Professional Documents
Culture Documents
Green Et Al 2008
Green Et Al 2008
Green Et Al 2008
a
Departments of Psychology, bAnthropology, University of Aarhus, cMR Research Centre, dCenter of Functionally Integrative Neuroscience,
Aarhus University Hospital and eRoyal Academy of Music, Aarhus, Denmark
Correspondence to Anders C.Green, Department of Psychology, University of Aarhus, Jens Chr. Skous Vej 4, DK- 8000 Aarhus C, Denmark
Tel: + 45 89424900; fax: + 45 89424901; e-mail: anders@psy.au.dk
Using functional magnetic resonance imaging, we contrasted ma- increased activity in limbic structures, namely left parahippocam-
jor and minor mode melodies controlled for liking to study the pal gyrus, bilateral ventral anterior cingulate, and in left medial pre-
neural basis of musical mode perception. To examine the in£uence frontal cortex. Dissonance explained some, but not all, of the
of the larger dissonance in minor melodies on neural activation dif- heightened activity in the limbic structures when listening to minor
ferences, we further introduced a strongly dissonant stimulus, in mode music. NeuroReport 19:711^715 c 2008 Wolters Kluwer
the form of a chromatic scale. Minor mode melodies were evalu- Health | Lippincott Williams & Wilkins.
ated as sadder than major melodies, and in comparison they caused
Keywords: dissonance, emotion, functional magnetic resonance imaging, limbic, mode, music
0959- 4965
c Wolters Kluwer Health | Lippincott Williams & Wilkins Vol 19 No 7 7 May 2008 711
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
NEUROREPORT GREEN ETAL.
such stimuli lack ecological validity, given that real music included realignment to the first image and unwarping,
unfolds over time. Third, we controlled liking for the music, coregistration with the T1-weighted images and slice
as a liking difference can by itself alter brain activation timing. Subsequently, segmentation of grey and white
[5,6,13]. matter, together with spatial normalization, was performed
By ensuring this high level of experimental control, and using the unified segmentation approach [14], and images
by investigating the possible relationship of mode with were smoothed (10-mm full width at half maximum
dissonance, this study aimed to clarify the neural under- Gaussian kernel).
pinnings of musical emotion. Experimental effects were estimated with a general linear
model [15]. Second-level random effects analyses yielded
statistical parametric maps (SPMs). Contrasts were set up
Methods to test for mode effect (minor vs. major). SPMs were
Stimuli thresholded at Po0.05, Family Wise Error corrected for the
A pool of 30 unknown single-voice melodies of 13 s duration entire volume at voxel level, whereas cluster level correction
each was used; half in major mode, half in minor. Melodies was also investigated. Voxel coordinates were transformed
were digital piano sequences, quantized and generated in (http://imaging.mrc-cbu.cam.ac.uk/imaging) from Montreal
Cubase SX (Steinberg Media Technologies GmbH, Ham- Neurological Institute to Talairach space. Peak–voxel ana-
burg, Germany), postprocessed in Adobe Audition. Three lyses were performed on the areas found in the second-level
different keys (A flat, E and C) were used to add variety; analysis; averaged b values for the main effects of major and
other parameters were kept identical, including ambitus minor were compared with those for the chromatic scale.
(a 10th), tempo (80 BPM) and amplitude.
The happiness or sadness expressed by the melodies was
judged by a separate group of 21 volunteers, on a scale of
Results
1 (saddest) to 5 (happiest).
Behavioural results
An ascending chromatic scale of uniformly distributed
Minor mode melodies were judged as sadder than major:
dissonant intervals (minor seconds), but otherwise with
mean major 2.62 (SD¼0.33) and mean minor 3.57 (SD¼0.40;
similar characteristics as the melodies, was used as the third
paired t-test: T¼8.001, Po0.001).
stimulus type.
Major and minor mode melodies were rated as equally
likeable, mean major 3.06 (SD¼0.38) and mean minor 3.26
Procedure and participants (SD¼0.48; paired t-test: T¼1.935, P40.05).
Twenty-one healthy nonmusicians, right-handed (self re-
port), with a mean age of 27.3 (range: 20–33) were instructed
to visually fixate on a cross on a screen, listen carefully to Functional MRI results
the stimuli over headphones, and immediately after each The contrast of minor over major melodies revealed
melody to rate how well they liked it on a scale from 1 (least activation in left parahippocampal gyrus (LPHG) (Fig. 1a).
liked) to 5 (most liked), using a response box with one A further activation site (Fig. 1b) at cluster level correction
button for each finger. Stimulus sequences were pseudo- encompassed left and right ventral anterior cingulate cortex
randomized for each participant individually, and consisted (VACC) (BA 24), and extended into left medial frontal gyrus
of 12 major and 12 minor mode melodies, six chromatic (LMFG) of the medial prefrontal cortex (BA 10). The
scale repetitions and a number of 2–8 s silences in between. opposite contrast of major over minor yielded no significant
Participants performed a practice session before the actual activation.
fMRI scan to familiarize themselves with all aspects of In the peak–voxel analysis, the mean response to the
the procedure. The ethical committee of Aarhus County chromatic scale fell between that of major and minor mode
approved the study. melodies with regard to three of the four areas mentioned
The reported study was a separate part of a larger above (Fig. 2). None of these differences between chromatic
investigation (forthcoming) of memory effects on music scale and the melodies were statistically significant (re-
perception. This involved a learning phase preceding the peated measures analysis of variance, P40.05). In LMFG,
fMRI scan, where participants listened to 18 of the melodies the chromatic scale elicited the greatest (least negative)
a varying number of times each. In the scanner, they listened response, followed by minor, and then major mode. Here,
to these again once each, along with a further six melodies the difference between chromatic scale and major was
(for a total of 24). As the distribution of major and minor significant (Po0.05).
mode was evenly balanced in every group of melodies,
these additional elements were considered to be of no
consequence to the results reported here. Discussion
In this study, we found limbic (LPHG, bilateral VACC) and
Image acquisition and data analysis limbic-related structures (LMFG) differentially activated by
Scanning was performed on a Signa Excite 1.5 T MR scanner listening to melodies in minor mode, compared with
(General Electric Medical Systems, Milwaukee, Wisconsin, equivalent major melodies. At the behavioural level, we
USA). T1-weighted anatomical images were acquired, and confirmed that the presented melodies in minor were
functional bold images were obtained using a gradient- perceived as sadder than those in major mode. This is
echo echoplanar imaging sequence; TR¼2700 ms, TE¼40 ms. consistent with the fact that limbic structures are central to
Thirty-four axial slices of 5-mm thickness (no gap), covering processing emotions [16]. We further propose that degree
the whole brain were acquired (64 64 matrix). of dissonance may account for some but not all of the
Functional MRI data analysis was done with SPM5 heightened activity of the brain regions responsive to minor
(Institute of Neurology, London, UK). Data preprocessing mode melodies. Thus, the study conclusively shows that
(a) (b)
6 6
5 5
4 4
3 3
2 2
1 1
0 0
Fig. Talairach Z score P
coordinates
(x,y,z)
(a) Voxel-level significant Voxel
activations (FWE) level
L parahippocampal gyrus −18,−44,8 4.65 0.039
(b) Cluster-level significant Cluster
activations level
R anterior cingulate (BA 24) 8, 33, 6 3.76 0.031
L anterior cingulate (BA 24) −2, 37, 6 3.70 0.031
L medial frontal gyrus (BA 10) −14, 47, 3 3.68 0.031
Fig. 1 Statistical activation maps for the minor^ major contrast of the random e¡ects analysis, withTalairach coordinates for the peak activation sites,
and their Z scores and P values.The right side of the images corresponds to the right hemisphere. FWE, Family Wise Error.
(a) 0.1 (b) 0.1 limbic structures are more involved in processing minor
than major mode melodies.
anterior cingulate activation was found when participants stimulus was in between and not significantly different from
listened to happy versus neutral music [9], and has been the response to the major and minor melodies (Fig. 2).
shown to correlate with intensity of musical ‘chills’ [21]. It is Hence, harmonic dissonance may be a contributing but
conceivable that the care taken in our study to optimize certainly not an exclusive factor in the observed minor-
experimental control, compared with earlier studies [7–10], related activity increase. Only in LMFG did we find traces of
has allowed for a finer differentiation of anterior cingulate the hypothesized response pattern. Musicologically, these
activity in response to musical emotion. findings suggest that there must be more to the minor mode
The involvement of LMFG possibly reflects its strong than its comparatively high degree of dissonance. It is worth
connectivity with the anterior cingulate cortex and other emphasizing here that the dissonant stimulus material was
limbic structures; a connectivity that has supported the a simple chromatic scale, whereas the minor (and major)
proposed role of the medial prefrontal cortex as a centre for mode stimuli were actual melodies. This indicates that in
integration of emotional content from these regions [19,22]. order for dissonance to exert its maximum influence on
This was corroborated with respect to musically induced brain activity, and hence perception, it needs to occur within
emotions by a recent study [23], which investigated the a ‘real’ musical context. An obvious possibility for future
emotionally enhancing effect of presenting congruent investigation would be to study dissonant melodies, not just
classical music together with affective pictures (as opposed scales.
to pictures alone). Both the medial frontal gyrus and the
parahippocampus were among the brain regions differen-
Conclusion
tially activated by the combined visual/auditive stimuli. This study showed that left parahippocampal gyrus is
Additionally, activation of left medial prefrontal cortex
crucially involved when people listen to melodies in the
(BA 10) has been shown before in a minor–major contrast sad minor mode, as opposed to the happier major mode.
[7]. The authors claimed that involvement of BA 10 was
Further areas implicated were bilateral ventral anterior
caused by a hypothesized need to introspectively evaluate cingulate and left medial prefrontal cortex. Compared with
the emotional (happy–sad) content. However, it is unclear to
earlier studies, the present design controlled liking for the
us why this introspection should be more pronounced when melodies, as well as a range of acoustic and musical
judging minor rather than major mode music. Conse- parameters. The greater dissonance of minor than major
quently, there is need for another explanation of what may account for some, but not all, of the specific brain
causes the increased emotional brain activity when listening activation related to listening to music in minor. The most
to minor mode music in particular. pertinent behavioural explanation for the observed brain
activity is the emotional quality of sadness associated with
the minor mode.
Dissonance as an explanatory factor
We tested whether the limbic activation in the minor–major
contrast might reflect the higher degree of dissonance in Acknowledgements
minor melodies. This was performed by comparing the The authors thank Bo Sommerlund, Torben E. Lund, Kim
response to a chromatic scale, which is strongly dissonant, Mouridsen, Christina Scheel Meyer and Pernille Bruhn. The
to major and minor mode melodies in the peak voxels of the study was partly funded by The Danish Research Council
brain areas activated in the minor–major contrast. Musico- for Communication and Culture, and The Danish National
logically, the major, minor and chromatic stimuli reflect Research Foundation.
three different levels of musical dissonance, where major
mode is the least dissonant and the chromatic scale the
most, as all its melodic intervals are dissonant. Although References
1. Meyer LB. Emotion and meaning in music. Chicago: University of Chicago
dissonance is usually associated with chords, it is also
Press; 1956.
relevant for intervals in single-voice melodies, such as our 2. Hevner K. The affective character of the major and minor mode in music.
stimuli, because subsequent notes are related to each other Am J Psychol 1935; 47:103–118.
in the sensory memory of the listener. Dissonance can 3. Krumhansl C, Kessler EJ. Tracing the dynamic changes in perceived tonal
induce musical tension, which according to musicology is a organization in a spatial representation of musical keys. Psychol Rev 1982;
main carrier of musical meaning and emotion [1,24]. The 89:334–368.
4. Aarden B. Dynamic melodic expectancy. PhD dissertation. Ohio State
higher degree of tension in the minor mode could also be a
University; 2003.
contributing factor to the emotional quality of sadness it 5. Koelsch S, Fritz T, von Cramon DY, Müller K, Friederici A. Investigating
elicits in listeners, including those in our study. To test emotion with music: an fMRI study. Hum Brain Mapp 2006; 27:239–250.
whether dissonance alone could explain the minor–major 6. Blood AJ, Zatorre RJ, Bermudez P, Evans AC. Emotional responses to
contrast, it was hypothesized that the chromatic scale would pleasant and unpleasant music correlate with activity in paralimbic brain
elicit the strongest response, minor mode the second regions. Nat Neurosci 1999; 2:382–387.
7. Khalfa S, Schön D, Anton J-L, Liégeois-Chauvel C. Brain regions involved
strongest, and major mode the weakest in the brain areas
in the recognition of happiness and sadness in music. Neuroreport 2005;
responding more to minor than major mode music. This 16:1981–1984.
should hold true for LPHG in particular, as this area has 8. Pallesen KJ, Brattico E, Bailey C, Korvenoja A, Koivisto J, Gjedde A,
repeatedly been implicated in perceiving dissonance et al. Emotion processing of major, minor, and dissonant chords:
[5,6,18], but also for the other three brain areas, in as much a functional magnetic resonance imaging study. Ann N Y Acad Sci 2005;
as they may constitute a common functional network. 1060:450–453.
9. Mitterschiffthaler MT, Fu CHY, Dalton JA, Andrew CM, Williams SCR.
Moreover, the medial frontal gyrus and anterior cingulate
A functional MRI study of happy and sad affective states induced by
have also been involved in processing dissonance [25]. classical music. Hum Brain Mapp 2007; 28:1150–1162.
The peak–voxel analysis showed that for LPHG and the 10. Mizuno T, Sugishita M. Neural correlates underlying perception of
two VACC areas, the mean response to the chromatic tonality-related emotional contents. Neuroreport 2007; 18:1651–1655.
11. Peretz I, Gaudreau D, Bonnel A-M. Exposure effects on music preference 19. Bush G, Luu P, Posner MI. Cognitive and emotional influences in anterior
and recognition. Mem Cognit 1998; 26:884–902. cingulate cortex. Trends Cogn Sci 2000; 4:215–222.
12. Peretz I, Zatorre RJ. Brain organization for music processing. Annu Rev 20. Allman JM, Hakeem A, Erwin JM, Nimchinsky E, Hof P. The anterior
Psychol 2005; 56:89–114. cingulate cortex. The evolution of an interface between emotion and
13. Menon V, Levitin DJ. The rewards of music listening: response and cognition. Ann N Y Acad Sci 2001; 935:107–117.
physiological connectivity of the mesolimbic system. Neuroimage 2005; 21. Blood AJ, Zatorre RJ. Intensely pleasurable responses to music correlate
28:175–184. with activity in brain regions implicated in reward and emotion. Proc Natl
14. Ashburner J, Friston KJ. Unified segmentation. Neuroimage 2005; 26: Acad Sci U S A 2001; 98:11818–11823.
839–851. 22. Krawczyk DC. Contributions of the prefrontal cortex to the neural
15. Friston KJ, Holmes A, Worsley KJ, Poline JB, Frith CD, Frackowiak RSJ. basis of human decision making. Neurosci Biobehav Rev 2002; 26:
Statistical parametric maps in functional imaging: a general linear 631–664.
approach. Hum Brain Mapp 1995; 2:189–210. 23. Baumgartner T, Lutz K, Schmidt CF, Jäncke L. The emotional power of
16. LeDoux JE. Emotion circuits in the brain. Annu Rev Neurosci 2000; music: how music enhances the feeling of affective pictures. Brain Res
23:155–184. 2006; 1075:151–164.
17. Nolte J. The human brain: an introduction to its functional anatomy. 5th ed. 24. Vuust P, Roepstorff A, Wallentin M, Mouridsen K, Østergaard L. It don’t
St Louis: Mosby; 2001. mean a thingy Keeping the rhythm during polyrhythmic tension,
18. Gosselin N, Samson S, Adolphs R, Noulhiane M, Roy M, Hasboun D, activates language areas (BA47). Neuroimage 2006; 31:832–841.
et al. Emotional responses to unpleasant music correlates with damage to 25. Foss AH, Altschuler EL, James KH. Neural correlates of the Pythagorean
parahippocampal cortex. Brain 2006; 129:2585–2592. ratio rules. Neuroreport 2007; 18:1521–1525.