Zurich Open Repository and

Archive
University of Zurich
University Library
Strickhofstrasse 39
CH-8057 Zurich
www.zora.uzh.ch

Year: 2018

The neural underpinnings of music listening under different attention conditions

Jäncke, Lutz ; Leipold, Simon ; Burkhard, Anja

DOI: https://doi.org/10.1097/WNR.0000000000001019

Posted at the Zurich Open Repository and Archive, University of Zurich

ZORA URL: https://doi.org/10.5167/uzh-151056
Journal Article
Published Version

Originally published at:

Jäncke, Lutz; Leipold, Simon; Burkhard, Anja (2018). The neural underpinnings of music listening under different
attention conditions. NeuroReport, 29(7):594-604.
DOI: https://doi.org/10.1097/WNR.0000000000001019
594 Integrative systems

The neural underpinnings of music listening under different

attention conditions
Lutz Jänckea,b,c,d, Simon Leipolda and Anja Burkharda

Most studies examining the neural underpinnings of music piece during both listening conditions. Thus, the focus of
listening have no specific instruction on how to process the attention had a strong influence on the neurophysiological
presented musical pieces. In this study, we explicitly responses to the pop song, but not on the responses to the
manipulated the participants’ focus of attention while they classical musical piece. The electroencephalographic
listened to the musical pieces. We used an ecologically valid responses during passive listening to the pop song are
experimental setting by presenting the musical stimuli interpreted as a neurophysiological and psychological state
simultaneously with naturalistic film sequences. In one typically observed when the participants are ‘drawn into the
condition, the participants were instructed to focus their music’. NeuroReport 29:594–604 Copyright © 2018
attention on the musical piece (attentive listening), whereas Wolters Kluwer Health, Inc. All rights reserved.
in the second condition, the participants directed their NeuroReport 2018, 29:594–604
attention to the film sequence (passive listening). We used
two instrumental musical pieces: an electronic pop song, Keywords: attention, electroencephalographic oscillations,
heart rate variability, music listening
which was a major hit at the time of testing, and a classical
a
musical piece. During music presentation, we measured Division of Neuropsychology, Department of Psychology, bInternational Normal
Aging and Plasticity Imaging Center (INAPIC), cUniversity Research Priority
electroencephalographic oscillations and responses from Program (URPP), Dynamic of Healthy Aging, University of Zurich, Zurich,
the autonomic nervous system (heart rate and high- Switzerland and dDepartment of Special Education, King Abdulaziz University,
Jeddah, Kingdom of Saudi Arabia
frequency heart rate variability). During passive listening to
the pop song, we found strong event-related Correspondence to Lutz Jäncke, PhD, Division of Neuropsychology, Department
of Psychology, University of Zurich, Binzmuehlestrasse 14/Box 25, CH-8050
synchronizations in all analyzed frequency bands (theta, Zurich, Switzerland
lower alpha, upper alpha, lower beta, and upper beta). The Tel: + 41 44 635 7557; fax: + 41 44 635 7409; e-mail: [email protected]

neurophysiological responses during attentive listening to Received 13 February 2018 accepted 5 March 2018
the pop song were similar to those of the classical musical

Introduction influence the brain activation in response to the incoming

During the past 30 years, many studies have been published stimuli [23–26]. In addition, a few music listening studies
examining the neural underpinnings of music listening. have shown that brain activations during music listening
Functional MRI and PET studies have shown that music depend on context effects [27,28]. Thus, there is a need for
listening is associated with hemodynamic responses in a the study of the neural underpinnings of music listening
distributed network. This network comprises the auditory under different listening conditions. To examine whether
cortex (as a core region with the primary and the secondary attentive or passive listening to music induces different
auditory cortex), brain regions associated with the control responses in the EEG and the autonomic nervous system
of attention (e.g. superior and inferior parietal cortex as well (ANS), we designed this exploratory experiment by
as areas in the frontal pole), memory (e.g. hippocampus, measuring EEG oscillations, heart rate (HR), and high-
medial, and inferior temporal cortex), working memory frequency heart rate variability (HF-HRV) during different
(e.g. frontal cortex), emotion (limbic system including the music listening conditions. The musical pieces were pre-
nucleus accumbens), and motor control (e.g. supplementary sented simultaneously with naturalistic film sequences. In
motor area and presupplementary motor area) [1–7]. Several one condition, the participants focused their attention on the
electroencephalography (EEG) studies have complemented music (attentive listening), whereas in the other condition,
these findings, either by estimating the intracortical sources the focus of attention was directed to the simultaneously
on the basis of surface EEG measures [8–10], by uncovering presented film sequence (passive listening). Furthermore,
specific functional network characteristics during music lis- we examined whether musical pieces of two different
tening [11–20], or by reporting increased alpha band power genres evoke diverging EEG and ANS responses in these
during musical imagery and listening [21,22]. Most of listening conditions. We used two instrumental musical
the neuroscientific studies investigating music listening pieces, one of which was well known to the participants
presented the music without specific instructions. However, because it was a major hit at the time of testing and the other
it is well known from studies using tightly controlled a classical instrumental piece that has been used in many
auditory stimuli that the direction and quality of attention experiments, but was not well known to the participants.
0959-4965 Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved. DOI: 10.1097/WNR.0000000000001019

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

 Music listening and attention Jäncke et al. 595

As this study was designed explicitly as an exploratory study, last 2000 ms of each musical piece. The two musical
we only defined general study questions and no specific pieces were chosen because they differ in several
hypotheses. aspects including acoustic features and publicity level.
The musical piece Sonnentanz is a well-known musi-
(1) Are the different listening conditions (attentive vs. cal piece, which was on the Swiss hit charts for several
passive) associated with different EEG oscillatory months, whereas the Mozart piece was not well known
activation patterns? to most of the examined participants. In addition, both
(2) Are the EEG responses different for the two musical pieces differ markedly in terms of major acoustic
pieces, especially in the context of the different features (intensity variation: Sonnentanz = 55%, KV
listening conditions? 448 = 70%; spectral entropy: Sonnentanz = 0.71,
(3) Are the different listening conditions (attentive vs. KV 448 = 0.64; spectral flatness: Sonnentanz = 0.20,
passive) associated with different ANS responses? KV 448 = 0.014). For the calculation of these acoustic
features, we used the toolbox seewave (version 2.0.5,
http://rug.mnhn.fr/seewave) in R. The two visual stimuli
Participants and methods were taken from a YouTube video showing a train ride
Sample from the driver’s perspective. The preprocessing of
Thirty individuals (25 women and five men) took part in the visual stimuli was performed using iMovie soft-
the study. The participants were students enrolled in ware (version 10.1, http://www.apple.com/de/imovie/) and
Psychology at the University of Zurich or in Arts at the Windows Movie Maker 2012. First, the audio track of
Zurich University of the Arts. The mean age of the par- the video was removed. Second, the video was deco-
ticipants was 22.2 years (range = 18–33 years). All parti- lorized. Third, two nonoverlapping sequences of 4 min
cipants were consistently right-handed according to the each were taken from the video, and finally, a visual
Annett Handedness Questionnaire [29]. None of the fade-in effect was applied to the first 1500 ms of each
participants reported any present or past psychiatric, sequence.
neurological, or audiological disorders. All participants
denied consumption of illegal drugs and regular medi-
cation. The participants’ history of musical training was Experimental procedure
assessed using an in-house questionnaire. Twenty-eight The participants were seated in a comfortable chair in
of the participants played an instrument or took singing a sound-shielded room with electromagnetic isolation.
lessons during childhood or youth. On average, those Before the experiment, 8 min of resting-state EEG
participants started playing when they were 8.3 years old and electrocardiography (ECG) were acquired: 4 min
and continued to play regularly for 7.3 years. Twenty-six with eyes open (EO) and 4 min with eyes closed. For
of the participants reported listening to pop music and the EO condition, participants were instructed to look
11 to classical music. The participants received course at a fixation cross, relax, and let their mind wander.
credits or monetary compensation for participation. The EO condition was used as a baseline condition for
Written informed consent was obtained from the parti- the subsequent experimental conditions. After the
cipants before participation. The study was approved by resting state, participants performed a task in which
the Ethics Committee of the University of Zurich and they simultaneously listened to musical pieces and
carried out following the principles of the Declaration of watched film sequences while EEG and ECG were
Helsinki. acquired. The musical stimuli were presented using
in-ear headphones (Sennheiser, Wedemark, Germany)
at a comfortable sound pressure level. The film stimuli
Stimulus material
were presented on a 19′ screen (Belinea, Wittmund,
The stimuli consisted of two musical pieces from differ-
Germany). Stimulus presentation was controlled by
ent genres (classical and electronic pop music) and two
Presentation software (version 17.1, http://www.neu
naturalistic film sequences. The two musical stimuli
robs.com/).
were (i) Sonate für zwei Klaviere in D-Dur, Allegro
con spirito, KV 448 (375a) by Wolfgang A. Mozart, The experiment comprised four conditions, which were
performed by Yaara Tal & Andreas Groethuysen and presented in a randomized order. In the Mozart-Attentive
(ii) Sonnentanz by the electronic pop music group condition, the musical piece by Mozart was presented
Klangkarussell. The preprocessing of the auditory simultaneously with one of the film sequences for
stimuli was performed using Audacity software (ver- 4 min, and the participants were instructed to attend
sion 1.3.14 Beta, http://www.audacity.de/ ) and included the music by counting the occurrence of certain
the following steps: first, the amplitude of the musical musical aspects (pauses, changes in loudness). In the
pieces was normalized. Second, only the first 4 min of Klangkarussell-Attentive condition, the musical piece
each musical piece was kept, with the rest of the piece by Klangkarussell was presented simultaneously with
being cut out. Third, a linear fade-in was applied to the other film sequence (i.e. the sequence that was
the first 500 ms and a linear fade-out was applied to the not used in the Mozart-Attentive condition). The

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

596 NeuroReport 2018, Vol 29 No 7

instructions for the Klangkarussell-Attentive condition baseline condition. Then, the data were again seg-
were the same as in the Mozart-Attentive condition. In mented into segments of 2 s length.
the Mozart-Passive condition, the physical stimulation
was identical to the Mozart-Attentive condition.
However, in the Mozart-Passive condition, the parti- Electroencephalography data analysis
cipants were instructed to focus their attention on Spectral analysis of the EEG data was carried out in Brain
the film sequence by counting the occurrence of cer- Vision Analyzer 2. For each 2 s segment, a frequency
tain visual aspects (railway signals, railway switches). spectrum was calculated by a fast Fourier transformation,
The physical stimulation of the Klangkarussell-Passive in which the segments were tapered with a Hanning
condition was identical to the Klangkarussell-Attentive window. The frequency resolution of the resulting
condition, with the instructions being the same as in spectra was 0.5 Hz. Next, the frequency spectra were
the Mozart-Passive condition. The combination of the averaged within each condition. For each electrode, this
musical pieces with the specific film sequence was fixed resulted in one frequency spectrum per experimental
within each participant, but counterbalanced across parti- condition and one additional spectrum for the baseline
cipants. The instructions were explained to the participants condition. Subsequently, the frequency spectra were
in detail before the experiment. After each condition, the combined spatially by averaging over electrodes to form
participants rated the valence and arousal induced by the the following electrode clusters (labels of the electrodes
musical pieces on a seven-point Likert scale. In addition, given in HydroCel GSN 130 terminology): frontal = 4, 5,
the participants rated the subjectively experienced diffi- 11, 12, 19, 20, 24, 28, 117, 118, 124; central = Cz, 7, 30,
culty to focus their attention on a seven-point Likert scale. 31, 36, 37, 55, 80, 87, 104, 105, 106; parietal = 52, 53, 60,
61, 62, 67, 72, 77, 78, 85, 86, 92; and occipital = 69, 70, 74,
75, 82, 83, 89. From the spectra of these electrode
Electroencephalography data acquisition and clusters, the mean amplitude of the following frequency
preprocessing bands was extracted: theta = 4–7 Hz, lower alpha =
EEG data were recorded using a high-density electrode 7.5–10 Hz, upper alpha = 10.5–12.5 Hz, lower beta = 13
net with 128 channels (HydroCel GSN 130; Electrical –16 Hz, and upper beta = 16.5–20 Hz.
Geodesics, Eugene, Oregon, USA) in combination with
an EEG amplifier (Net Amps 300; Electrical Geodesics) Following the spectral analysis, event-related synchroni-
and Net Station software (version 4.4, http://www.egi.com/ zation (ERS) and event-related desynchronization (ERD)
clinical-division/net-station). The sampling rate of the EEG values were computed for the amplitudes in each fre-
data was 500 Hz, electrode Cz was used as the online quency band and for each experimental condition [28,31].
reference electrode, and an online bandpass filter of The following formula was used for calculation:

ðAmplitude in experimental condition Amplitude in baseline conditionÞ  100

ERS=ERD ¼
Amplitude in baseline condition

0.1–100 Hz was applied. Electrode impedance was main- The ERS/ERD values characterize changes (%) in the
tained below 50 kΩ. mean amplitude in the experimental conditions com-
pared with the baseline condition. ERS/ERD values are
The EEG data were preprocessed using Brain Vision
relative values that have the advantage that unspecific
Analyzer 2 software (version 2.1.1, http://www.brainpro inter-individual differences in amplitude are attenuated.
ducts.com/downloads, Brain Products GmbH, Gilching, The ERS/ERD values were subjected to subsequent
Germany). First, 20 electrodes from the outermost ring statistical analysis.
were excluded from further analysis because of their
tendency to be affected by artifacts. Subsequently,
EEG data were bandpass-filtered from 0.5 to 30 Hz. ECG data acquisition, preprocessing, and data analysis
Artifacts caused by eye blinks and horizontal eye ECG data were collected using three electrodes (EL504;
movements were corrected using independent com- Biopac Systems, Goleta, California, USA), two of which
ponent analysis [30]. Artifact-contaminated electrodes were placed on the left and right forearm. A third elec-
were interpolated with the Hjorth nearest neighbor’s trode served as a ground electrode and was placed on
method. The remaining artifacts were removed using the right elbow. The ECG data were sampled at
an automatic raw data inspection. After preprocessing, 200 Hz (ECG100C; Biopac Systems) and recorded using
an average reference was computed, and the data AcqKnowledge software (version 4.4.2, http://www.biopac.
were segmented into segments of 4 min length com/product-category/research/software/). Brain Vision Analyzer
corresponding to the experimental conditions and the 2 was used to segment the data into the experimental

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

 Music listening and attention Jäncke et al. 597

conditions and the baseline condition. Subsequently, the also calculated effect sizes. Effect sizes in the context
QRS complexes, more specifically the R peaks, were of the rANOVAs are given using the generalized η2
detected automatically in the raw ECG data. In addition, as recommended for a repeated-measures design [34].
the data were visually inspected to ensure that no R peaks A generalized η2 more than 0.02 is considered a small
were missed by the automatic detection algorithm. The effect, a generalized η2 more than 0.13 is considered a
timing of the R peaks was extracted for further analysis. moderate effect, and a generalized η2 more than 0.26 is
considered a large effect [34].
Construction of the time series of intervals between suc-
cessive R peaks (RR time series) and the calculation of HR In addition to the rANOVAs, effect sizes as measured
and the heart rate variability (HRV) measure was performed by Cohen’s d [35] were calculated for each EEG fre-
using the R package RHRV (version 4.2.3, http://rhrv.r-forge. quency band to investigate to what extent the ERS/
r-project.org/). The RR intervals were used to construct an ERD values in the experimental conditions differed
RR time series. Next, outliers were removed automatically from 0 (using the R package lsr, version 0.5, https://
using default RHRV settings. Finally, the RR time series cran.r-project.org/web/packages/lsr/index.html ). Cohen’s d
was interpolated to construct an equidistantly sampled time measures were also calculated for the changes in the
series. From the RR time series, HR (i.e. mean beats/min) HR/HRV measure from baseline. In this context, a d
and the power in the high-frequency band (HF-HRV: more than 0.2 is considered as small, a d more than 0.5
0.15–0.4 Hz) were computed. Both HR measures were then is considered moderate, whereas a d more than 0.8 is
transformed using the natural logarithm. We did not include considered large.
the low-frequency (LF-HRV: 0.04–0.15 Hz) or very low-
frequency components of HRV (VLF-HRV: ≤ 0.04 Hz) Results
because their interpretation as indicators of autonomic Behavior
activity still remains unclear [32]. Analogous to the EEG The participants rated the musical pieces positive in all
data analysis, we calculated the relative changes in experimental conditions (valence ratings; Klangkarussell-
the experimental conditions compared with the baseline Attentive: mean = 4.9, SD = 0.92, Klangkarussell-Passive:
condition. The following formula was used for HR and the mean = 4.77, SD = 0.82, Mozart-Attentive: mean = 4.97,
HRV measure: SD = 0.85, Mozart-Passive: mean = 4.8, SD = 0.89).

ðHR=HRV in experimental condition HR=HRV in baseline conditionÞ100

Change in HR=HRV measure ¼
HR=HRV in baseline condition

Statistical analysis The participants also indicated moderately increased arousal

All statistical analyses were carried out in R (version 3.3.2, in all experimental conditions (arousal ratings; Klangkarussell-
http://www.R-project.org). For the analysis of the EEG Attentive: mean = 3.93, SD = 1.34, Klangkarussell-Passive:
data, we performed, separately for each frequency band, mean = 3.37, SD = 1.35, Mozart-Attentive: mean = 4.07,
repeated-measures analysis of variance (rANOVA) with SD = 1.20, Mozart-Passive: mean = 3.50, SD = 1.33). There
the ERS/ERD values as the dependent variable. In these was no substantial difference between the valence and
analyses, the following factors were used: (i) Music = Mozart arousal ratings of the two musical pieces. The participants
versus Klangkarussell, (ii) Condition = attentive versus rated the subjectively experienced difficulty of the experi-
passive, and (iii) Electrodes = frontal, central, parietal, and mental conditions as follows: Klangkarussell-Attentive:
occipital. The rANOVAs were calculated using the R mean = 3.53, SD = 1.5, Klangkarussell-Passive: mean = 2.63,
package ez (version 4.4-0, https://cran.r-project.org/web/packa SD = 1.27, Mozart-Attentive: mean = 4.23, SD = 1.83,
ges/ez/index.html ). For the analysis of HR and HF-HRV data, Mozart-Passive: mean = 3.07, SD = 1.51. The comparison of
we applied the same analysis logic as for the above-mentioned these subjective difficulty evaluations showed that the
EEG measures. Here, we calculated the rANOVAs with Mozart-Passive condition was evaluated as more difficult than
the following factors: (i) Music = Mozart, Klangkarussell and the Klangkarussell-Passive condition (Wilcoxon signed-rank
(ii) Condition = attentive, passive. test, P = 0.048). There was no difference between the
We will only discuss rANOVA results associated with a Mozart-Attentive and the Klangkarussell-Attentive conditions
P value less than or equal to 0.05. These results are in terms of subjective difficulty. Combining the difficulty
termed in our paper as significant, although we are fully evaluations for the two experimental conditions showed that
aware of the fact that this term is critical in the context of listening to the Mozart musical piece was rated as more dif-
experiments such as ours [33]. As we must take into ficult and demanding than listening to the Klangkarussell
consideration that P values depend on sample size, we piece (Wilcoxon signed-rank test, P = 0.01).

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

598 NeuroReport 2018, Vol 29 No 7

Fig. 1

Event-related synchronization (ERS) and event-related desynchronization (ERD) in the different experimental conditions and frequency bands. The hot
colors indicate ERS and the cold colors indicate ERD. (a) ERS/ERD Klangkarussell-Attentive, (b) ERS/ERD Klangkarussell-Passive, (c) ERS/ERD
Mozart-Attentive and (d) ERS/ERD Mozart-Passive.

Event-related synchronization and event-related Table 1 Summary of the effect sizes (Cohen’s d) of the event-
related synchronization and event-related desynchronization values
desynchronization for the different experimental conditions, frequency bands, and
Figure 1 shows the ERS/ERD results separately for each electrode clusters
musical piece and experimental condition. As one can see Frequency bands
from Fig. 1, oscillations in the theta frequency band
strongly synchronize (ERS) in all experimental conditions, Condition Theta Lower alpha Upper alpha Lower beta Upper beta

especially at frontal electrodes. For the Klangkarussell- Klangkarussell-Attentive

Passive condition, theta band ERS was observed at nearly F + 0.84 + 0.40 − 0.09 + 0.45 + 0.38
C + 0.47 + 0.17 − 0.45 + 0.06 − 0.02
all electrodes. For the lower alpha band, we found no or P + 0.38 − 0.02 − 0.51 − 0.07 − 0.07
only very small ERS in all experimental conditions. In the O + 0.61 + 0.15 − 0.28 + 0.06 + 0.09
Mozart-Passive condition, there were small lower alpha All + 0.56 + 0.16 − 0.31 + 0.10 + 0.08
Klangkarussell-Passive
band ERDs at occipital electrodes. For the upper alpha F + 0.86 + 0.51 + 0.22 + 0.61 + 0.52
band, several electrodes (especially at occipital positions) C + 0.84 + 0.35 + 0.12 + 0.49 + 0.35
P + 0.66 + 0.22 − 0.03 + 0.32 + 0.24
showed ERDs. For the lower and upper beta bands, we O + 0.72 + 0.42 0.00 + 0.44 + 0.40
found ERS during the Klangkarussell-Passive condition at All + 0.73 + 0.37 + 0.07 + 0.46 + 0.37
frontal electrode sites (lower and upper beta band). Mozart-Attentive
F + 0.71 + 0.50 − 0.01 + 0.36 + 0.35
C + 0.33 + 0.12 − 0.41 − 0.03 − 0.04
Table 1 shows the effect sizes (Cohen’s d ) characterizing P + 0.35 0.00 − 0.44 − 0.14 − 0.10
the changes in the EEG measures in the experimental O + 0.56 + 0.26 − 0.24 + 0.12 + 0.18
conditions compared with baseline separately for each All + 0.49 + 0.22 − 0.25 + 0.08 + 0.12
Mozart-Passive
frequency band and electrode cluster. F + 0.51 + 0.19 − 0.05 + 0.30 + 0.22
C + 0.33 − 0.10 − 0.32 + 0.04 − 0.08
In the following, the results of the rANOVAs of the ERS/ P + 0.37 − 0.20 − 0.45 − 0.09 − 0.26
ERD values are described separately for each frequency O + 0.64 + 0.07 − 0.38 + 0.06 − 0.02
All + 0.47 0.00 − 0.29 + 0.08 − 0.02
band. A summary of the rANOVA results is shown in
Table 2, which lists all P values separately for each effect The direction of the effect is shown by the sign [plus (+ ) = ERS, minus
(− ) = ERD]. Moderate and large effects are indicated in bold.
and frequency band. As recommended by Krauth [36] all, across all electrode clusters; C, central; ERS, event-related synchronization;
and consistent with the exploratory nature of our study, ERD, event-related desynchronization; F, frontal; P, parietal; O, occipital.

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

 Music listening and attention Jäncke et al. 599

Table 2 Summary of the repeated-measures analysis of variance factor Electrodes is mainly driven by ERDs at posterior
results broken down for each main effect and interaction,
separately for each frequency band electrodes (see Fig. 3c). No further main effects or interac-
tions were found.
Frequency bands
In the lower beta frequency band, we found significant
Lower Upper Lower Upper
Effect Theta alpha alpha beta beta
main effects for the factor Music [F(1,29) = 4.62, P = 0.04,
generalized η2 = 0.02], for the factor Electrodes [F(3,87) =
Condition 0.16 0.69 0.26 0.16 0.42 4.54, P = 0.005, generalized η2 = 0.02], and a significant
Music 0.02 0.06 0.10 0.04 0.09
Electrodes < 0.001 0.01 0.02 0.005 0.007 Condition × Music interaction [F(1,29) = 5.01, P = 0.03,
Condition : Music 0.07 0.02 0.04 0.03 0.02 generalized η2 = 0.02]. The Condition × Music interaction
Condition : Electrodes 0.43 0.94 0.83 0.87 0.76
Music : Electrodes 0.69 0.90 0.85 0.70 0.68
is driven by a substantial ERS in the Klangkarussell-
Condition : Music : Electrodes 0.31 0.41 0.90 0.84 0.48 Passive condition (see Fig. 2d). The significant effect for
the factor Electrodes is mainly driven by ERSs at frontal
The P values for each main effect and interaction are shown. Effects associated
with P ≤ 0.05 are indicated in bold letters. and central electrodes (see Fig. 3d). There were no further
significant main effects or interactions.

we report all P values to provide the reader with com- Finally, in the upper beta frequency band, we found a main
plete information on our analyses. effect for the factor Electrodes [F(3,87) = 4.34, P = 0.01,
generalized η2 = 0.02] and an interaction between the factors
The rANOVA for the theta frequency band showed sig- Condition and Music [F(1,29) = 6.58, P = 0.02, generalized
nificant main effects for the factors Music [F(1,29) = 6.24, η2 = 0.02]. No further main effect or interactions were sig-
P = 0.02, generalized η2 = 0.02] and Electrodes [F(3,87) = nificant. The Condition × Music interaction is driven by a
6.44, P < 0.001, generalized η2 = 0.03]. The main effect for large ERS during the Klangkarussell-Passive condition (see
Music is driven by larger ERS during the Klangkarussell- Fig. 2e). The significant effect for the factor Electrodes is
Passive condition (see Fig. 2a). The main effect for mainly driven by ERSs at frontal electrodes and ERDs at
Electrodes is driven by larger ERS values at posterior parietal electrodes (see Fig. 3e). Overall, the pattern was
electrodes (see Fig. 3a). No further main effect or interac- similar compared with the lower beta band.
tions were significant. In Fig. 2a, the ERS values for the
theta band are shown broken down for the four conditions. Heart rate and heart rate variability
As shown in Fig. 2a, the theta band synchronizes much Table 3 lists the effect sizes (Cohen’s d ) describing the
stronger during the Klangkarussell-Passive condition HR/HRV measure changes in each experimental condi-
than during the other three conditions. During the tion in comparison with the baseline condition. There
Klangkarussell-Passive condition, there was strong ERS as were small HR increases (as indicated by Cohen’s d )
shown by a moderate effect size (d = 0.73) compared with relative to baseline in the Klangkarussell conditions and a
the ERS values during the other conditions, which showed moderate change in the Mozart-Attentive condition. In
small or moderate effect sizes (d = 0.47–0.56). the Mozart-Passive condition, there was no HR change.
For the lower alpha frequency band, we found a significant The HF-HRV measure showed strong decreases from
main effect for the factor Electrodes [F(3,87) = 3.82, baseline in all experimental conditions, with the stron-
P = 0.01, generalized η2 = 0.02] and a significant interaction gest changes during the Klangkarussell-Passive and the
between the factors Condition and Music [F(1,29) = 5.91, Mozart-Attentive conditions.
P = 0.02, generalized η2 = 0.01]. We found no further The rANOVA for the HR showed no significant main effects
main effects or interactions. As shown in Fig. 2b, the for the factors Condition and Music, and no significant
Condition × Music interaction is driven by a large ERS interaction. The rANOVA for the HF-HRV showed a sig-
during the Klangkarussell-Passive condition in addition nificant main effect for Condition [F(1,29) = 4.28, P = 0.047,
to a lack of change from baseline in the Mozart-Passive generalized η2 =0.1], which is driven by slightly larger power
condition. The significant effect for the factor Electrodes is decreases during attentive music listening (see Fig. 4).
driven by ERDs at parietal electrodes (see Fig. 3b).
In the upper alpha frequency band, the rANOVA showed Discussion
a significant main effect for the factor Electrodes The most important finding of this exploratory study
[F(3,87) = 3.55, P = 0.02, generalized η2 = 0.02] and a sig- is that the same musical piece can evoke entirely dif-
nificant Condition × Music interaction [F(1,29) = 4.72, ferent neural activations depending on the direction of
P = 0.04, generalized η2 = 0.01]. As shown in Fig. 2c, the attention. This attention-effect was mainly present for
Condition × Music interaction is characterized by ERDs in the electronic musical piece by Klangkarussell, but not
three experimental conditions (Klangkarussell-Attentive, for the classical musical piece by Mozart. For the
Mozart-Attentive, and Mozart-Passive). In the Klangkarussell- Klangkarussell musical piece, we consistently found
Passive condition, there was no change from baseline in either stronger ERS during the passive listening condition than
direction (i.e. no ERS or ERD). The significant effect for the during the attentive listening condition. This effect was

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

600 NeuroReport 2018, Vol 29 No 7

Fig. 2

Event-related synchronization/event-related desynchronization separately for each frequency band. (a) Theta, (b) lower alpha, (c) upper alpha, (d)
lower beta, (e) upper beta. Error bars visualize the between-subject SEM.

present in all investigated frequency bands. For the strong inhibition of the parasympathetic and an activation
Mozart piece, we only identified small or absent ERS of the sympathetic nervous system. HF-HRV changes
values in nearly all frequency bands. In the upper alpha were stronger for the Mozart-Attentive compared with
band, we identified small to moderately strong ERD the Mozart-Passive condition.
values for the Mozart piece during both attentive con-
ditions. The HF-HRV measure showed strong decreases The uncovered attention-dependent EEG response for
compared with baseline during all conditions, indicating a the Klangkarussell-Passive listening condition resembles

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

 Music listening and attention Jäncke et al. 601

Fig. 3

Event-related synchronization/event-related desynchronization separately for each frequency band and electrode cluster. (a) Theta, (b) lower alpha, (c)
upper alpha, (d) lower beta, (e) upper beta.

the synchronizations across all frequency bands found in involved in generating this internal state. We have
a previous study of our group [37]. In this earlier study, speculated that this state would fit with a psychological
we found a general EEG synchronization while the par- state in which the listeners are ‘drawn into’ the musical
ticipants listened to a famous and well-appreciated aria. piece. A similar, or even the same, state could have been
We have interpreted this synchronization pattern as a induced during the Klangkarussell-Passive condition.
neurophysiological state associated with increased inter- This particular musical piece was well appreciated by the
nal but reduced external attention. This state is accom- participants (most of them listen more to pop music than
panied by increased inhibition of brain networks not to classical music) and was a major hit during the time of

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

602 NeuroReport 2018, Vol 29 No 7

testing. Thus, this musical piece might have required musical piece did not induce the particular neurophy-
fewer attentional resources, and therefore, the partici- siological and psychological state induced by the
pants could have listened in a ‘by-the-way style’ that did Klangkarussell musical piece.
not affect the performance on the primary task (here:
counting the occurrence of certain aspects of the film However, it should be kept in mind that the listening
sequences). This interpretation also fits with the sub- conditions used in this experiment are different com-
jective assessment of the experienced difficulty by the pared with the listening condition used in the afore-
participants during the two conditions. The participants mentioned study. In the earlier study, the participants’
rated the Mozart-Passive condition as more demanding listening to the musical piece was unrestricted, while the
and difficult than the Klangkarussell-Passive condition. listening conditions in this study are more of a dual-task
Therefore, it is quite plausible that the Klangkarussell- situation. Therefore, one might also interpret the present
Passive condition provides more of an opportunity to get findings in the context of the dual-task literature. Power
into this psychological state. increases in the theta and beta bands (partly with
decreases in the upper alpha band) have frequently been
The Mozart musical piece, in contrast, was associated
reported in the dual-task literature [39–41] and with
with small to moderate ERDs in the upper alpha band
increased mental workload [42,43]. Increased activity in
during both listening conditions, particularly at posterior
the alpha band (especially over parietal regions) is often
electrodes. This ERD of the upper alpha band indicates
associated with increased activity of the tonic alertness
an activation of the dorsal attention network (comprising
the superior parietal lobule, the intraparietal sulcus area, network (comprising the dorsal anterior cingulum, the
and parts of the dorsal visual cortex), which is most pro- anterior insula, the thalamus, and the anterior prefrontal
minently involved in selective spatial attention [38]. cortex) [38]. Thus, in the context of these findings,
Perhaps the participants directed their attention more passive listening to the Klangkarussell musical piece
strongly to the film sequence and were not drawn into would have induced an increased workload and more
the Mozart musical piece. In other words, the classical tonic alertness, requiring increased activation and invol-
vement of the responsible neural networks. However,
this interpretation seems to be less plausible as the ANS
Table 3 Effect sizes as measured by Cohen’s d describing changes responses indicate ANS arousal for all conditions. If
in the heart rate/heart rate variability measure in the experimental mental workload was increased during one of these lis-
conditions in comparison with the baseline condition
tening conditions, we would have anticipated stronger
Condition HR HF-HRV
HF-HRV responses as well as stronger HR increases
Klangkarussell-Attentive + 0.29 − 0.94 [44,45]. In summary, mental workload is a less plausible
Klangkarussell-Passive + 0.29 − 1.31 explanation for this pattern of EEG and ANS responses.
Mozart-Attentive + 0.37 − 1.32
Mozart-Passive − 0.03 − 0.92 We favor the hypothesis that the music by Klangkarussell
had the power to induce a specific psychological and
The direction of the effect is indicated by the sign [plus (+ ) = increase, minus
(− ) = decrease]. neurophysiological state that comes close to what we
HR, heart rate; HF-HRV, high-frequency heart rate variability. have described as being ‘drawn into the music’.

Fig. 4

Heart rate (HR) and high-frequency heart rate variability (HF-HRV). (a) HR changes relative to baseline in the experimental conditions. (b) High-frequency
HF-HRV changes relative to baseline in the experimental conditions. Error bars represent the between-subject SEM.

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

 Music listening and attention Jäncke et al. 603

In this study, we refrained from estimating intracortical 10 Rogenmoser L, Zollinger N, Elmer S, Jäncke L. Independent component
sources on the basis of surface EEG signals to base our processes underlying emotions during natural music listening. Soc Cogn
Affect Neurosci 2016; 11:1428–1439.
interpretation on more robust EEG data. However, from 11 Altenmüller E, Schürmann K, Lim VK, Parlitz D. Hits to the left, flops to the
functional MRI studies it can be seen that directing right: different emotions during listening to music are reflected in cortical
attention away from the auditory stimuli results in less lateralisation patterns. Neuropsychologia 2002; 40:2242–2256.
12 Bhattacharya J, Petsche H. Universality in the brain while listening to music.
and more variable auditory cortex activation. Therefore, Proc R Soc Lond B Biol Sci 2001; 268:2423–2433.
we hypothesize that in the passive music listening con- 13 Bhattacharya J, Petsche H. Enhanced phase synchrony in the
ditions, auditory cortex activation may have decreased, electroencephalograph gamma band for musicians while listening to music.
Phys Rev E Stat Nonlin Soft Matter Phys 2001; 64:012902.
whereas activation in the visual areas may have increased. 14 Bhattacharya J, Petsche H. Musicians and the gamma band: a secret affair?
However, this auditory attention-effect has to be shown Neuroreport 2001; 12:371–374.
in future experiments using musical stimuli [23,24]. 15 Bhattacharya J, Petsche H, Pereda E. Long-range synchrony in the gamma
band: role in music perception. J Neurosci 2001; 21:6329–6337.
Although the results of our study must be substantiated 16 Carrus E, Koelsch S, Bhattacharya J. Shadows of music-language interaction
on low frequency brain oscillatory patterns. Brain Lang 2011; 119:50–57.
by follow-up experiments using more participants, dif- 17 Mikutta C, Altorfer A, Strik W, Koenig T. Emotions, arousal, and frontal alpha
ferent musical pieces, and further experimental manip- rhythm asymmetry during Beethoven’s 5th symphony. Brain Topogr 2012;
ulations, we would like to emphasize that more 25:423–430.
18 Mikutta CA, Maissen G, Altorfer A, Strik W, Koenig T. Professional musicians
experiments should be conducted using similar para- listen differently to music. Neuroscience 2014; 268:102–111.
digms to the one used in this study. Future experiments 19 Sarnthein J, vonStein A, Rappelsberger P, Petsche H, Rauscher FH,
should study neurophysiological and autonomic respon- Shaw GL. Persistent patterns of brain activity: an EEG coherence study of
the positive effect of music on spatial-temporal reasoning. Neurol Res 1997;
ses to music in situations resembling everyday situations.
19:107–116.
When we listen to music in everyday situations, we often 20 Tsang CD, Trainor LJ, Santesso DL, Tasker SL, Schmidt LA. Frontal EEG
simultaneously perform other tasks. In the present responses as a function of affective musical features. Ann N Y Acad Sci
2001; 930:439–442.
experiment, we have attempted to model such an
21 Schaefer RS, Vlek RJ, Desain P. Music perception and imagery in EEG:
everyday situation to study the EEG and ANS responses alpha band effects of task and stimulus. Int J Psychophysiol 2011;
to music in a more ecologically valid setting. By doing 82:254–259.
this, we have shown that the neurophysiological 22 Schaefer RS, Desain P, Farquhar J. Shared processing of perception and
imagery of music in decomposed EEG. Neuroimage 2013; 70:317–326.
responses to a particular musical piece strongly depend 23 Jäncke L, Mirzazade S, Shah NJ. Attention modulates activity in the primary
on the attention focus. Thus, we have provided scientific and the secondary auditory cortex: a functional magnetic resonance imaging
arguments for studying the neural underpinnings of study in human subjects. Neurosci Lett 1999; 266:125–128.
24 Jäncke L, Buchanan TW, Lutz K, Shah NJ. Focused and nonfocused
music perception in more ecologically valid experiments. attention in verbal and emotional dichotic listening: an FMRI study. Brain
Lang 2001; 78:349–363.
25 Johnson JA, Zatorre RJ. Attention to simultaneous unrelated auditory and
Acknowledgements visual events: Behavioral and neural correlates. Cereb Cortex 2005;
This research was supported by the Swiss National 15:1609–1620.
Foundation (grant no. 320030_163149 granted to L.J.). 26 Petkov CI, Kang X, Alho K, Bertrand O, Yund EW, Woods DL. Attentional
modulation of human auditory cortex. Nat Neurosci 2004; 7:658–663.
27 Hutcherson CA, Goldin PR, Ochsner KN, Gabrieli JD, Barrett LF, Gross JJ.
Conflicts of interest Attention and emotion: does rating emotion alter neural responses to
There are no conflicts of interest. amusing and sad films? Neuroimage 2005; 27:656–668.
28 Markovic A, Kühnis J, Jäncke L. Task context influences brain activation
during music listening. Front Hum Neurosci 2017; 11:342.
References 29 Annett M. A classification of hand preference by association analysis. Br J
1 Baumgartner T, Lutz K, Schmidt CF, Jäncke L. The emotional power of Psychol 1970; 61:303–321.
music: how music enhances the feeling of affective pictures. Brain Res 30 Jung TP, Makeig S, Humphries C, Lee TW, McKeown MJ, Iragui V, et al.
2006; 1075:151–164. Removing electroencephalographic artifacts by blind source separation.
2 Koelsch S, Fritz T, V Cramon DY, Müller K, Friederici AD. Investigating Psychophysiology 2000; 37:163–178.
emotion with music: an fMRI study. Hum Brain Mapp 2006; 27:239–250. 31 Pfurtscheller G, Lopes da Silva FH. Event-related EEG/MEG synchronization
3 Koelsch S, Skouras S, Fritz T, Herrera P, Bonhage C, Küssner MB, et al. The and desynchronization: basic principles. Clin Neurophysiol 1999;
roles of superficial amygdala and auditory cortex in music-evoked fear 110:1842–1857.
and joy. Neuroimage 2013; 81:49–60. 32 Billman GE. Heart rate variability – a historical perspective. Front Physiol
4 Salimpoor VN, van den Bosch I, Kovacevic N, McIntosh AR, Dagher A, 2011; 2:86.
Zatorre RJ. Interactions between the nucleus accumbens and auditory 33 Benjamin DJ, Berger JO, Johannesson M, Nosek BA, Wagenmakers EJ,
cortices predict music reward value. Science 2013; 340:216–219. Berk R, et al. Redefine statistical significance. Nat Hum Behav 2017;
5 Salimpoor VN, Zald DH, Zatorre RJ, Dagher A, McIntosh AR. Predictions and 2:6–10.
the brain: how musical sounds become rewarding. Trends Cogn Sci 2015; 34 Bakeman R. Recommended effect size statistics for repeated measures
19:86–91. designs. Behav Res Methods 2005; 37:379–384.
6 Trost W, Ethofer T, Zentner M, Vuilleumier P. Mapping aesthetic musical 35 Cohen J. Statistical power analysis for the behavioral sciences. New York,
emotions in the brain. Cereb Cortex 2012; 22:2769–2783. NY: Routledge; 1988.
7 Trost W, Frühholz S, Cochrane T, Cojan Y, Vuilleumier P. Temporal dynamics 36 Krauth J. Distribution-free statistics An application-oriented approach.
of musical emotions examined through intersubject synchrony of brain Amsterdam; New York; Oxford: Elsevier; 1988.
activity. Soc Cogn Affect Neurosci 2015; 10:1705–1721. 37 Jäncke L, Kühnis J, Rogenmoser L, Elmer S. Time course of EEG oscillations
8 Jäncke L. The dynamic audio-motor system in pianists. Ann N Y Acad Sci during repeated listening of a well-known aria. Front Hum Neurosci 2015; 9:401.
2012; 1252:246–252. 38 Sadaghiani S, Scheeringa R, Lehongre K, Morillon B, Giraud AL,
9 Jäncke L, Alahmadi N. Detection of independent functional networks during Kleinschmidt A. Intrinsic connectivity networks, alpha oscillations, and tonic
music listening using electroencephalogram and sLORETA-ICA. Neuroreport alertness: a simultaneous electroencephalography/functional magnetic
2016; 27:455–461. resonance imaging study. J Neurosci 2010; 30:10243–10250.

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.

604 NeuroReport 2018, Vol 29 No 7

39 Casutt G, Martin M, Jäncke L. Driving simulator training is associated with 43 Gevins A, Smith ME, Leong H, McEvoy L, Whitfield S, Du R, et al.
reduced inhibitory workload in older drivers. Geriatrics 2016; 1:16. Monitoring working memory load during computer-based tasks
40 Lin CT, Chen SA, Chiu TT, Lin HZ, Ko LW. Spatial and temporal EEG dynamics with EEG pattern recognition methods. Hum Factors 1998;
of dual-task driving performance. J Neuroeng Rehabil 2011; 8:11. 40:79–91.
41 Pizzamiglio S, Naeem U, Abdalla H, Turner DL. Neural correlates of single- 44 Koelsch S, Jäncke L. Music and the heart. Eur Heart J 2015;
and dual-task walking in the real world. Front Hum Neurosci 2017; 11:460. 36:3043–3049.
42 Gevins A, Leong H, Du R, Smith ME, Le J, DuRousseau D, et al. Towards 45 Thayer JF, Lane RD. Claude Bernard and the heart–brain connection: further
measurement of brain function in operational environments. Biol Psychol elaboration of a model of neurovisceral integration. Neurosci Biobehav Rev
1995; 40:169–186. 2009; 33:81–88.

Copyright r 2018 Wolters Kluwer Health, Inc. All rights reserved.