The experience of beauty derived from sorrow

Abstract

We studied the neural mechanisms that are engaged during the experience of beauty derived from sorrow and from joy, two experiences that share a common denominator (beauty) but are linked to opposite emotional valences. Twenty subjects viewed and rerated, in a functional magnetic resonance imaging scanner, 120 images which each had classified into the following four categories: beautiful and sad; beautiful and joyful; neutral; ugly. The medial orbito-frontal cortex (mOFC) was active during the experience of both types of beauty. Otherwise, the two experiences engaged different parts of the brain: joyful beauty engaged areas linked to positive emotions while sorrowful beauty engaged areas linked to negative experiences. Separate regions of the cerebellum were engaged during experience of the two conditions. A functional connectivity analysis indicated that the activity within the mOFC was modulated by the supplementary motor area/middle cingulate cortex, known to be engaged during empathetic experiences provoked by other peoples' sadness. Hum Brain Mapp, 2017. © 2017 Wiley Periodicals, Inc.

Full text

The Experience of Beauty Derived from Sorrow
Tomohiro Ishizu
1,2
*and Semir Zeki
1
*
1
Wellcome Laboratory of Neurobiology, Department of Cell and Developmental Biology,
University College London, London, United Kingdom
2
Japan Society for the Promotion of Science, Tokyo, Japan
r r
Abstract: We studied the neural mechanisms that are engaged during the experience of beauty derived
from sorrow and from joy, two experiences that share a common denominator (beauty) but are linked
to opposite emotional valences. Twenty subjects viewed and rerated, in a functional magnetic reso-
nance imaging scanner, 120 images which each had classified into the following four categories: beauti-
ful and sad; beautiful and joyful; neutral; ugly. The medial orbito-frontal cortex (mOFC) was active
during the experience of both types of beauty. Otherwise, the two experiences engaged different parts
of the brain: joyful beauty engaged areas linked to positive emotions while sorrowful beauty engaged
areas linked to negative experiences. Separate regions of the cerebellum were engaged during experi-
ence of the two conditions. A functional connectivity analysis indicated that the activity within the
mOFC was modulated by the supplementary motor area/middle cingulate cortex, known to be
engaged during empathetic experiences provoked by other peoples’ sadness. Hum Brain Mapp 38:4185–
4200, 2017.V
C2017 The Authors Human Brain Mapping Published by Wiley Periodicals, Inc.
Key words: aesthetic experience; emotion; functional magnetic resonance imaging
r r
You came to me to learn the Pleasure of Life and the Pleasure of
Art. Perhaps I was chosen to teach you something much more
wonderful, the meaning of Sorrow, and its beauty
Oscar Wilde (1896),
(in a letter to ‘Bosie’, Lord Alfred Douglas)
INTRODUCTION
Over the past few years, we and others have addressed
the question of what neural mechanisms are engaged dur-
ing aesthetic experiences, and especially during the experi-
ence of beauty. In addressing the question, we were
inspired by the question posed by the English art critic,
Clive Bell. In his book Art [1914], Bell asked whether there
is anything common to all objects that are experienced as
beautiful or that arouse the aesthetic emotion. Translating
this into neural terms we, likewise, sought to understand
whether there is a common brain system in which activity
correlates with the experience of beauty. Surprising though
it may seem, although the experience of beauty derived
from different sources entails activity in different areas of
the brain, depending upon the source, there is one com-
mon area, located in the medial orbito-frontal cortex
(mOFC) of the emotional brain, in which activity correlates
parametrically with the experience of beauty, whether
derived from sensory sources such as music or visual art
[e.g., Ishizu and Zeki, 2011; Kawabata and Zeki, 2004],
from moral sources [Tsukiura and Cabeza, 2011; Wang
et al., 2015] or from highly cognitive sources such as
Additional Supporting Information may be found in the online
version of this article.
Contract grant sponsor: Wellcome Trust, London.
Tomohiro Ishizu is currently at Department of Basic Psychological
Research and Research Methods, University of Vienna
(tomohiro.ishizu@univie.ac.at).
*Correspondence to: Tomohiro Ishizu, Gower Street, London,
WC1E 6BT, United Kingdom. E-mail: t.ishizu@ucl.ac.uk or Semir
Zeki, Gower Street, London, WC1E 6BT, United Kingdom. E-mail:
s.zeki@ucl.ac.uk
Received for publication 10 August 2016; Revised 3 May 2017;
Accepted 11 May 2017.
DOI: 10.1002/hbm.23657
Published online 23 May 2017 in Wiley Online Library (wileyonli-
nelibrary.com).
rHuman Brain Mapping 38:4185–4200 (2017) r
V
C2017 The Authors Human Brain Mapping Published by Wiley Periodicals, Inc.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and
reproduction in any medium, provided the original work is properly cited.

mathematics [Zeki et al., 2014]. In the work reported here,
we were inspired by the quote from Oscar Wilde given
above, to explore the neural activity that correlates with
the experience of beauty derived from sorrow. The quote
implies that sorrowful beauty belongs in a separate, or
separable, category, which can be defined as the experi-
ence of a positive (aesthetic) emotion with a negative com-
ponent, that of sadness. To study the neural correlate of
such an experience, we had to incorporate another distinct
category—that of beauty aroused from joy, which can be
categorised as a positive emotion with a positive compo-
nent—for comparison.
Important though the distinction between sorrowful and
joyful beauty is, it is one that is not often made, or not
made emphatically enough. This is surprising because the
distinction is easily recognized and experienced by most,
even if both categories arouse the aesthetic emotion. In
sculpture, for example, Michelangelo’s great Piet
ain Rome
is suffused with pathos, tenderness and sadness, whereas
the Three Graces of Canova are joyful and playful. In
music, the waltzes of Johann Strauss are light-hearted and
engaging while the adagio from Beethoven’s Ninth Sym-
phony is permeated with contemplative sorrow. The list is
endless and includes works in literature, poetry, dance,
and theatre. Photography, especially, has provided many
examples that can be easily classified emotionally by any
viewer as sorrowful or joyful, with beauty as their com-
mon denominator; good examples are Dorothea Lange’s
iconic images of the Great Depression, and especially The
Migrant Mother, on the one hand and Bill Brandt’s East
End Girl Dancing on the other. Aside from beauty, these
two separate categories share another common denomina-
tor, empathy, which makes it possible for humans to
become aware of the feelings of others and indeed experi-
ence those same feelings—whether of joy or sorrow—to
varying degrees.
It seemed to us that this distinction provides fertile
ground for a neurobiological enquiry into the brain mecha-
nisms that are engaged when two contrasting affective
states, a negative one (sorrow) and a positive one (joy)
both result in the experience of beauty. This enquiry paral-
lels, in a sense, our previous enquiry into the distinction,
in neural terms, between the sublime and the beautiful,
two categories that have been discussed much more exten-
sively in philosophies of aesthetics. The common descrip-
tion of the sublime as containing a negative affect
(‘pleasure from displeasure’ or ‘beauty mingled with hor-
ror’) is reflected in a pattern of brain activity that is differ-
ent from that which is engaged during the experience of
the beautiful [Ishizu and Zeki, 2014]. This made it interest-
ing to enquire whether we can also detect differences in
the pattern of brain activity during aesthetic experiences
derived from two opposite states. Our general hypothesis
was that there would be profound differences with the
two experiences but that, given the pre-eminence of activ-
ity in the mOFC during the experience of the beautiful,
the latter would be active in both states, even in spite of
the evident differences between the two. Moreover, since
experiencing beauty derived from positively or negatively
valenced emotion inevitably requires mentalizing others’
emotional states or interpreting their intentions, empathy
is another common denominator to the experience of
beauty in sorrowful and joyful sources. We, thus, expected
to find activity in brain regions which past studies have
implicated in empathetic experiences.
MATERIALS AND METHODS
Participants
Twenty one healthy right-handed volunteers (11
females; 10 males; mean age, 28.6 years) from different
cultures and ethnic backgrounds (2 Taiwanese, 4 Indian, 6
Japanese, 2 Middle Eastern, and 7 West and North Euro-
peans) participated but data from one volunteer was
excluded because of excessive noise during scanning, leav-
ing us with 20 volunteers; all had normal or corrected-to-
normal vision, and none had a history of neurological or
psychiatric disorder. Written informed consent was
obtained from all and the study was approved by the
Ethics Committee of University College London, and con-
formed to the Code of Ethics of the World Medical Associ-
ation (Declaration of Helsinki). All data were anonymized.
Preliminary Psychophysical Testing and
Postscanning Ratings
Although pictures depicting war scenes, weapons, vio-
lence or strong political, and religious attributes may often
be deemed sorrowful or beautiful, we excluded them from
this study because it is known that viewing scenes depict-
ing violence, or which remind viewers of violence, can
induce immediate brain responses, such as activity within
frontal and limbic system, even when viewed passively
(e.g., Kelly et al., 2007]; we wanted to exclude such auto-
matic brain responses which are unrelated to the current
task. We, therefore, used instead pictures of events like
funerals, abandoned children and buildings, and sad faces,
as well as landscapes and daily scenes, in both mono-
chrome and in color. The 800 pictures that we used were
drawn from photographic magazines and books, including
The Family of Man (in which the photographs referred to in
the Introduction can be found), The Modern Century, and
The Great LIFE Photographers.
During a first visit to the laboratory, between 3 and 7
days prior to scanning, each subject was instructed about
the experiment and, in a psychophysical test, rated the
stimuli according to their aesthetic and emotional valence
through two questionnaires. In the aesthetic evaluation,
participants classified a picture into five groups according
to the intensity of the aesthetic experience aroused in
them, using a Likert scale extending from 5 (‘very
rIshizu and Zeki r
r4186 r

beautiful’) to 1 (very ugly), with 3 being ‘neutral’; in the
emotional evaluation, 5 corresponded to ‘very joyful’, 3 to
‘neutral’, and 1 to ‘very sorrowful’. Thus, we obtained, for
each subject, an emotional and an aesthetic rating for each
of the 800 stimuli. Participants gave the aesthetic and the
emotional ratings in counterbalanced order. It should be
noted that they were instructed to give emotional ratings
according to the feelings that they experienced when view-
ing the images, not to those of what people in the images
might feel.
Each stimulus remained on the computer screen until
participants responded to the second evaluation, after
which an inter-trial interval of 1 s followed; they were
then asked to press a button as soon as possible after they
had made their evaluation and were also asked to indicate
their familiarity with each picture (‘have you seen this pic-
ture before?’) and familiar pictures were excluded.
From these 800 rated pictures, we selected, for each sub-
ject, 120 which fell into the four categories of ‘sorrowful
beauty’, ‘joyful beauty’, ‘neutral’, and ‘ugly’. Pictures fall-
ing into the sorrowful beauty category were the ones that
had been given a rating of 1 on the emotional score and 5
on the aesthetic score; pictures falling into the joyful
beauty category had a rating of 5 on the emotional scale
and 5 on the aesthetic scale while those rated as neutral
had 3 on both scales. Stimuli rated as ugly had a score of
3 (neutral) for the emotional rating and 1 for the aesthetic
score (see Fig. 1). Each of the four categories had 30 stim-
uli, making a total of 120 stimuli which each participant
viewed in the scanner. We excluded five participants out
of twenty-six after the preliminary psychophysical tests
showed that they did not have sufficient trials for each of
the four categories. The detailed behavioral data obtained
in the preliminary psychophysics are found in Supporting
Information.
During the scanning session, participants were asked to
rerate each picture aesthetically, after viewing it, but this
time on a scale of 3 to 1 (beautiful, neutral, ugly). Immedi-
ately after scanning, they rerated the stimuli (which were
presented in the same order as in the scanner) for their
emotional valence, using a 3-point Likert scale (3 as joyful,
2 as neutral, 1 as sorrowful).
Paradigm and Procedure
Stimuli were generated using Cogent 2000 (http://
www.vislab.ucl.ac.uk/cogent_2000.php) running in MAT-
LAB (MathWorks, Natick, MA); they were back-projected
onto a screen using an LCD projector, through an angled
mirror. The resolution of the screen was 1,400 31,050 pix-
els; the height of each stimulus was 198while the width
varied.
The session began with subjects viewing a flat black
screen for 20 s to allow for T1 equilibration effects to sub-
side (the corresponding first six brain volumes were dis-
carded). After this 20 s blank period, an instruction about
the aesthetic judgment appeared on the screen, to inform
participants that a session had started. A fixation point
then appeared at the centre of the screen for 1 s against a
black background, after which visual stimuli were pre-
sented in a pseudorandom order for 6 s. We had 30
images in each of the four experimental conditions, mak-
ing 120 images in total. We had six functional scanning
sessions for each subject. Each functional session had 20
trials. To make a set of 20 images, we selected 5 images
randomly out of 30 from each of the experimental condi-
tions so that each condition had the same number of trials
through a functional session. After this procedure, we ran-
domized the sequence of stimulus presentation within a
functional scanning. The stimulus presentation was fol-
lowed by an interval with a jitter of 5–7 s, during which
participants gave their aesthetic ratings.
Following each stimulus presentation, participants were
asked to rate it on a 3-point Likert scale, by pressing one
of three buttons with their right index, middle or ring fin-
ger. The response period lasted 5–7 s and participants
could make their rating at any time during that period; it
ended with a blank period of 20 s, during which the scan-
ner continued to acquire blood-oxygen-level dependent
(BOLD) signals. The stimuli were presented in six sessions.
Each session consisted of 20 stimuli with a 20 s resting
period between the first and the last 10 trials during which
participants were instructed not to close their eyes. Prior
Figure 1.
Preliminary behavioral data summed over 20 subjects. Frequency
distribution of beauty rating (xaxis) versus emotion rating (y
axis). Size of each circle is proportional to the number of trials
for that rating.
rExperience of Sorrowful Beauty r
r4187 r

to the scanning, participants had a short practice session
with a different set of visual stimuli to those used in the
scanning session.
Functional Magnetic Resonance
Imaging Scanning
Scanning data were acquired in a 3-T Siemens Magne-
tom Trio magnetic resonance imaging scanner (Siemens,
Erlangen, Germany) fitted with a 32-channel head-coil. An
echo-planar imaging (EPI) sequence was applied for func-
tional scans to obtain BOLD signals (echo time, 30 ms;
repeat time, 3.36 s), using 48 slices to cover the whole
brain. The voxel resolution was 3 33-mm in-plane resolu-
tion, with a 2 mm slice thickness and 1-mm inter-slice
gap. Magnetic resonance imaging signal losses in the orbi-
tofrontal cortex (OFC) and amygdala were reduced by
applying a z shim gradient moment and slice tilt [Weis-
kopf et al., 2006]. T1-weighted anatomical images were
acquired at the end of the experimental sessions for each
subject (176 slices; resolution, 1 3131 mm; echo time,
2.48 ms; repeat time, 7.92 ms). Field maps were also
acquired with the Siemens standard gradient-echo field
map sequence to correct for geometric distortion of EPI
images [Hutton et al., 2002]. We also recorded the heart
and respiration rates for each subject.
Functional Magnetic Resonance Imaging
Data Analysis
All data were analysed with SPM8 (Statistical Parametric
Mapping, http://www.fil.ion.ucl.ac.uk/spm/software/
spm8/). The EPI images for each subject were realigned
and normalized into Montreal Neurological Institute
(MNI) space, smoothed with a Gaussian smoothing kernel
of 9 3939 mm, and filtered with a high-pass cut-off
(128 s) to remove drift terms. The stimulus for each subject
was modelled as a set of regressors in a general linear
model first-level (within subject) analysis. The experiment
was a block design, and boxcar functions were used to
define the stimulations; these modelled the onsets and
durations of the visual stimuli. Head movement parame-
ters calculated from the realignment preprocessing step,
physiological recordings, and response periods were
included as regressors of no interest. Stimulus functions
were convolved with a canonical haemodynamic response
function. Contrast images were taken to random-effects
second-level (between subject) analyses to produce statisti-
cal maps at the group level.
To carry out categorical contrast analyses according to
the intensity of the aesthetic and emotional experience, rat-
ings were coded as 1, 2, 3 for ‘ugly’, ‘neutral’, and ‘beauti-
ful’, and 1, 2, and 3 for ‘sorrowful’, ‘neutral’, and ‘joyful’.
We then categorized stimuli rated 3 in the aesthetic rating
and 3 in the emotional rating into ‘joyful beauty’, those
rated 3 and 1 into ‘sorrowful beauty’, those rated 2 and 2
into ‘neutral’ and those rated 1 and 2 as ‘ugly’.
ROI Analysis
Since, we hypothesised the involvement of the mOFC
during the experience of both sorrowful and joyful beauty,
we constructed a region-of-interest (ROI) in the mOFC
centered on the coordinates at which activity was found in
previous studies to correlate with the experience of beauty
derived from visual sources, field A1 (–6 41 211) [Ishizu
and Zeki, 2011], to learn whether activity there also corre-
lates with the experience of two differently valenced emo-
tions which have beauty as a common denominator. We
created a ROI mask with 10-mm sphere to extract average
contrast estimates from the mOFC ROI across subjects. We
then compared activity within this region between sorrow-
ful beauty >ugliness, as well as joyful beauty >ugliness,
against zero, to learn whether mOFC was active with both
types of beauty experience. We also compared these two
contrasts directly to learn whether the strength of activity
differed significantly during the experience of the two
emotionally distinct kinds of beauty.
Whole Brain Analysis
We also conducted whole brain categorical analyses to
chart brain activations unique to joyful beauty and sorrow-
ful beauty separately, using the contrasts of sorrowful beau-
ty >ugliness and joyful beauty >ugliness, respectively.
To characterize common brain responses involved in the
above two contrasts, we conducted a conjunction analysis
performed by a test for independently significant effects as
in a logical AND ([sorrowful beauty >ugliness] \[joyful
beauty >ugliness]) based on the minimum statistic [Nichols
et al., 2005]. Since we had an a priori assumption of the
involvement of the mOFC in the experience of beauty [e.g.,
Ishizu and Zeki, 2011, 2013], we used a small volume correc-
tion (SVC) on the mOFC with a 16 mm sphere centred on
coordinates (–6 41 211), taken from Ishizu and Zeki [2011].
Functional Connectivity Analysis
In addition to the regional activity analyses, we also
studied the functional connectivity between mOFC and
other brain regions to determine the contribution that the
latter may make to modulating mOFC activity as a func-
tion of the aesthetic and emotional valence of the stimulus,
either through input to it or output from it. For this, we
performed a psychophysiological interaction (PPI) [Friston
et al., 1997]; this tests which regions show activation pat-
terns that co-vary with mOFC activity, when stimuli are
rated as joyfully (or sorrowfully) beautiful or ugly. We
assessed changes in functional connectivity between the
seed region in the mOFC and other brain regions in two
contrasts: joyful beauty >ugliness and sorrowful
rIshizu and Zeki r
r4188 r

beauty >ugliness; the analysis for each was performed sep-
arately. The PPI employed a design matrix consisting of
three regressors representing (1) the extracted time-series of
neural activity within an 8-mm spherical region centred on
the mOFC (‘physiological variable’); (2) a second regressor
representing the psychological context of interest, that is,
joyful beauty >ugliness and sorrowful beauty >ugliness
(‘psychological variables’); (3) a third regressor representing
the interaction of the two previous variables (i.e., the interac-
tion between the psychological and physiological variables -
‘PPI term’). Head-movement parameters were also included
in the model as regressors of no interest. This enabled us to
identify areas in which the correlation in BOLD activity
with the mOFC seed region increases during trials in which
a participant experienced sorrowful (or joyful) beauty rela-
tive to those during which they experienced ugliness. The
PPI was carried out separately for each subject and entered
into random-effects group level analysis. To obtain the data
for the physiological variable, we de-convolved the time-
series of neural activity from the 8-mm ROI within the
mOFC, centred on the coordinates of subject-specific activa-
tions in the region. Thus, to define the ROI, we used the con-
trast (joyful beauty 1sorrowful beauty) >ugliness to locate
the closest local maximum to the coordinates [0 48 216]
which we obtained in the group-level analysis. The results
of this analysis are displayed in Table I and Figure 6. It
should be noted that, although we selected the mOFC as the
seed region, the PPI cannot determine directly the direction-
ality of the connectivity or the interaction between the seed
region and other brain regions; our discussion is, therefore,
limited to interactions between the mOFC and functionally
connected regions without specifying their polarity.
We report cluster level activations significant at P<0.05
family-wise error (FWE) corrected, although some of these
(indicated in the table) were also significant at the peak
level at P<0.05 FWE corrected. The coordinates of all acti-
vations are reported in MNI space.
RESULTS
Behavioral Results
Ideally, the status of the ratings given to the 120 stimuli
viewed in the scanner should be the same as that given dur-
ing the preliminary viewing test, that is, each condition (e.g.,
beautiful and sorrowful) should appear 30 times. In reality,
such an ideal situation was not reached. Based on the aes-
thetic ratings given during the scanning sessions and the
postscanning emotional ratings, we obtained the following
ratings over 20 participants: 32.4 for ‘joyful beauty’ (3–3),
31.4 for ‘sorrowful beauty’ (3–1), 26.9 for ‘neutral’ (2–2), and
29.3 for ‘ugliness’ (2–1) (Fig. 2), on average. These slight var-
iations in the ratings during the preliminary tests and dur-
ing the scanning functional magnetic resonance imaging
(fMRI) session are not of great concern since there was a rea-
sonably distributed number of trials in each condition. The
stimuli consisted of pictures of a face(s), people including
faces, and landscapes without human figures. Figure 2
shows details of the number of images in terms of stimulus
contents. It is known that viewing different types of visual
stimuli, for example, faces versus scenes, results in activa-
tion of different brain areas, reflecting the functional special-
ization of the visual brain [Kawabata and Zeki, 2004; Zeki
et al., 1991]. All four experimental conditions in this study,
however, had similar proportions of images belonging to
the different categories, that is, human figures (including
faces and people) and landscapes (see Fig. 2). The most
prevalent stimulus images we had for each subject con-
tained human figures, that is, people or faces (92.7% on
average (90.7–93.9%) across the experimental conditions),
whereas, very few consisted of scenes without a human fig-
ure (7.3% on average (6.1–9.3%)). We conducted a 2-way
analysis of variance with 2 stimulus contents (human figure,
landscape) and 4 response conditions (joyful beauty, sad
beauty, neutral, ugliness). There was no significant differ-
ence in response conditions and interactions, while a main
effect of stimulus contents alone was observed (df 519,
F51,631, P<0.001). Therefore, the brain responses found
by contrasting the conditions we report here cannot be
explained by differences in stimulus contents.
Neuroimaging Results
mOFC
To learn whether there was a difference in the strength of
activity within the mOFC between the experience of sorrow-
ful and joyful beauty, we extracted, for each subject, the con-
trast estimates within the defined ROI in the mOFC and
compared across conditions. First, one sample t-tests
revealed that the extracted averaged contrasts estimates for
‘sorrowful beauty’ and ‘joyful beauty’, both against ‘ugli-
ness’, were different from zero (sorrowful beauty, t524.3,
df 519, P<0.001; joyful beauty, t539.7, df 519, P<0.001),
showing that, as we expected, the mOFC is engaged during
the experience of both joyful beauty and sorrowful beauty.
This area has been reported to be active in previous studies
on aesthetic experiences using a variety of stimuli [Ishizu
and Zeki, 2011; Kawabata and Zeki, 2004; O’Doherty et al.,
2003; Tsukiura and Cabeza, 2011; Zeki et al., 2014]. A direct
comparison between the two conditions showed, however,
that the averaged contrast estimates for joyful beauty gave
higher values than that for sorrowful beauty (paired t-test,
t510.4, df 538, P<0.001) (Fig. 3).
In short, the ROI results suggest that the mOFC is com-
monly activated with the experience of beauty but that the
strength of the activation may be modulated by the
valenced emotions.
Whole brain contrasts
Our main aim was to learn whether the brain regions
previously reported as active during the experience of
rExperience of Sorrowful Beauty r
r4189 r

TABLE I. Location, MNI coordinates, cluster size and values for the activations produced by the categorical contrasts of sorrowful beauty>
ugliness, joyful beauty >ugliness, sorrowful beauty >joyful beauty, joyful beauty >sorrowful beauty, and by the conjunction analysis sorrowful
beauty >ugliness \joyful beauty >ugliness
Cluster
p(FWE-cor) p(FDR-cor) equivk p(unc)
peak
p(FWE-cor) p(FDR-cor) TequivZ p(unc) X Y Z (mm)
Categorical contrasts:
Sorrowful beauty >Ugliness
0.004 0.009 137 0.001 0.027 0.249 6.971 4.853 0.000 33 47 7 R MFG
0.092 0.255 6.182 4.523 0.000 42 50 10 IFG
0.132 0.255 5.949 4.417 0.000 45 44 22 dIPFC
0.001 0.004 178 0.000 0.058 0.255 6.475 4.650 0.000 26210 25 L Caudate
0.558 0.521 4.894 3.889 0.000 26234 31 PCC
0.563 0.521 4.885 3.884 0.000 9 213 25
0.010 0.016 110 0.001 0.177 0.256 5.755 4.326 0.000 36 252 232 R Cereberum (lobule VI)
0.208 0.268 5.647 4.274 0.000 27 246 226
0.333 0.373 5.315 4.110 0.000 21 240 226
0.019 0.023 94 0.003 0.719 0.555 4.628 3.741 0.000 6 38 28 R dACC
0.861 0.605 4.360 3.585 0.000 0 38 16
Joyful beauty >Ugliness
0.000 0.000 897 0.000 0.000 0.010 9.551 5.715 0.000 239 234 25 L Hippocampus
0.001 0.015 8.734 5.472 0.000 15 219 28 Caudate
0.012 0.064 7.463 5.041 0.000 21 237 7 Hippocampus
0.000 0.000 739 0.000 0.021 0.064 7.168 4.930 0.000 3 35 13 R pgACC
0.042 0.084 6.716 4.750 0.000 2320 22
0.137 0.118 5.962 4.423 0.000 3 26 22
0.072 0.122 60 0.010 0.021 0.064 7.167 4.929 0.000 236 264 238 L Cerebellum(Lobule VII)
0.054 0.084 6.561 4.686 0.000 227 258 241
0.019 0.041 89 0.003 0.234 0.155 5.606 4.255 0.000 30 258 238 R Cerebellum(Lobule VII
0.374 0.229 5.267 4.086 0.000 39 261 238
0.411 0.229 5.194 4.048 0.000 27 249 229
0.000 1.000 136 0.000 0.003 0.214 5.928 4.407 0.000 2647217 L mOFC
0.004 0.214 5.726 4.313 0.000 2341220
0.013 0.265 5.004 3.948 0.000 2347 28
0.058 1.000 4.094 3.424 0.000 2344 4
0.046 1.000 7 0.336 0.030 0.335 4.506 3.671 0.000 2932 1L rACC
Sorrowful beauty >Joyful beauty
0.001 0.001 229 0.000 0.068 0.256 4.783 4.458 0.000 230 264 52 L SPL
0.155 0.256 4.505 4.228 0.000 239 255 49 IPL
0.585 0.416 3.940 3.746 0.000 212 261 52 Precuneus
0.018 0.016 112 0.003 0.182 0.256 4.449 4.181 0.000 30 23 28 R MFG
0.425 0.341 4.105 3.889 0.000 45 32 31
0.955 0.739 3.450 3.314 0.000 45 23 40
0.004 0.005 163 0.001 0.195 0.256 4.423 4.159 0.000 248 32 25 L MFG
0.430 0.341 4.100 3.884 0.000 227 23 28
0.477 0.341 4.049 3.841 0.000 236 29 28
rIshizu and Zeki r
r4190 r

TABLE I. (continued).
Cluster
p(FWE-cor) p(FDR-cor) equivk p(unc)
peak
p(FWE-cor) p(FDR-cor) TequivZ p(unc) X Y Z (mm)
Joyful beauty >Sorrowful beauty
0.032 0.117 96 0.006 0.148 0.480 4.522 4.242 0.000 51 228 25 R TPJ/SMG
Conjunction([joyful beauty >ugliness] \[sad beauty>ugliness])
0.436
[SVC, 264121]
0.547
L (mOFC)]
28 0.104 0.028 0.155 5.056 4.679 0.000 236 264 235 L Cerebellum VII
0.023 0.211 16 0.211 0.017 0.335 3.933 3.740 0.000 2635217 L mOFC
0.026 0.335 3.772 3.600 0.000 212 38 214
PPIs:
mOFC_Sorrowful beauty >Ugliness
0.000 0.000 1,466 0.000 0.000 0.004 10.330 5.926 0.000 39 282 4 R Middle occipital gyrus
0.002 0.030 8.310 5.336 0.000 23282 1 Lingual gyrus
0.004 0.030 8.092 5.263 0.000 3 276 1
0.000 0.001 282 0.000 0.547 0.547 4.754 3.812 0.000 2624 61 L SMA (encroaching to MCC)
0.580 0.550 4.699 3.781 0.000 6 11 67
0.722 0.658 4.464 3.646 0.000 29546
[SVC, 33 47 1 ’ (dIPFC)]
0.040 0.393 8 0.393 0.024 0.236 4.450 3.640 0.000 33 53 16 R dIPFC/MFG
mOFC_Joyful beauty >Ugliness
0.000 0.000 886 0.000 0.004 0.037 8.484 5.238 0.000 51 261 22 R MTG
0.076 0.125 6.610 4.591 0.000 9 288 1
0.134 0.125 6.199 4.423 0.000 15 288 25
0.014 0.009 110 0.002 0.068 0.125 6.690 4.622 0.000 29 62 10 L rMPFC
0.866 0.387 4.459 3.579 0.000 6 59 19
0.985 0.685 3.961 3.288 0.001 212 50 16
0.000 0.000 612 0.000 0.133 0.125 6.202 4.425 0.000 242 273 211 L Occipital gyrus
0.182 0.125 5.974 4.327 0.000 227 279 217
0.333 0.131 5.509 4.117 0.000 236 267 28
| 0.013 0.009 112 0.002 0.188 0.125 5.950 4.317 0.000 239 213 52 L Precentral gyrus
0.192 0.125 5.934 4.310 0.000 248 219 49
0.955 0.548 4.168 3.412 0.000 230 222 64
Also shown are areas exhibiting greater functional connectivity with the mOFC in psychophysiological interactions (PPIs), in the contrasts sorrowful beauty >ugliness and
joyful beauty >ugliness.
rExperience of Sorrowful Beauty r
r4191 r

beauty, and in particular the mOFC, were differentially
engaged during the experience of sorrowful and joyful
beauty. But we were also interested in learning whether,
besides the mOFC, significantly different brain regions
were involved when the experience was that of sorrowful
as opposed to joyful beauty, especially given that the two
arouse different empathetic feelings.
To chart brain activations that correlate with the experi-
ence of sorrowful and joyful beauty, we performed separate
categorical contrasts of (1) sorrowful beauty >ugliness, and
of joyful beauty >ugliness; we also contrasted the activity
produced by (2) joyful beauty versus sorrowful beauty, to
directly compare brain regions that are uniquely active dur-
ing each kind of aesthetic experience. We then used a con-
junction analysis [Nichols, et al., 2005] to characterize brain
activations common to both sorrowful and joyful beauty
using the contrast [sorrowful beauty >ugliness] \[joyful
beauty >ugliness]. All activations are listed in Table I.
Sorrowful beauty versus ugliness. The contrast sorrowful
beauty >ugliness resulted in activity in lateral frontal lobe
including the middle frontal gyrus (MFG); this band of
activity extended to the inferior frontal gyrus (IFG) and
the dorso-lateral prefrontal cortex (dlPFC). The posterior
cingulate cortex (PCC), encroaching upon the caudate
Figure 2.
Behavioral data collected in the fMRI experiment, showing aver-
aged number of trials with each kind of stimulus (people, faces,
and landscape without a human figure), with standard deviations
for each of the four experimental conditions across 20 subjects.
Figure 3.
The upper panel shows the activation within the medial orbitofron-
tal cortex correlating with the experience of joyful and sorrowful
beauty. Statistical parametric maps rendered onto canonical ana-
tomical sections showing the t-statistic for (left) joyful beauty >ugli-
ness, (middle) sorrowful beauty >ugliness, and (right) the results of
a conjunction analysis for joyful beauty >ugliness \sorrowful
beauty >ugliness. Random effects analysis with 20 subjects. Display
threshold P<0.001 (uncorrected). (Lower right) Region of interest
in the mOFC. (Lower left) Averaged contrast estimates for the
contrasts joyful beauty >ugliness (JB >UG) and sorrowful beau-
ty >ugliness (SB >UG) within the defined ROI (–6 41 211), over
20 subjects. Joyful beauty caused a higher BOLD signal than sor-
rowful beauty. * P<0.05. Error bars are standard error (SE).
[Color figure can be viewed at wileyonlinelibrary.com]
rIshizu and Zeki r
r4192 r

(head and body), was also active. In addition, there was
activity in parts of the cerebellum (lobule VI) and dorsal
ACC (see Figs. 4, 7, and 8).
In short, some cortical regions which previous studies had
found to be active during sorrowful experiences were also
active in the contrast of sorrowful beauty versus ugliness.
Joyful beauty versus ugliness. The contrast joyful beauty
>ugliness led to activation in the right mOFC, in a region
adjoining rostro-ventral anterior cingulate cortex (ACC);
this entire zone has been reported to be active in previous
studies of aesthetic experiences [e.g., Ishizu and Zeki,
2011; Tsukiura and Cabeza, 2011; Zeki et al., 2014]. The
body of the right caudate nucleus and pregenual ACC
(pgACC), which have been found to be active in the expe-
rience of visual beauty and aesthetic and evaluative judg-
ments [Cunningham et al., 2004; Ishizu and Zeki, 2011;
Jacobsen et al., 2006; Vartanian and Goel, 2004], were also
active. In addition, there were activations in bilateral pos-
terior hippocampus and parts of the cerebellum (lobule
VII crus I and II) (see Figs. 5, 7, and 8).
In short, in addition to the activation within the mOFC,
we observed a similar pattern of cortical and subcortical
activations in the contrast of joyful beauty versus ugliness
to that reported in previous studies of visual beauty.
Areas uniquely active for each kind of beauty. The con-
trast sorrowful >joyful beauty produced activation in: the
left inferior parietal lobe (IPL) encroaching on the precu-
neus, a region which has often been observed in studies of
emotional/social pain (e.g., Lamm et al., 2011); it also pro-
duced activity in bilateral MFG, parts of which have been
found to be involved during emotional states [Acevedo
et al., 2014; Sabatinelli et al., 2011].
The reverse contrast, of joyful >sorrowful beauty led to
activation in the right temporoparietal junction (TPJ),
including the supramarginal gyrus (SMG), a region
thought to be involved in controlling empathy toward
others [e.g., Silani et al., 2013] (Fig. 6).
In summary, the experience of joyful and sorrowful
beauty had, as correlates, distinct patterns of cortical activ-
ity. The former included the TPJ and the SMG alone while
the latter included the left parietal lobe, the precuneus,
and bilateral MFG.
Areas commonly active during the experience of both types
of beauty. The conjunction analysis ([sorrowful beau-
ty >ugliness] \[joyful beauty >ugliness]) showed com-
mon activation in the mOFC with the application of SVC,
and left cerebellum (lobule VII), whereas the ACC, which
Figure 4.
Sites active during the experience of sorrowful beauty. Statistical
parametric maps rendered onto canonical anatomical sections
showing the t-statistic for the contrast sorrowful beauty >ugli-
ness. Random effects analysis with 20 subjects. Display threshold
P<0.001 (uncorrected). Abbreviations: IFG, frontal gyrus; MFG,
middle frontal gyrus; dACC, dorsal anterior cingulate cortex;
PCC, posterior cingulate cortex. [Color figure can be viewed at
wileyonlinelibrary.com]
Figure 5.
Sites active during the experience of joyful beauty. Statistical
parametric maps rendered onto canonical anatomical sections
showing the t-statistic for the contrast joyful beauty >ugliness.
Random effects analysis with 20 subjects. Display threshold
P<0.001 (uncorrected). Abbreviation: dACC, dorsal anterior
cingulate cortex. [Color figure can be viewed at wileyonlineli-
brary.com]
rExperience of Sorrowful Beauty r
r4193 r

was active in both conditions, did not survive this thresh-
old in the conjunction analysis (Fig. 3).
Functional Connectivity
The conjunction analysis revealed that the mOFC is
engaged during the experience of both sorrowful and joy-
ful beauty. We undertook a PPI analysis to learn more
about the possible differential neural engagements during
the experience of the two types of beauty, by examining
the functional connectivity between the mOFC and other
brain regions. Because the stimuli experienced as being
sorrowful contained, for the most part, pitiful and empa-
thetic scenes (such as pictures of funerals, an abandoned
child, a crying man, separations, etc.), we expected that, in
this condition, the mOFC will show greater functional con-
nectivity with regions related to the experience of sad
(negative) emotional empathy and the interpretation of
others’ intentions, especially under sad or painful situa-
tions (‘negative empathy’ as opposed to ‘positive empathy’
[Morelli et al., 2015]); these regions are the middle cingu-
late cortex (MCC), the supplementary motor area (SMA)
[Fan et al., 2011], and the dlPFC [Lieberman, 2007 for a
review; Weissman et al., 2008]. Using sorrowful beauty
and ugliness as the psychological parameters, we tested
whether the physiological coupling between mOFC, as the
seed region, and other regions besides the ones mentioned
above, would change between the experience of sorrowful
beauty and ugliness. The results showed that there was
indeed increased functional connectivity between the
mOFC and SMA, encroaching onto MCC, and several
other regions (see Fig. 9 and Table I). With the application
of SVC using the coordinates based on a previous PPI
study in the bilateral dlPFC [Kirk et al., 2011], dlPFC also
showed a significant increased connectivity with the
mOFC during the experience of sorrowful beauty. We
then compared the averaged b-value in the SMA/MCC
during the experience of sorrowful and joyful beauty and
found a stronger connectivity between the SMA/MCC and
the mOFC in sorrowful beauty relative to joyful beauty
(t(19) 54.03, P<0.01).
In contrast, the experience of joyful beauty produced
increased functional connectivity between the mOFC and
the anterior rostral medial prefrontal cortex (anterior
rMPFC) and middle temporal gyrus (MTG), among several
other regions (see Table I). The averaged b-value in the
rMPFC showed stronger connectivity in joyful beauty than
sorrowful beauty (t(19) 55.31, P<0.01). All other PPI
results are listed in Table I.
DISCUSSION
The experience of beauty may, in general terms, be
regarded as a positive, rewarding, and pleasurable one. It
is perhaps not surprising, therefore, that regardless of
source, it correlates with activity in field A1 of mOFC [Ish-
izu and Zeki, 2011], a region of the emotional brain which
has been generally associated with pleasure and reward
(e.g., O’Doherty et al., 2001); activity in it has been shown
to correlate parametrically with the declared intensity of
the experience of beauty derived from a variety of stimuli,
such as faces, colors, motion, paintings, music, architec-
tures, moral judgments, and mathematical equations
[Ikeda et al., 2015; Ishizu and Zeki, 2011; Kawabata and
Zeki, 2004; Kuhn and Gallinat, 2012; O’Doherty et al.,
2003; Zeki and Stutters, 2012; Tsukiura and Cabeza, 2011;
Vartanian et al., 2013 for a meta-analysis; Zeki et al., 2014]
though apparently not from the performing (dance) arts
[e.g., Calvo-Merino, et al., 2008; Cross et al., 2011]. More-
over, a recent study has reported an increase in aesthetic
ratings of visual stimuli [Nakamura and Kawabata, 2015]
following the application of anodal transcranial direct cur-
rent stimulation to the mOFC, presumably because of
enhanced neural activity within it.
In the work reported here, we wanted to go a step
beyond and enquire into whether the experience of beauty
linked to different and indeed opposite emotional states
would also correlate with activity in A1 of mOFC. Sorrow-
ful beauty is commonly associated with negative empathy
Figure 6.
Sites revealed to be active in the contrast sorrowful versus joy-
ful beauty. Statistical parametric maps rendered onto canonical
anatomical sections showing the t-statistic for (upper) the con-
trasts joyful beauty >sorrowful beauty and (lower) sorrowful
beauty >joyful beauty. Random effects analysis with 20 subjects.
Display threshold P<0.001 (uncorrected). Abbreviations: rTPJ,
right temporoparietal junction; IPL, inferior parietal lobe; SPL,
superior parietal lobe; MFG, middle frontal gyrus. [Color figure
can be viewed at wileyonlinelibrary.com]
rIshizu and Zeki r
r4194 r

while joyful beauty is linked to positive empathy. The two
experiences studied here are, thus, associated with oppo-
site empathetic sources but share a common denominator,
that of beauty. Given that field A1 of mOFC correlates
with the experience of beauty regardless of source, it was
natural to hypothesize that the experience of beauty linked
to emotional states of opposite valence would also corre-
late with activity in the same area, which is indeed what
we found. But the intensity of activity in A1 of mOFC
(defined as the ROI) was greater during the experience of
beauty derived from joy than that derived from sorrow.
This made it interesting to enquire into the strength of
connectivity between the mOFC and areas of the brain
that have been associated with experience of two opposite
states of empathy.
Negative and Positive Empathy Reflected in the
Pattern of Brain Activity
Empathy has been studied in fair detail recently but
most do not refer explicitly to a distinction between nega-
tive and positive empathy. That the two are separate is, of
course, a common human experience; it is reflected here in
the distinct patterns of activation that correlate with the
experience of sorrowful and joyful beauty, besides the
common correlate in activity of A1 of mOFC. This distinc-
tion can be discerned in (a) the general pattern of cortical
activity, (b) in activity within the anterior cingulate cortex,
and (c) in the cerebellum.
a. General cortical activation patterns. Contrasting the pat-
tern produced by the experience of sorrowful beauty with
that produced by the experience of ugliness resulted in an
extensive pattern of activity that includes, in addition to
the reward-related regions, bilateral MFG, extending to
IFG, right dlPFC and PCC; all three areas are known to be
active when empathizing with others, especially in nega-
tive emotional conditions. The IFG, in particular, has been
reported to be active during listening to minor chords
(rated as sad and indicative of sorrow) compared to major
ones (which were rated as ‘happy’), even though both
were rated as aesthetically beautiful by the subjects
[Suzuki et al., 2008]. This region has also been reported to
be active when viewing pictures of humans suffering from
harm and threat [Nummenmaa et al., 2008], while the
MFG has been reported to be active during the viewing of
sad faces [Acevedo et al., 2014; Sabatinelli et al., 2011]. The
dlPFC has been linked to handling complex social situa-
tions [Lieberman, 2007 for a review; Weissman et al.,
2008], control of emotional states [Goldin et al., 2008;
Keightley et al., 2003], inferring others’ intentions, and the-
ory of mind [Guroglu et al., 2011]. The PCC, encroaching
upon the caudate (head and body), was also active; this
region has been implicated in a range of functions includ-
ing the experience of high valence emotional stimuli [Mad-
dock et al., 2003], theory of mind [Fletcher et al., 1995;
Greene et al., 2001], and sad autobiographical recall
[Farrow et al., 2001; Maddock, 1999]. Activity in the IPL,
also observed with the experience of sorrowful beauty
when contrasted with joyful beauty, has been associated
with emotional or social pain and, together with the IFG,
has been considered as constituting an ‘emotion contagion
network’ underlying our ability to empathize emotionally
[Shamay-Tsoory, 2011], though negatively.
By contrast, the experience of joyful beauty correlated
with activity in the right TPJ and in the SMG, both of
which have been considered to be involved in controlling
empathy toward others, by overriding an emphasis on the
self (ego-centricity) [e.g., Silani et al., 2013]. Activity in the
TPJ, which has been considered to play an important role
in interpersonal emotional and cognitive interactions
[Saxe, 2006, for a review], has been reported to increase
when subjects view happy faces compared to angry or dis-
appointed ones [Lelieveld et al., 2013] and both TPJ and
SMG have been demonstrated to be part of a larger corti-
cal zone, which includes the right parietal area, that is
active when adopting other peoples’ emotional states [e.g.,
Ruby and Decety, 2004].
b. Anterior cingulate cortex: Activation within the ACC
found during the experience of sorrowful and joyful
beauty can be separated into dorsal (dACC) for the former
and the pregenual subdivision (pgACC) for the latter (see
Fig. 7). It has been suggested that the dACC is active dur-
ing the experience of emotionally distressing conditions
Figure 7.
Sites active within the ACC during the experience of sorrowful
and joyful beauty. Statistical parametric maps rendered onto
canonical anatomical sections showing the t-statistic for (red)
the contrasts joyful beauty >ugliness and (blue) sorrowful beau-
ty >ugliness. Random effects analysis with 20 subjects. Display
threshold P<0.001 (uncorrected). Abbreviations: dACC, dorsal
anterior cingulate cortex; pgACC, perigenual anterior cingulate
cortex. [Color figure can be viewed at wileyonlinelibrary.com]
rExperience of Sorrowful Beauty r
r4195 r

such as physical and social pain (acknowledging others’
pain) [Eisenberger and Lieberman, 2004; Lamm et al., 2011
for a review], whereas the pgACC activity (including
activity in the adjacent subgenual ACC and mOFC/ven-
tromedial prefrontal cortex (vmPFC)) correlates with the
experience of positive emotions [Etkin et al., 2011 for a
review]. A previous study reports activity in this region
when viewing aesthetically pleasing stimuli [e.g., Varta-
nian et al., 2013]. ACC’s diverse cognitive and emotional
functions make it difficult to define each subdivision’s
involvement in a precise function. But previous studies
suggest that the ventral and sub/pregenual areas are
involved in processing of emotion, especially positive emo-
tion while the dorsal subdivision, by contrast, is strongly
associated with negative emotional states as well as cogni-
tive components. This separation, again, reflects anatomi-
cally the involvement of separate regions in positive and
negative components of aesthetic experiences.
c. Cerebellum: The dichotomy in neural activity that cor-
relates with the two contrasting experiences is also
reflected in cerebellar activity. We found that the experi-
ence of sorrowful and joyful beauty engaged different
parts of the cerebellum, lobule VI for sorrowful beauty
and lobule VII for joyful beauty (see Fig. 8). Although
there is no current consensus regarding the pattern of cer-
ebellar activity during aesthetic experiences, several past
studies have reported activity in cerebellum during aes-
thetic experiences, including ones derived from visual and
literary beauty as well as during the judgment of beauty
[Bohrn et al., 2013; Ishizu and Zeki, 2013; Vartanian and
Goel, 2004]. It is noteworthy that sorrowful beauty
engaged lobule VI, which past studies have shown to be
more responsive to negatively charged stimuli such as sad-
ness, fear, and anger than to positive ones [Baumann and
Figure 8.
Sites active within the cerebellum during the experience of sor-
rowful and joyful beauty. Statistical parametric maps rendered
onto canonical anatomical sections showing the t-statistic for
(red) the contrasts joyful beauty >ugliness and (blue) sorrowful
beauty >ugliness. Random effects analysis with 20 subjects. Dis-
play threshold P<0.001 (uncorrected). [Color figure can be
viewed at wileyonlinelibrary.com]
Figure 9.
The upper panel shows the functional connectivity observed in
this study, with the mOFC as the seed region. Areas in orange
(anterior rMPFC) showed greater functional connectivity with
the mOFC during the experience of joyful beauty, whereas areas
shown in blue (SMA/MCC) showed greater connectivity during
the experience of sorrowful beauty. Display threshold P<0.001
(uncorrected). Lower panel (left) shows the averaged beta-
estimates measuring the correlation between BOLD activity in
the SMA/MCC and the mOFC. Lower panel (right) shows the
averaged beta-estimates measuring the correlation between
BOLD activity in the anterior rMPFC and the mOFC. Blue bars
denote the averaged beta-estimates with sorrowful beauty and
red bars denote those with joyful beauty. Error bars are stan-
dard error (SE). Abbreviations: SMA, supplementary motor area;
MCC, middle cingulate cortex; rMPFC, rostral medial prefrontal
cortex; mOFC, medial orbito-frontal cortex. [Color figure can
be viewed at wileyonlinelibrary.com]
rIshizu and Zeki r
r4196 r

Mattingley, 2012; Park et al., 2010] and to perspective tak-
ing to others’ pain [Lamm et al., 2007], which can be
regarded as negative empathy in the context of the current
study. Although the cerebellum is, in general, more active
during what may be regarded as negative emotions
[Stoodley, 2012], crus II (lobule VII) has been reported to
be more strongly engaged when viewing pictures repre-
senting happiness than disgust [Schienle and Scharmuller,
2013] and is uniquely active with joyful beauty in the cur-
rent experiment.
Patient studies have suggested that lesions in lobule VI
and VII can lead to cerebellar cognitive affective syn-
drome, when patients suffer from various emotional and
behavioral deficits, including flattening of emotions or
impulsive behaviors [Schmahmann, 2004; Schmahmann
et al., 2007]. It has been suggested that those deficits in
emotion, possibly produced by disrupting the cerebellar-
limbic connection [Stoodley and Schmahmann, 2009], can
affect the ability to communicate and empathize with the
perspectives of others [Oberman and Ramachandran,
2007].
Hence, both cortical and cerebellar activations reflect, in
a sense, the common human experience which can sepa-
rate sorrowful from joyful beauty. Even though this is a
distinction that is not commonly made or emphasized in
philosophies of aesthetics, it is, nevertheless, one that cor-
relates with distinct patterns of activation during the expe-
rience of the two different kinds of beauty.
Functional connectivity
Given this dichotomy, we naturally expected that some of
the areas that were active above would show a positive,
state-dependent positive connectivity with mOFC. It is
known that activity in the mOFC (and vmPFC) can be mod-
ulated by signals from other brain regions [Harvey et al.,
2010]. The areas showing a greater functional connectivity
with the mOFC during the experience of sorrowful beauty
were the SMA, MCC, and dlPFC, regions related to the
experience of sad (negative) emotional empathy and the
interpretation of others’ intentions, especially under sad or
painful situations [Lieberman, 2007 for a review; Fan et al.,
2011; Weissman et al., 2008]. One study using diffusion-
weighted and functional MRI showed a direct connection
between the SMA and OFC area [Johansen-Berg et al., 2004].
It is known that aesthetic judgments under the influence of
monetary value and sponsorship to artworks lead to
increased functional connectivity between the mOFC and
the dlPFC, activity in the former being influenced by that of
the latter [Harvey et al., 2010; Kirk et al., 2011]. Hare et al.,
[2009, 2010] reported that the dlPFC modulated value sig-
nals encoded in the vmPFC/mOFC when subjects were
given information about the health status of a food item and
conducted self-controlling dietary choice.
Joyful beauty, in contrast, did not exhibit a functional con-
nectivity with the regions relating to negative empathy, but
showed a greater connectivity with anterior part of the
rMPFC, a region thought to be involved in mentalizing other
people’s psychological perspective [e.g., Amodio and Frith,
2006 for a review; Skerry and Saxe, 2015]. It is not clear
whether different patterns of activity in the rMPFC correlate
with the experience of positive and negative emotion in
mentalizing but some studies suggest that it responds more
to positive and aesthetically pleasing stimuli [Kreplin and
Fairclough, 2013; Vessel et al., 2012]. An enhanced func-
tional connection between mOFC/vmPFC, MPFC, and
MTG during the judgment of facial attractiveness has been
reported in a previous PPI study [Smith et al., 2014].
It has been suggested that the mOFC encodes the aes-
thetic value on the basis of a common neural scale regard-
less of its source [e.g., Ishizu and Zeki, 2011; Pegors et al.,
2015; Zeki et al., 2014], leading to an enquiry on how rele-
vant information, such as emotional context, modulates
perceived aesthetic value and how such a modulation is
represented in neural terms, besides activity within the
mOFC [Pegors et al., 2015]. The findings from recent studies,
some mentioned above, have suggested that the mOFC inter-
acts with other brain systems during evaluation of reward
values in a ‘context-dependent’ manner [Levy and Glimcher,
2012; Smith et al., 2014]. We have revealed that there is an
increased functional connectivity between the mOFC and
anterior rMPFC in joyful beauty and between the mOFC and
the SMA/MCC and the dlPFC in sorrowful beauty. This find-
ing suggests that (1) the brain engages two specialised sys-
tems, a reward-related one (the mOFC) and empathy-related
regions; these are dissociable from each other depending
upon whether the experience has a positive or negative emo-
tional valance (the SMA/MCC and rMPFC), and (2) the
empathy-related regions may modulate activity within the
mOFC through functional connectivity, to enable us to experi-
ence the contradictory aesthetic and emotional values.
In summary, therefore, functional connectivity between
mOFC and other cortical areas during the experience of
beauty is dictated by whether the experienced beauty is
joyful or sorrowful.
It is interesting to discuss briefly the neural correlates of
pleasure evoked by listening to sad music, which is also
regarded as being a positive aesthetic experience with a neg-
ative emotional valence [Schubert, 1996]. Such a contradic-
tory experience can be seen in many forms of art, including
paintings and films [e.g., Hanich et al., 2014; Leder et al.,
2014] but is most notable and relatively well studied in
music. Among previous behavioral and neuroimaging stud-
ies on sad music [e.g., Kawakami et al., 2013; Suzuki et al.,
2008; Taruffi and Koelsch, 2014], a recent one [Sachs et al.,
2015] has argued that sad music is found pleasurable when
(1) it is perceived as non-life-threatening and with, no
immediate real life implication; (2) it is aesthetically pleas-
ing; and (3) it has certain psychological benefits, such as
mood regulation caused by recollection of personal past
events, which leads to activation within hippocampus/par-
ahippocampal gyrus. We did not find hippocampal activity
in the sorrowful beauty condition in the current study;
rExperience of Sorrowful Beauty r
r4197 r

instead, we found activation within SMA/MCC, indicating
empathy and perspective taking toward other people. This
may be due to a difference in the nature of visual and musi-
cal perception. With visual stimuli, viewers can immediately
empathize with sufferers or wounded people depicted in an
image. By contrast, music, having no figurative representa-
tion, may make listeners adopt a more ‘self-referential
mode’ and recall personal-relevant memories. This points to
possible interesting future studies, of how the brain reacts to
the experience of sadness evoked by different sources.
CONCLUSION
It is gratifying to us that an inspiration derived from a
literary source should have led to work which has given
us a little, but not much, more knowledge about the brain
mechanisms that are engaged during aesthetic experiences.
ACKNOWLEDGMENT
We thank Yen Yu for his helpful comments. This study
was supported by the Wellcome Trust, London.
CONFLICT OF INTEREST
The authors declare no conflict of interest.
REFERENCES
Acevedo BP, Aron EN, Aron A, Sangster MD, Collins N, Brown
LL (2014): The highly sensitive brain: An fMRI study of sen-
sory processing sensitivity and response to others’ emotions.
Brain Behav 4:580–594.
Amodio DM, Frith CD (2006): Meeting of minds: The medial fron-
tal cortex and social cognition. Nat Rev Neurosci 7:268–277.
Baumann O, Mattingley JB (2012): Functional topography of pri-
mary emotion processing in the human cerebellum. Neuro-
image 61:805–811.
Bohrn IC, Altmann U, Lubrich O, Menninghaus W, Jacobs AM
(2013): When we like what we know–a parametric fMRI analy-
sis of beauty and familiarity. Brain Lang 124:1–8.
Calvo-Merino B, Jola C, Glaser DE, Haggard P (2008): Towards a
sensorimotor aesthetics of performing art. Conscious Cogn 17:
911–922.
Cross ES, Kirsch L, Ticini LF, Schutz-Bosbach S (2011): The impact
of aesthetic evaluation and physical ability on dance percep-
tion. Front Hum Neurosci 5:102.
Cunningham WA, Raye CL, Johnson MK (2004): Implicit and
explicit evaluation: fMRI correlates of valence, emotional inten-
sity, and control in the processing of attitudes. J Cogn Neuro-
sci 16:1717–1729.
Eisenberger NI, Lieberman MD (2004): Why rejection hurts: A
common neural alarm system for physical and social pain.
Trends Cogn Sci 8:294–300.
Etkin A, Egner T, Kalisch R (2011): Emotional processing in ante-
rior cingulate and medial prefrontal cortex. Trends Cogn Sci
15:85–93.
Fan Y, Duncan NW, de Greck M, Northoff G (2011): Is there a
core neural network in empathy? An fMRI based quantitative
meta-analysis. Neurosci Biobehav Rev 35:903–911.
Farrow TF, Zheng Y, Wilkinson ID, Spence SA, Deakin JF, Tarrier
N, Griffiths PD, Woodruff PW (2001): Investigating the func-
tional anatomy of empathy and forgiveness. Neuroreport 12:
2433–2438.
Fletcher PC, Happe F, Frith U, Baker SC, Dolan RJ, Frackowiak
RS, et al. (1995): Other minds in the brain: A functional imag-
ing study of “theory of mind” in story comprehension. Cogni-
tion 57:109–128.
Friston KJ, Buechel C, Fink GR, Morris J, Rolls E, Dolan RJ (1997):
Psychophysiological and modulatory interactions in neuroim-
aging. Neuroimage 6:218–229.
Goldin PR, McRae K, Ramel W, Gross JJ (2008): The neural bases
of emotion regulation: Reappraisal and suppression of nega-
tive emotion. Biol Psychiatry 63:577–586.
Greene JD, Sommerville RB, Nystrom LE, Darley JM, Cohen JD
(2001): An fMRI investigation of emotional engagement in
moral judgment. Science 293:2105–2108.
Guroglu B, van den Bos W, van Dijk E, Rombouts SA, Crone EA
(2011): Dissociable brain networks involved in development of
fairness considerations: Understanding intentionality behind
unfairness. Neuroimage 57:634–641.
Hanich J, Wagner V, Shah M, Jacobsen T, Menninghaus W
(2014): Why we like to watch sad films. The pleasure of being
moved in aesthetic experiences. Psychol Aesthet Creat Arts 8:
130–143.
Hare TA, Camerer CF, Rangel A (2009): Self-control in decision-
making involves modulation of the vmPFC valuation system.
Science 324:646–648.
Hare TA, Camerer CF, Knoepfle DT, Rangel A (2010): Value com-
putations in ventral medial prefrontal cortex during charitable
decision making incorporate input from regions involved in
social cognition. J Neurosci 30:583–590.
Harvey AH, Kirk U, Denfield GH, Montague PR (2010): Monetary
favors and their influence on neural responses and revealed
preference. J Neurosci 30:9597–9602.
Hutton C, Bork A, Josephs O, Deichmann R, Ashburner J, Turner
R (2002): Image distortion correction in fMRI: A quantitative
evaluation. Neuroimage 16:217–240.
Ikeda T, Matsuyoshi D, Sawamoto N, Fukuyama H, Osaka N
(2015): Color harmony represented by activity in the medial
orbitofrontal cortex and amygdala. Front Hum Neurosci 9:382.
Ishizu T, Zeki S (2011): Toward a brain-based theory of beauty.
PLoS One 6:e21852.
Ishizu T, Zeki S (2013): The brain’s specialized systems for
aesthetic and perceptual judgment. Eur J Neurosci 37:
1413–1420.
Ishizu T, Zeki S (2014): A neurobiological enquiry into the origins
of our experience of the sublime and beautiful. Front Hum
Neurosci 8:891.
Jacobsen T, Schubotz RI, Hofel L, Cramon DY (2006): Brain corre-
lates of aesthetic judgment of beauty. Neuroimage 29:276–285.
Johansen-Berg H, Behrens TE, Robson MD, Drobnjak I, Rushworth
MF, Brady JM, et al. (2004): Changes in connectivity profiles
define functionally distinct regions in human medial frontal cor-
tex. Proc Natl Acad Sci USA 101:13335–13340.
Kawabata H, Zeki S (2004): Neural correlates of beauty.
J Neurophysiol 91:1699–1705.
Kawakami A, Furukawa K, Katahira K, Okanoya K (2013): Sad
music induces pleasant emotion. Front Psychol 4:311.
rIshizu and Zeki r
r4198 r

Keightley ML, Winocur G, Graham SJ, Mayberg HS, Hevenor SJ,
Grady CL (2003): An fMRI study investigating cognitive mod-
ulation of brain regions associated with emotional processing
of visual stimuli. Neuropsychologia 41:585–596.
Kelly CR, Grinband J, Hirsch J (2007): Repeated exposure to
media violence is associated with diminished response in an
inhibitory frontolimbic network. PLoS One 2:e1268.
Kirk U, Harvey A, Montague PR (2011): Domain expertise insu-
lates against judgment bias by monetary favors through a
modulation of ventromedial prefrontal cortex. Proc Natl Acad
Sci USA 108:10332–10336.
Kreplin U, Fairclough SH (2013): Activation of the rostromedial
prefrontal cortex during the experience of positive emotion in
the context of esthetic experience. An fNIRS study. Front Hum
Neurosci 7:879.
Kuhn S, Gallinat J (2012): The neural correlates of subjective pleas-
antness. Neuroimage 61:289–294.
Lamm C, Batson CD, Decety J (2007): The neural substrate of
human empathy: effects of perspective-taking and cognitive
appraisal. J Cogn Neurosci 19:42–58.
Lamm C, Decety J, Singer T (2011): Meta-analytic evidence for com-
mon and distinct neural networks associated with directly expe-
rienced pain and empathy for pain. Neuroimage 54:2492–2502.
Leder H, Gerger G, Brieber D, Schwarz N (2014): What makes an
art expert? Emotion and evaluation in art appreciation. Cogn
Emot 28:1137–1147.
Lelieveld GJ, Van Dijk E, Guroglu B, Van Beest I, Van Kleef GA,
Rombouts SA, et al. (2013): Behavioral and neural reactions to
emotions of others in the distribution of resources. Soc Neuro-
sci 8:52–62.
Levy DJ, Glimcher PW (2012): The root of all value: A neural com-
mon currency for choice. Curr Opin Neurobiol 22:1027–1038.
Lieberman MD (2007): Social cognitive neuroscience: A review of
core processes. Annu Rev Psychol 58:259–289.
Maddock RJ (1999): The retrosplenial cortex and emotion: New
insights from functional neuroimaging of the human brain.
Trends Neurosci 22:310–316.
Maddock RJ, Garrett AS, Buonocore MH (2003): Posterior cingu-
late cortex activation by emotional words: fMRI evidence from
a valence decision task. Hum Brain Mapp 18:30–41.
Morelli SA, Lieberman MD, Zaki J (2015): The emerging study of
positive empathy. Soc Personal Psychol Compass 9:57–68.
Nakamura K, Kawabata H (2015): Transcranial direct current
stimulation over the medial prefrontal cortex and left primary
motor cortex (mPFC-lPMC) affects subjective beauty but not
ugliness. Front Hum Neurosci 9:654.
Nichols T, Brett M, Andersson J, Wager T, Poline JB (2005): Valid
conjunction inference with the minimum statistic. Neuroimage
25:653–660.
Nummenmaa L, Hirvonen J, Parkkola R, Hietanen JK (2008): Is
emotional contagion special? An fMRI study on neural systems
for affective and cognitive empathy. Neuroimage 43:571–580.
Oberman LM, Ramachandran VS (2007): The simulating social
mind: The role of the mirror neuron system and simulation in
the social and communicative deficits of autism spectrum dis-
orders. Psychol Bull 133:310–327.
O’Doherty J, Kringelbach ML, Rolls ET, Hornak J, Andrews C
(2001): Abstract reward and punishment representations in the
human orbitofrontal cortex. Nat Neurosci 4:95–102.
O’Doherty J, Winston J, Critchley H, Perrett D, Burt DM, Dolan RJ
(2003): Beauty in a smile: The role of medial orbitofrontal cor-
tex in facial attractiveness. Neuropsychologia 41:147–155.
Park JY, Gu BM, Kang DH, Shin YW, Choi CH, Lee JM, et al.
(2010): Integration of cross-modal emotional information in the
human brain: An fMRI study. Cortex 46:161–169.
Pegors TK, Kable JW, Chatterjee A, Epstein RA (2015): Common
and unique representations in pFC for face and place attrac-
tiveness. J Cogn Neurosci 27:959–973.
Ruby P, Decety J (2004): How would you feel versus how do
you think she would feel? A neuroimaging study of
perspective-taking with social emotions. J Cogn Neurosci 16:
988–999.
Sabatinelli D, Fortune EE, Li Q, Siddiqui A, Krafft C, Oliver
WT, et al. (2011): Emotional perception: Meta-analyses of
face and natural scene processing. Neuroimage 54:
2524–2533.
Sachs M, Damasio A, Habibi A (2015): The pleasure of sad music:
A systematic review. Front Hum Neurosci 9:404.
Saxe R (2006): Uniquely human social cognition. Curr Opin Neu-
robiol 16:235–239.
Schienle A, Scharmuller W (2013): Cerebellar activity and connec-
tivity during the experience of disgust and happiness. Neuro-
science 246:375–381.
Schmahmann JD (2004): Disorders of the cerebellum: Ataxia,
dysmetria of thought, and the cerebellar cognitive affec-
tive syndrome. J Neuropsychiatry Clin Neurosci 16:
367–378.
Schmahmann JD, Weilburg JB, Sherman JC (2007): The neuropsy-
chiatry of the cerebellum - insights from the clinic. Cerebellum
6:254–267.
Schubert E (1996): Enjoyment of negative emotions in music: An
associative network explanation. Psychol Music 24:18–28.
Shamay-Tsoory SG (2011): The neural bases for empathy. Neuro-
scientist 17:18–24.
Silani G, Lamm C, Ruff CC, Singer T (2013): Right supramarginal
gyrus is crucial to overcome emotional egocentricity bias in
social judgments. J Neurosci 33:15466–15476.
Skerry AE, Saxe R (2015): Neural representations of emotion are
organized around abstract event features. Curr Biol 25:
1945–1954.
Smith DV, Clithero JA, Boltuck SE, Huettel SA (2014): Functional
connectivity with ventromedial prefrontal cortex reflects sub-
jective value for social rewards. Soc Cogn Affect Neurosci 9:
2017–2025.
Stoodley CJ (2012): The cerebellum and cognition: Evidence from
functional imaging studies. Cerebellum 11:352–365.
Stoodley CJ, Schmahmann JD (2009): Functional topography in
the human cerebellum: A meta-analysis of neuroimaging stud-
ies. Neuroimage 44:489–501.
Suzuki M, Okamura N, Kawachi Y, Tashiro M, Arao H,
Hoshishiba T, et al. (2008): Discrete cortical regions associated
with the musical beauty of major and minor chords. Cogn
Affect Behav Neurosci 8:126–131.
Taruffi L, Koelsch S (2014): The paradox of music-evoked sadness:
An online survey. PLoS One 9:e110490.
Tsukiura T, Cabeza R (2011): Shared brain activity for aesthetic
and moral judgments: Implications for the Beauty-is-Good ste-
reotype. Soc Cogn Affect Neurosci 6:138–148.
Vartanian O, Goel V (2004): Neuroanatomical correlates of aes-
thetic preference for paintings. Neuroreport 15:893–897.
Vartanian O, Navarrete G, Chatterjee A, Fich LB, Leder H,
Modrono C, et al. (2013): Impact of contour on aesthetic judg-
ments and approach-avoidance decisions in architecture. Proc
Natl Acad Sci USA 110(Suppl 2):10446–10453.
rExperience of Sorrowful Beauty r
r4199 r

Vessel EA, Starr GG, Rubin N (2012): The brain on art: Intense
aesthetic experience activates the default mode network. Front
Hum Neurosci 6:66.
Wang T, Mo L, Mo C, Tan LH, Cant JS, Zhong L, et al.
(2015): Is moral beauty different from facial beauty? Evi-
dence from an fMRI study. Soc Cogn Affect Neurosci 10:
814–823.
Weiskopf N, Hutton C, Josephs O, Deichmann R (2006): Optimal
EPI parameters for reduction of susceptibility-induced BOLD
sensitivity losses: A whole-brain analysis at 3 T and 1.5 T.
Neuroimage 33:493–504.
Weissman DH, Perkins AS, Woldorff MG (2008): Cognitive control
in social situations: A role for the dorsolateral prefrontal cor-
tex. Neuroimage 40:955–962.
Zeki S, Stutters J (2012): A brain-derived metric for preferred
kinetic stimuli. Open Biol 2:120001.
Zeki S, Watson JD, Lueck CJ, Friston KJ, Kennard C, Frackowiak
RS (1991): A direct demonstration of functional specialization
in human visual cortex. J. Neurosci 11:641–649.
Zeki S, Romaya JP, Benincasa DM, Atiyah MF (2014): The experi-
ence of mathematical beauty and its neural correlates. Front
Hum Neurosci 8:68.
rIshizu and Zeki r
r4200 r