,11].
An increasing number of procedures have been devel-
oped to discriminate emotions from physiological signals.
Investigating the effect of emotional stimuli on electroen-
cephalogram (EEG) signals , the accuracy of 82.37%