Culture modulates brain activity during empathy with anger

34 Pages·2011·1.12 MB·English

Culture modulates brain activity
during empathy with anger
Moritz de Greck1, Zhenhao Shi1, Gang Wang1,
Xiangyu Zuo1, Xuedong Yang2, Xiaoying Wang2,
Georg Northoff3, Shihui Han1
1 Department of Psychology, Peking University, Beijing, China
2 Department of Radiology, Peking University First Hospital, Bei-
jing, China
3 Institute of Mental Health Research, University of Ottawa, Otta-
wa, Canada
This is a free version of the paper NIMG-11-838R1
which has been peer-reviewed and accepted
by NeuroImage.
Corresponding authors:
Dr. med. Moritz de Greck
& Shihui Han, PhD
Department of Psychology
Peking University
5 Yiheyuan Road
100871 Beijing
China
Phone: [phone]
Fax: [phone]
Email: [email protected], [email protected]
1
Culture modulates empathy with anger
Abstract
Interdependent cultures (such as regions, including the left inferior
the Chinese) and independent cul- frontal cortex, the left supplemen-
tures (such as the German) differ tary motor area, and the left insula,
in their attitude towards harmony that showed comparable hemody-
that is more valued in interdepen- namic responses in both groups.
dent cultures. Interdependent and However, the left dorsolateral pre-
independent cultures also differ in frontal cortex had stronger hemo-
their appreciation of anger - an dynamic responses in Chinese sub-
emotion that implies the disrup- jects in the contrast ‘intentional
tion of harmony. empathy for familiar angry’ > ‘ba-
seline’. Germans, in contrast, sho-
The present study investigated if wed stronger hemodynamic respon-
interdependent and independent ses in the right temporo-parietal
cultures foster distinct brain acti- junction, right inferior and superi-
vity associated with empathic pro- or temporal gyrus, and left middle
cessing of familiar angry, familiar insula for the same contrast. He-
neutral, and unfamiliar neutral fa- modynamic responses in the lat-
ces. ter three brain regions correlated
Using functional MRI, we scanned with interdependences scores over
Chinese and German healthy sub- all subjects.
jects during an intentional empathy
task, a control task (the evaluati- Our results suggest that enhanced
on of skin color), and a baseline emotion regulation during empathy
condition. The subject groups were with anger in the interdependent li-
matched with regard to age, gen- festyle is mediated by the left dor-
der, and education. solateral prefrontal cortex. Increa-
sed tolerance towards the expres-
Behaviorally, Chinese subjects de- sion of anger in the independent
scribed themselves as significantly lifestyle, in contrast, is associated
more interdependent compared to with increased activity of the right
German subjects. The contrast ‘in- inferior and superior temporal gy-
tentional empathy for familiar an- rus and the left middle insula.
gry’ > ‘baseline’ revealed several
Key words:
fmri, transcultural, emotion, empa-
thy, anger
2
Culture modulates empathy with anger
Introduction
Interdependence describes a lifestyle, in conducted by Suh et al. Germans sho-
which an individual is attuned to the wed the strongest correlation of affective
social environment, adjusts his behavior well-being and life satisfaction.
to others, and takes the perspective of In addition, two studies showed that, re-
others. Independence, in contrast, refers lative to individuals from independent
to a lifestyle, in which an individual pri- cultures, individuals in interdependent
marily refers to his own thoughts and cultures can tolerate disharmony less
feelings (Markus and Kitayama, 2010; and are more prone to react with depres-
Chiao et al., 2009). Typically, most Asi- sive reactions to negative social events
an cultures engage an interdependent (Tafarodi and Smith, 2001; Chen et al.,
lifestyle, whereas in most Western cul- 2006). Individuals from interdependent
tures an independent lifestyle is preva- cultures also seek less social support in
lent(MarkusandKitayama,1991;Chiao the case of stressing events in order to
et al., 2009). Interestingly, both lifesty- maintain social harmony (Taylor et al.,
lesdonotcompletelyexcludeeachother. 2004).
Rather, they can coexist to some degree
in one and the same individual (Singelis, Given the differential significance of
1994). Interdependent and independent harmony for interdependent and inde-
lifestylesareconsideredtoberesponsible pendent cultures, one might also expect
for a large amount of cultural differences cultural differences in the appreciation
in emotional experience, cognition and of anger (Kövecses, 2000) because a cru-
behavior (Markus and Kitayama, 1991). cial characteristic of anger is the disrup-
tion of harmony (Markus and Kitaya-
A key concept, which is appreciated very ma, 1991). Indeed, it is argued that the
differently in interdependent and inde- expression of anger is less prevalent in
pendent cultures, is harmony. Harmony, interdependent than in independent cul-
which describes the balance, positive tures (Markus and Kitayama, 1991). In
mood and social peace within a group, a study investigating the reaction of 11
is important in interdependent cultures month old infants towards vocal expres-
and often rooted in their cultural traditi- sions of emotions in the voice of their
ons (Markus and Kitayama, 1991, Uno, mothers, infants from interdependent
1991 as described in Kim and Markus, cultures reacted stronger to the vocal
1999). Independent cultures, in contrast, expression of anger (but not joy or fe-
rather stress the importance of uniquen- ar) compared to independent cultures
ess (Kim and Markus, 1999). (Miyake et al., 1986). The authors con-
For instance, as shown by Kwan et al. cluded, that the expression of anger hap-
(1997), relationship harmony is more pens less often in interdependent cultu-
important for life satisfaction in inter- res(andisrelatedtoextremesituations).
dependent cultures (in this study Hong Moreover, in interdependent cultures (in
Kong students) compared to indepen- this case China), the control of anger
dent cultures (namely US students). In is related to high social functioning of
independent cultures (such as the Uni- schoolchildren(Zhouetal.,2004).Inad-
ted states or Germany), in contrast, life dition, individuals from interdependent
satisfaction is closer related to the affec- cultures tolerate less anger. When anger
tive well-being of the individuals (Suh was expressed in simulated negotiations
et al., 1998). Interestingly, in the study (as part of recent study conducted by
3
Culture modulates empathy with anger
Adam and colleagues), Asians and Asi- self-otherdistinctionintheJapanesecul-
an Americans made smaller concessions. ture”.
In contrast, European Americans made
larger concessions (Adam et al., 2010). The current study investigated diffe-
Moreover, there are differences between rences in brain activity during em-
interdependent and independent cultu- pathy with anger between individuals
res concerning the suppression of anger. stemming from interdependent and in-
The suppression of anger can lead to dependent cultures. For this, we used
depression in individuals from interde- fMRI and an empathy paradigm, which
pendent and independent cultures; ho- was recently introduced by our group
wever,thelinkbetweensuppressedanger (de Greck et al., 2011). The paradigm
and depression is significantly stronger allowed us to investigate automatic as
in interdependent cultures (Cheung and well as intentional empathic responses,
Park, 2010; Park et al., 2010). and to control for the effects of emotion
Interestingly, there is a special psychia- and familiarity. (As recently shown by
tric disorder “hwa-byung” (English: “fire Xu et al. (2009), in particular familiari-
disease” or “anger disease”), which is ex- ty can significantly modulate empathic
plicitly associated with the suppression processes.) We recruited two groups of
of anger and strictly bound to the Ko- healthy subjects from an interdependent
rean (i.e. interdependent) culture (Min, culture (namely China) and from an in-
2008; Min et al., 2009). dependent culture (namely Germany).
With regard to the avoiding attitude to-
A suitable approach to investigate cultu- wards anger prevalent in interdependent
ral differences in emotional processing is cultures, we hypothesized less activity in
to implement an empathy task. Regar- Chinese subjects (compared to German
dingthis,empathyimpliesthecapability subjects) during empathy with anger in
to understand and share the emotional regions typically involved in emotional
states of other creatures without losing empathy and emotional processing such
the ability to differentiate between one’s as insula, anterior cingulate cortex, in-
own feelings and the feelings of others ferior frontal cortex and superior tem-
(Decety and Jackson, 2004; Preston and poral sulcus (Blair et al., 1999; Carr
de Waal, 2002). et al., 2003; de Greck et al., 2011; Hoo-
To our knowledge, no study examined ker et al., 2008, 2010; Jabbi et al., 2007;
cultural differences in emotional em- Jabbi and Keysers, 2008; Ochsner et al.,
pathy so far. One study however, ex- 2004a; Phillips et al., 1997; Sprengel-
plored differences in brain activity du- meyer et al., 1998; Wicker et al., 2003).
ring “Theory of Mind” (TOM) - a pa- In addition, we expected more activity
radigm, which is related to empathy - in Chinese in brain regions connected to
between interdependent and indepen- emotion regulation, such as the prefron-
dent individuals. Investigating American talcortex(MacDonaldetal.,2000;Ochs-
and Japanese children using a cartoon ner et al., 2004b; Ochsner and Gross,
TOMtask,Kobayashietal.(2007)found 2005; Vanderhasselt et al., 2006).
stronger hemodynamic responses in the Finally, considering the familiarity of
right temporo-parietal junction (TPJ) in empathy and TOM, we expected stron-
American children. The authors sugge- ger activity in the right TPJ in German
stedthatdiminishedTPJactivityduring subjects (Kobayashi et al., 2007).
TOM might reflect a “demoted sense of
4
Culture modulates empathy with anger
Methods
Participants Thestudywasapprovedbyalocalethics
A group of Chinese students (n=16) and committee. After a detailed explanation
agroupofGermanstudents(n=16)were of the study design and any potential
recruited in this study. Both groups we- risks, all subjects gave their written in-
re scanned in Beijing, China, using the formed consent. All subjects were reim-
same fMRI scanner. Table 1 illustrates bursed for their participation.
subjects’ information about the two cul-
tural groups.
Table 1
Chinese Germans
Number 16 16
Age mean 22.9 years mean 23.3 years1
95% CI2: 22.3 - 23.5 years 95% CI: 22.2 - 24.3 years
Gender 10 f / 6 m3 10 f / 6 m
Culture 16 Han Chinese 16 Germans
raised in China by raised in Germany by
Chinese parents German parents
Occupation 16 students 16 students
Table 1: Characteristics of the two subject groups
1 There was no significant difference regarding the age of both groups
(t(30)=0.681, p =0.501). 2 The 95% confidence interval. 3 In
[two-tailed]
addition, there was no significant difference with regard to the gender
distribution in both groups (χ2(1)=1, p=0.317).
Paradigm 7 blocks of 312 s duration each. Prior to
entering the scanner each subject read
Experimental design detailed information of the paradigm in
The fMRI experiment was divided into their native language and completed a
5
Culture modulates empathy with anger
couple of trial runs in order to familia- Ekman, 1988). Stimuli for German sub-
rize fully with the task. While lying in jects for the condition ‘unfamiliar neu-
the scanner, the stimuli were displayed tral’ were photographed and preproces-
using the software package ‘Presentati- sed for presentation by our own group
on’ (Neurobehavioral Systems, Albany, (de Greck et al., 2011). These pictures
CA, USA) and were projected onto a weretakeninfrontofacomparableback-
matte screen via an LCD projector, vi- groundandundercomparableconditions
sible through a mirror mounted on the to match them as close as possible to the
head coil. Each block started with 10s four pictures taken from the JACNeuF
pause to control for epi-saturation ef- battery.
fects. A total number of 24 trials (12 Stimuli for Chinese subjects for the con-
intentional empathy trials and 12 trials dition‘familiarneutral’wereexactlytho-
skin color evaluation trials) were pre- se stimuli, as presented to German sub-
sented in a randomized order in each jects in the condition ‘unfamiliar neu-
block. Figure 1 illustrates the intentio- tral’. Analogously, stimuli for Chinese
nal empathy task, the control task and subjects for the condition ‘unfamiliar
the baseline condition. neutral’ were exactly the same stimuli as
presentedtoGermansubjectsinthecon-
Stimuli dition ‘familiar neutral’ (i.e. these stimu-
Two sets of stimuli were used for Chine- li were taken from the JACNeuF-picture
se and German subjects. Each stimulus battery).StimuliforChinesesubjectsfor
set consisted of 12 different face stimuli the condition ‘familiar angry’ were also
- four stimuli (two female, two male) of photographed and preprocessed for pre-
each condition (namely ‘familiar angry’, sentation by our own group, as described
‘familiar neutral’, and ‘unfamiliar neu- above. Each stimulus was presented twi-
tral’). ce during each block: once during inten-
Stimuli for Germans subjects for the tional empathy, once during skin color
conditions ‘familiar angry’ and ‘familiar evaluation.
neutral’ were taken from the “Japane- The term “familiar” as used in our study
se and Caucasian Facial Expressions of refers to the concept of “race-based fa-
Emotion (JACFEE) and Neutral Faces miliarity” and not “personal familiarity”
(JACNeuF)”-battery provided by Mat- (Liew et al., 2011).
sumoto and Ekman (Matsumuto and
6
Culture modulates empathy with anger
Figure 1. Paradigm
A cue indexed the task at the start of each trial. The cue had the shape of a
black circle with a small white circle in the ‘North’- or ‘South’-position. The
white circle in the ‘North’-position indexed an intentional empathy trial; the
white circle in the ‘South’-position cued a skin color evaluation trial. In
intentional empathy trials, subjects were instructed to empathize with
perceived emotional or neutral faces. After a 4 s lasting viewing period,
subjects were supposed to rate their subjective impression of empathy
capability in the evaluation period, which lasted for 3.5 s. By virtually
moving a red bar, they were instructed to make a statement on a visual
analogue scale. In control trials subjects were instructed to concentrate on
the skin color of the presented faces. Analogue to the intentional empathy
task, a 4s lasting viewing period was followed by a 3.5s lasting evaluation
period. After every trial a short inter trial interval of 1.2s to 1.8s duration
was presented. The face stimuli consisted of familiar angry, familiar neutral,
and unfamiliar neutral faces. Both groups of subjects had different sets of
stimuli: For Chinese subjects familiar faces consisted of Chinese faces and
7
Culture modulates empathy with anger
unfamiliar faces consisted of Caucasian faces. For German subjects familiar
faces consisted of Caucasian faces and unfamiliar faces consisted of Chinese
faces. Because of time constraints, we did not include unfamiliar angry
stimuli. After every 6 trials, a baseline trial was presented; during baseline
trials, the fixation cross was displayed for 6s or 7s.
Psychological scales (24 slices parallel to the AC-PC plane,
slice thickness 5mm, TR 2000ms, TE
Interpersonal Reactivity Index (IRI) 30ms, flip angle α=90◦ , 64 x 64 voxels
The “Interpersonal Reactivity Index” per slice with 3.75mm x 3.75mm x 5mm)
(IRI, Davis, 1983) is a commonly used were acquired in seven scanning sessions
self evaluation questionnaire to measu- containing 156 volumes per session for
re the subjective impression of different each subject. In addition, T1-weighted
empathic skills. The IRI uses four sub- images of each subject were recorded.
scales related to ‘empathic fantasy’, ‘em-
pathic concern’, ‘personal distress’, and fMRI data analysis
‘perspective taking’. The statistical analysis of the
fMRI data was performed using
Self-Construal Scale (SCS) the software packages “Analysis of
The “Self-Construal Scale” (SCS, Singe- Functional NeuroImages” (AFNI,
lis,1994)basesontheconceptofinterde- http://afni.nimh.nih.gov/afni/, Cox,
pendentandindependentself-construals, 1996),“Python” (http://www.python.org),
which was introduced by Markus & Ki- “PERL” (http://www.perl.org) and “R”
tayama (Markus and Kitayama, 1991, (http://www.r-project.org/, R Develop-
2010). The SCS implements two scales ment Core Team, 2009).
related to ‘interdependence’ and ‘inde- The first three volumes were discarded
pendence’. to compensate for saturation effects. All
functional images were slice-time cor-
Behavioral data analysis rected with reference to the acquisition
Behavioral data where analy- time of the first slice and corrected for
zed using the software packages motion artifacts by realignment to the
“PERL” (http://www.perl.org) and “R” first volume. The images were spatially
(http://www.r-project.org/, R Develop- normalized to a standard EPI-template
ment Core Team, 2009). Repeated mea- provided by AFNI (‘TT_EPI’) and re-
surements Analyses of Variance (ANO- sampled to 3mm x 3mm x 3mm. Final-
VAs) and post-hoc independent samples ly, all functional images were smoothed
t-tests were used to test for significant with an isotropic 6mm full-width half
differences; Spearman correlations we- maximum Gaussian kernel. T1-weighted
re implemented to analyze associations images of each subject were normalized
between different behavioral scores. to a standard T1-template provided by
AFNI (‘TT_avg152T1’).
fMRI data acquisition For each subject, regressors of interest
The study was conducted using a Ge- were created by the convolution of a
neral Electrics 3 Tesla Magnetic Reso- canonical, fixed shape hemodynamic re-
nance Imaging Scanner. Functional data sponse function with the according sti-
8
Culture modulates empathy with anger
mulus time functions (Josephs et al., zed T1-weighted image.
1997). Regarding this, all relevant peri- In a second step, we performed a stati-
ods (namely viewing periods with cor- stical analysis of the raw fMRI signals.
rect later responses for both tasks and Using the significant clusters from the
all three conditions, evaluation periods different contrasts as regions of interest,
with correct responses for both tasks we extracted fMRI signal timecourses
and all three conditions, viewing and from activations found in the second le-
evaluation periods for tasks with incor- vel analysis using sphere shaped regions
rect responses, and the baseline event) of interest with a radius of 5mm. The
were included in the model. In additi- timecourses were linearly interpolated
on, six movement parameters resulting and normalized with respect to a time
from motion correction, as well as nine window ranging from 0s to 30s after the
regressors for the 3rd degree polynomial onset of each event. fMRI signal chan-
model of the baseline of each block were ges of every event were calculated with
included as regressors to account for any regard to the fMRI signal value of the
residual effects of head motion and ba- onset of the according event. Mean nor-
seline fluctuations respectively. Contrast malized fMRI signal values from two
images were calculated by employing li- following time steps (6s to 8s after on-
near contrasts to the parameter estima- set of the according event) were inclu-
tes for the regressors of each event. The ded in the statistical analysis. We used
resulting contrast images were then sub- paired t-tests, to analyze the effects of
mitted to a second level random-effects the factors ‘task’ ([‘intentional empathy
analysis. Here, one-sample t-tests (inclu- forfamiliarangry’+‘intentionalempathy
ding the 16 Chinese and the 16 German for familiar neutral’+‘intentional empa-
subjects in one group) and independent thy unfamiliar neutral’]-[control for fa-
two sample t-tests (comparing the 16 miliar angry’+‘control for familiar neu-
Chinese and the 16 German subjects) tral’+‘control for unfamiliar neutral’]),
were applied (Friston et al., 1994). To ‘emotion’ (‘intentional empathy for fa-
control for the multiple testing problem, miliar angry’-‘intentional empathy for
we performed a false discovery rate cor- familiar neutral’), and ‘familiarity’ (‘in-
rection (Nichols and Hayasaka, 2003) tentional empathy for familiar neutral’-
and calculated family-wise error proba- ‘intentional empathy unfamiliar neu-
bilities. The anatomical localization and tral’). In addition, Spearman correlati-
labeling of significant activations were onswereappliedtoanalyzetheassociati-
assessed with reference to the standard on of different behavioral scores (namely
stereo-tactic atlas of Talairach & Tour- the intra-scanner empathy ratings for
noux (Talairach and Tournoux, 1988) angry faces, the IRI ‘personal distress’
andbysuperimpositionofthegroupcon- scoreandtheSCS‘interdependence’sco-
trast images on a mean brain generated re) with hemodynamic responses of our
by an average of each subject’s normali- regions of interest.
9
Culture modulates empathy with anger
Results
Behavioral results lower ratings of ‘empathic concern’
(t(30)=1.935, p =0.062(*)) com-
[two-tailed]
Intra-scanner empathy ratings pared to Germans. We did not find signi-
We used a 2x3 factorial repeated measu- ficant differences with regard to the IRI
re analysis of variance (ANOVA) with scales ‘empathic fantasy’ (t(30)=0.689,
Group (Chinese vs. Germans) as a p =0.496) and ‘perspective ta-
[two-tailed]
between-subjects factor and Conditi- king’ (t(30)=0.261, p =0.796)
[two-tailed]
on (‘familiar angry’, ‘familiar neutral’, though (Figure 2.b).
and ‘unfamiliar neutral’) as a within- Subjects, who score high on ‘personal
subjects factor to analyze mean intra- distress’ typically agree to the state-
scanner empathy ratings. We detec- mentsofthefollowingkind:“Isometimes
ted a significant main effect of Condi- feel helpless when I am in the middle of
tion (F(2,60)=50.793, p<0.001**) and a very emotional situation”, “Being in
a significant interaction of Group x a tense emotional situation scares me”,
Condition (F(2,60)=4.017, p=0.023*); and “I tend to lose control during emer-
the main effect for Group was not gencies”.
significant (F(1,30)=1.209, p=0.280). Subjects, who score high on ‘empathic
Post-hoc t-tests revealed a significant concern’, in contrast, typically agree to
group difference only for the conditi- these statements: “I often have tender,
on ‘unfamiliar neutral’. German sub- concerned feelings for people less for-
jects rated higher subjective impres- tunate than me”, “When I see someone
sion of empathy for unfamiliar neu- being taken advantage of, I feel kind of
tral faces compared to Chinese sub- protective towards him”, and “I would
jects (t(30)=2.782, p =0.009**), describe myself as a pretty soft-hearted
[two-tailed]
whilst there were no differences between person” (Davis, 1983).
both groups for the conditions ‘familiar
angry’ (t(30)=0.339, p =0.737) Self-Construal Scale (SCS)
[two-tailed]
and ‘familiar neutral’ (t(30)=0.317, Independent samples t-tests confirmed
p =0.764; Figure 2.a). higher ratings of Chinese compared
[two-tailed]
to Germans in the ‘interdependence’-
Interpersonal Reactivity Index (IRI) scale (t(30)=3.469, p =0.002**),
[two-tailed]
The 2x4 factorial ANOVA with Group whilst both groups did not differ
(Chinese vs. Germans) as a between- with regard to the ‘independence’-scale
subjects factor and Sub-scale (‘empa- (t(30)=0.710, p =0.483; Figure
[two-tailed]
thic fantasy’, ‘empathic concern’, ‘per- 2.c). Subjects, who score high on the
sonal distress’, and ‘perspective ta- ‘interdependence’-scale, typically agree
king’) as a within-subjects factor re- to the statements which include the fol-
vealed a significant effect of Sub-scale lowing: “It is important for me to main-
(F(3,90)=46.634, p<0.001**) and a si- tain harmony within my group”, “I will
gnificant interaction of Group x Sub- sacrifice my self-interest for the benefit
scale (F(3,90)=3.418, p=0.021*). The of the group I am in”, and “Even when I
main effect of Group was not signifi- strongly disagree with group members, I
cant (F(1,30)=0.039, p=0.844). Post- avoid an argument” (Singelis, 1994).
hoc t-tests showed significant higher ra-
tings of Chinese for ‘personal distress’ Correlation analyses of behavioral sco-
(t(30)=2.496, p =0.018*) but res
[two-tailed]
10