Personality modulates the effects of emotional arousal and valence on brain activation

13 Pages·2012·0.65 MB·English

doi:10.1093/scan/nsr059 SCAN(2012)7,858^870
Personality modulates the effects of emotional
arousal and valence on brain activation
Elizabeth G. Kehoe,1 John M. Toomey,2 Joshua H. Balsters,3 and Arun L. W. Bokde1
1Cognitive Systems Group, Discipline ofPsychiatry, Schoolof Medicine and theTrinityCollegeInstitute ofNeuroscience (TCIN),
TrinityCollegeDublin, 2Schoolof Computing, DublinInstitute ofTechnology and 3TheSchool ofPsychology andThe TrinityCollege
Instituteof Neuroscience (TCIN),Trinity CollegeDublin,Dublin, Ireland.
Theinfluenceofpersonalityontheneuralcorrelatesofemotionalprocessingisstillnotwellcharacterized.Weinvestigatedthe D
o
relationship between extraversion and neuroticism and emotional perception using functional magnetic resonance imaging w
n
(fMRI)inagroupof23young,healthywomen.Usingaparametricmodulationapproach,weexaminedhowthebloodoxygenation lo
a
leveldependent(BOLD)signalvariedwiththeparticipants(cid:2)ratingsofarousalandvalence,andwhetherlevelsofextraversionand d
e
d
nlinekusrohtiicgihsmexwtrearveerrseiloantetdotloowtheersleevmelosdouflarteiotincsu.loInthpaalartmiciuc–lacro,rwtiecawliasrhoeudsatol,taensdtEnyesuernotcikc’issmbiotologiniccarelathseedorryeoafctpievirtsyoonfaltihtye,lwimhibcihc from
systemandstrongerreactionstoemotionalarousal.Individualshighinneuroticismdemonstratedreducedsustainedactivationin h
theorbitofrontalcortex(OFC)andattenuatedvalenceprocessingintherighttemporallobewhileviewingemotionalimages,but ttps
anincreasedBOLDresponsetoemotionalarousalintherightmedialprefrontalcortex(mPFC).TheseresultssupportEysenck’s ://a
c
theory,aswellasourhypothesisthathighlevelsofneuroticismareassociatedwithattenuatedrewardprocessing.Extraversion a
d
wasinverselyrelatedtoarousalprocessingintherightcerebellum,butpositivelyassociatedwitharousalprocessingintheright em
insula, indicating thattherelationship betweenextraversion andarousal isnotassimple asthatproposed by Eysenck. ic
.o
u
p
.c
Keywords: arousal; extraversion; fMRI;neuroticism; valence om
/s
c
a
n
/a
INTRODUCTION Cremers et al., 2010). However, there has been little focus rtic
Extraversion and neuroticism are two of the most widely todateontherelationshipbetweenneuroticismandpositive le-a
studied personality traits (Canli et al., 2001; Hamann and emotional perception. There is evidence that depressed pa- bs
Canli,2004)andtheyarecommontoseveraldominantthe- tients demonstrate reduced neural responding to positive trac
ories of personality including Costa and McCrae’s Big Five emotional stimuli (Shestyuk et al., 2005), and given that t/7/7
model (2003) and Eysenck’s biological approach (1967, neuroticismisariskfactorfordepression,wewereinterested /8
5
1994). Extraversion and neuroticism have been found to ininvestigatingwhetherhighlevelsofthistraitareassociated 8/1
correlate robustly with positive and negative emotionality, with an attenuated neural response to positive valence. 67
6
respectively (Costa and McCrae, 1980; Rusting and Larsen, According to Eysenck’s biological theory of personality 30
0
1997),arelationshipwhichseemstoremainstableacrossthe (EysenckandEysenck,1991;Eysenck,1994),theneuroticism b
y
lifespan(WilsonandGullone,1999);anddifferencesinthese dimension is also posited to affect how individuals respond gu
traits are known to influence emotional and cognitive pro- to emotional arousal. Whereas extraversion is considered es
cessing (Bradley and Mogg, 1994; Amin et al., 2004; Canli, to be linked to differences in the functioning of the t on
2004; Kumari et al., 2004, 2007). reticulothalamic–corticalarousalsystem,highlevelsofneur- 04
M
Although neuroticism describes differences in a personal- oticism are theorized to reflect increased reactivity of the a
y
ity dimension rather than a clinical disorder, it can be of limbic system, which predisposes individuals high in neur- 2
0
high clinical relevance as it is a risk factor for developing oticismtoreactstronglytoemotionallyarousingexperiences 19
anxiety and depression disorders (Derryberry and Reed, and take longer to return to pre-arousal states. In spite of
1994; Kendler et al., 2004). Several neuroimaging studies these predictions about the relationship between personality
have found that this trait is most notably associated with andreactivitytoemotionalarousal,theneuralbasesofthese
biases towards negative emotional processing (Canli et al., relationships have never been investigated using functional
2001; Haas et al., 2007, 2008; Chan et al., 2008, 2009; magnetic resonance imaging (fMRI). Therefore, we were
interested in examining whether differences in this trait are
related to how emotional arousal is processed in the brain.
Received4April2011;Accepted25August2011
AdvanceAccesspublication23September2011 Eysenck’s biological theory of extraversion (1967, 1994)
WewouldliketothankMrSojoJosephsforhisassistanceinacquiringtheneuroimagingdataandthe proposes that differences in this personality dimension are
TrinityCentreforHighPerformanceComputingformakingcomputerresourcesavailableforthestudy.
reflectiveofdifferencesinareticulothalamic–corticalarousal
CorrespondenceshouldbeaddressedtoElizabethG.Kehoe,LloydInstitute,TrinityCollegeInstituteof
Neuroscience,TrinityCollegeDublin,Dublin2,Ireland.E-mail:[email protected] system, with extraverts experiencing both lower baseline
(cid:2)TheAuthor(2011).PublishedbyOxfordUniversityPress.ForPermissions,pleaseemail:[email protected]
Personalityinfluencesonemotionalperception SCAN(2012) 859
levels of cortical arousal, as well as low arousability of the motivational systems in the brain which drive behavioural
cortex, i.e. they show a smaller change in cortical activity in responses, an approach and an inhibition system, which
response to arousing stimuli than introverts (Eysenck, 1967; differ between individuals, representing differences in sensi-
EysenckandEysenck;1991;Hagemannetal.,2009).Itisthis tivitytoenvironmentalcuesofrewardandpunishment.This
chronic intrinsic under-arousal which is thought to drive theoryalsohasastrongbiologicalcomponent,however,due
highly extraverted people to engage in typically extraverted to the focus on behavioural inhibition in the avoidance of
behaviours in order to enhance their low arousal states punishment cues, we choose Eysenck’s model as having po-
(Eysenck, 1994). Furthermore, this under-arousability en- tentiallymorepredictivepowerinthecontextofourexperi-
ables extraverts to tolerate much higher levels of arousal mental questions, namely concerning the processing of
than introverts, who withdraw to avoid further increases in rewarding stimuli, and arousal in particular. Finally, the
arousal which they find difficult to withstand (Eysenck, Big Five model of Costa and McCrae (McCrae and Costa,
D
1967, 1994). 2003), although very popular in the field of personality has o
w
Several neuroimaging studies have supported this theory been criticized for being more of a descriptive model, with nlo
of extraversion (Kumari et al., 2004; O’Gorman et al., 2006; less scope for explanation as it contains no specific predic- ad
e
lHeaagguemesan(2n00e9t)alf.o,u2n0d09)th;aftoralepxhaampaclet,ivHityagaems amnenasaunrdedcobly- tioInnspaabrotuictutlhare,bwieolwogericeailnbtearseisstoedf pinertseosntianligtyth(eBrloelcakt,io2n0s1h0i)p. d from
resting-stateelectroencephalography(EEG),whichisindica- betweenextraversionandtheperceptionofemotionalarou- http
tiveoflowerarousalstates(Barryetal.,2011;DeCesareiand sal, and predicted that individuals high in this trait would s
Codispoti, 2011) was positively associated with extraversion experience less brain activation in response to increasing ://ac
a
levels in left and right frontal sites, indicating that extra- levels of arousal, in accord with the predictions of de
m
verted individuals had lower baseline levels of cortical arou- Eysenck’s arousal hypothesis (Eysenck, 1994). Second, we ic
sal; while Kumari et al. (2004) found that levels of wished to examine the relationship between neuroticism .ou
p
extraversion were negatively associated with resting fMRI and positive valence and arousal processing. We predicted .c
o
signals in the thalamus and in Broca’s area extending to that high levels of neuroticism would be associated with m
/s
Wernicke’s area. Given the predictions of Eysenck’s arousal reduced neural activation in response to positive valence, ca
n
hypothesis,onemightexpectthatinthecontextofemotion- but an increased response to emotional arousal. /a
al processing, extraversion levels would have a pervasive in- rtic
le
fluenceontheperceptionofemotionalarousal;however,the MATERIALS AND METHODS -ab
relationship between extraversion and the neural substrates Participants stra
ofemotionalarousalprocessingareunknown.Therefore,we Twenty-three young (mean age¼23.04(cid:2)3.46 years, age ct/7
sought to investigate this question using fMRI. /7
range¼19–29 years), healthy, right-handed women took /8
Inthepresentstudy,weusedfMRItoexaminehowlevels 5
of extraversion and neuroticism are related to the neural part in the study. Only women were included as there have 8/1
been considerable differences found between genders in 67
substrates of emotional arousal and valence processing in a 6
emotional reactivity (Bradley et al., 2001b) in the neural 3
groupof23younghealthywomen.Weutilizedanamplitude 00
representation of emotion (Beck et al., 1996; Cahill et al., b
modulationapproachtodissociatetheeffectsofarousaland y
2001; Wager et al., 2003; Wrase et al., 2003) and in levels of g
valence on the blood oxygenation level-dependent (BOLD) u
neuroticism (Lynn and Martin, 1997). They had no history es
signal during an emotional image viewing task. We exam- ofpsychiatricorneurologicalillnessandwerenotdepressed t o
inedsingletrialdynamics,wherebythesubject’svalenceand [BeckDepressionInventoryscores:4.83(cid:2)3.38;cut-offscore n 04
arousal ratings for each stimulus image were included as a for mild depression¼14 (Beck et al., 1996)]. The study had M
parametric weight in our model. We then examined how ay
full ethical approval from the St James Hospital and the 2
these effects varied as a function of neuroticism and extra- 0
Adelaide and Meath Hospital, incorporating the National 1
9
version levels.
Children’s Hospital Research Ethics Committee. All of the
WeusedtheEysenckpersonalityquestionnairetomeasure
participants gave written informed consent and were paid
levelsofneuroticismandextraversion(EysenckandEysenck,
E40 for taking part.
1991). The primary reason we used Eysenck’s scale and the-
oretical framework is that this personality theory has a
strong biological component and causal explanation, with Personality measurement
individual differences in personality thought to arise from The participants completed the Eysenck personality ques-
differences in brain function. Specifically, Eysenck proposed tionnaire, revised edition short scale (EPQ-R, Eysenck and
that differences in the extraversion dimension are reflective Eysenck, 1991), a self-report questionnaire which measures
of differences in the functioning of the reticular arousal levels of extraversion, neuroticism and psychoticism on
system, while neuroticism reflects differences in the limbic scales ranging from 0 to 12, with 0 indicating the lowest
system’s emotional responses (Eysenck, 1967, 1994). Gray’s level of the trait and 12 the highest level of the trait. We
(1982, 1990, 1997) theory proposes that there are two basic focused our analyses on neuroticism and extraversion.
860 SCAN(2012) E.G.Kehoeetal.
Stimuli 95 trials and lasting (cid:3)20min. In each trial, an image was
The stimuli consisted of 190 coloured photographs. They presented in the centre of a white background for 3000ms,
were a combination of 98 images from the International and after a delay of 1000–3000ms (pseudo-random jitter),
Affective Picture System (IAPS, Lang et al., 2007) and 92 apromptappearedonscreenfor2000msaskingthepartici-
images gathered by the experimenter from various sources. pants to classify the image they had just seen as ‘Living’
They were either positive or neutral in valence and they or ‘Non-living’. The participants were instructed to make
varied in arousal levels. The stimuli were limited to neutral their response by pressing either the left or right button on
andpositive valence only for several reasons. First, although a MR-compatible button response box held in their right
arousal and valence tend to be highly correlated (Bradley hand, to correspond with the left/right position of the
et al., 1992; Ribeiro et al., 2005), we included images ‘Living/Non-living’wordonthescreen.Thisshallowencod-
which were positive or neutral in valence and which varied ingtaskwasintendedtomaintaintheparticipants’focusfor
D
in arousal level, allowing us to match the stimuli along one the duration of the task, without explicitly drawing their o
w
dimension while varying along the other and so investigate attention to the emotional content of the stimuli nlo
their effect on the BOLD signal separately. However, the (Kensinger et al., 2007). The onset of both the images and ad
e
cino-tvhaeriactaisoenooffnaergoautisvael iwmitahgevs,alwenhciechistepnadrtitcoulbaerlyrastterdonags thhaeemproodmynpatsmwicerreesjpitotenrseed(tJooseenpshusreanodptHimenalsosanm, 1p9li9n9g).ofthe d from
Tmhoerreefoarroe,uistinwgasthnaont ppoosssitibivlee tiomhagaevse e(nLoanugghetneagla.,tiv2e00lo8w). Post-scanning image rating https
arousalexemplars.Second,wewereparticularlyinterestedin Theparticipantsdidnotratetheimageswhileinthescanner ://ac
a
examiningtherelationshipbetweenneuroticismandpositive as emotional evaluation has been found to result in attenu- d
e
m
valence. Third, we wanted to be sure that any linear associ- ationoftheneuralresponsetoemotionalstimuli,thoughtto ic
ationswith valencein theimagingparadigm could beinter- beduetothetop–downinfluencesofcognitivere-evaluation .o
u
p
preted clearly as being associated with increasing valence and judgement (Hariri et al., 2000; Taylor et al., 2003). .c
o
from neutral to positive, and not confounded by the inclu- Instead, they returned 2–3 days later and rated all of the m
/s
sion of negative stimuli. images they had seen during the scanning session, as well c
a
n
The IAPS was supplemented to increase the number of as48negativeimagesfromtheIAPSalongthedimensionsof /a
positive, low arousal and neutral, higher arousal images. valence, arousal and dominance. (The 2–3 days gap was to rtic
le
None of the images we used were of a sexual nature. The facilitate the collection of memory recall information that is -a
b
new images were chosen to be as similar as possible to the not part of this report). The negative images were included s
tra
IAPS images and so, included very similar content and a intheratingtasktoprovidecontrasttotheothers,toensure c
mixture of objects, scenes, animals and people as does the thattheparticipantsunderstoodthefullremitofthevalence t/7/7
IAPS. dimension. A computerized version of the Self-Assessment /8
5
8
As far as was possible the images of different valence and Manikin(SAM,Langetal.,2008)wasusedtooperationalize /1
6
arousal levels were chosen to contain the same number of valence, arousal and dominance. 7
6
3
people, animals, scenes and objects. Before the experiment, 0
0
all of the images were rated by 15 age-matched controls, MRI scanning protocol b
y
of which 8 were women, along the dimensions of valence, Imaging data were acquired using a Philips Intera Achieva gu
e
arousalanddominance.Theseratingswerecollectedtoboth 3.0TMRsystem(Best,TheNetherlands). TheBOLDsignal st o
devise approximate ratings for the new images and also to changes were measured using a T2*-weighted echo-planar n
0
assess the level of agreement between the IAPS standard imaging sequence with TR¼2000ms and TE¼30ms. 4 M
ratings,whicharedevisedfromaUSsample,andtheratings Each volume of data covered the entire brain with 39 a
y
of a group of young Irish adults. Further details of the rat- slices, and the slices were acquired in interleaved sequence 20
1
ings of the new images and the IAPS images, as well as ex- from inferior to superior direction. In total 598 volumes 9
amples of images from our stimulus set can be found in the were acquired during each of the two runs, with voxel di-
Supplementary Data. The stimuli were delivered using mensions of 3.5(cid:4)3.5(cid:4)3.85mm and a 0.35mm gap be-
Presentation v. 13.0 (Neurobehavioral Systems, Albany, tween the slices. A T1W/IR sequence was used to collect a
CA, USA). 3Dhigh-resolutionanatomicalimagewithvoxeldimensions
equal to 0.9(cid:4)0.9(cid:4)0.9mm for structural localization.
Experimental design
Functional MRI paradigm Analysis
Theparticipantsviewed190colouredimagesas theyunder- Behavioural data
went fMRI. To avoid long-lasting mood states, the images The experimental log files were parsed using python scripts
werepseudo-randomizedsothatnomorethanthreeimages (version 2.6.2, http://www.python.org/) to extract perform-
ofthesamevalenceorarousaltypewerepresentedinarow. ance and rating information. These data were then used to
Thetaskconsistedoftwoexperimentalruns,eachcontaining create individually tailored regressors for each participant
Personalityinfluencesonemotionalperception SCAN(2012) 861
D
o
w
n
lo
a
Fig.1 IllustrationshowingaschemaofthefMRIanalysisandparametricmodellingofthehaemodynamicresponsefunction.Eachtrialconsistedofanimagepresentedfor d
e
3000ms,apseudo-randomjitterof1000–3000msbeforetheonsetofthepromptfor2000ms,followedbyaninter-trialintervalofbetween5000msand8000ms.Thevalence d
and arousal ratings were included in the GLM as amplitude modulators and are shown in blue and red, respectively, whereas the constant or average BOLD response is fro
m
representedby thesolid grey response. TheBOLD response totheprompts is shownasa dashedgrey line. h
ttp
s
://a
c
a
based on their subjective ratings of the images. Statistical low frequency noise and (iii) two regressors to model the d
e
m
analyses of the behavioural results were conducted using mean differences between the two runs. ic
Minitab (version 15, Coventry, UK). All statistical analyses were calculated in the participants’ .o
u
p
native space, then the regressor coefficients maps were nor- .c
o
malizedintostandardstereotactic spacebywarpingthemto m
MRI data analysis the MNI brain template (Montreal Neurological Institute/ /sc
a
n
The MRI data were analysed using Analysis of Functional International Consortium for Brain Mapping 152 standard /a
NeuroImaging (AFNI) (Cox, 1996) (http://afni.nimh.nih.gov/ atlas as provided in the FSL software package) using FSL’s rtic
le
afni/)andFSL(FMRIBSoftwareLibrary-http://www.fmrib.ox. linear registration tool, FLIRT. The transformation matrix -a
b
ac.uk/fsl/).Thefirstfourdynamicswereobtainedtocorrectfor (12 parameter affine) from native space to MNI space was stra
T1 equilibration effects and were subsequently discarded. The calculated using the high resolution structural images from c
t/7
dataweremotion-correctedbyrealignmenttothefirstvolume each subject. /7
of the first run, concatenated into a single run, global mean The group statistical analyses were based on a /85
8
adjusted by proportional scaling and smoothed with a 6-mm random-effects model, and activation maps of the constant /1
6
full-width-at-half-maximum Gaussian kernel. and modulated responses were calculated with a series of 76
3
A general linear model (GLM) analysis was conducted in independent t-tests. The relationship between personality 00
AFNI. Figure 1 shows a schema of the experimental design. and brain activation was examined with a series of whole by
g
Two regressors of interest were included to model the vari- brain correlations using AFNI’stool 3dRegAna. Thepartici- u
e
ance due to the image and prompt trials(cid:2)these two regres- pants’levelsofextraversionandneuroticismwereseparately st o
sors modelled the mean BOLD signal change from baseline regressed against the constant BOLD response, the n 0
4
across all trials. To model the additional effects of arousal arousal-modulated response and the positive valence- M
and valence on the BOLD response during the image trials, modulatedresponse.Inordertocontrolforthenearinverse ay
2
subjectiveratingsofarousalandvalenceforeachimagewere correlation between extraversion and neuroticism, when 0
1
included as single trial parametric weights. This resulted in examining the relationship between extraversion and the 9
three image regressors: a constant unmodulated regressor BOLD signal, neuroticism was included as a covariate of
describing themeanBOLDresponse duringtheimagepres- no interest in the model, while extraversion was included
entations, regardless of the arousal or valence of the image as a covariate of no interest when examining the effects of
(this will be referred to in this manuscript as the constant neuroticism on brain activation.
BOLD response), BOLD activation during the image trials Significant voxels passed a voxel-wise statistical threshold
modulated by arousal, and BOLD activation during the of P(cid:5)0.01. To correct for multiple comparisons across the
image trials modulated by valence. A separate regressor brain, each cluster had to have a minimum size of 708ml of
was included for all of the images rated as negative, so that contiguous statistically significant voxels to be considered
these trials were not included in the amplitude modulation statistically significant. This minimum cluster size was cal-
analysis. Several regressors of no interest were also included culated using a Monte Carlo simulation (in AFNI) to ob-
in the GLM to model the following sources of variance: tain a (familywise error) corrected P<0.05 statistical
(i) six motion parameters, (ii) eight regressors to model significance in the t-tests. The SPM anatomy toolbox
862 SCAN(2012) E.G.Kehoeetal.
D
o
w
n
lo
a
d
e
d
fro
m
h
ttp
s
://a
Fig.2 Thegrouplevelresultsindependentofthepersonalitytraits.AconjunctionmapwascreatedtoshowthebrainregionswithastatisticallysignificantconstantBOLDsignal c
a
notmodulatedbytheemotionalcontentofimages,andbrainregionswheretheBOLDsignalwasmodulatedbythearousalandvalencecharacteristicsofthestimuli.Thecolours d
e
represent the following: Purple¼Constant BOLD response; yellow¼regions modulated by arousal; red¼regions activated in both the constant response and the m
arousal-dependentresponse;blue¼regionsmodulatedbyvalence;green¼regionsactivatedinboththeconstantresponseandthevalence-dependentresponse.Thenumbers ic.o
oftheslicesindicate thedirectionalong theinferior–superioraxis in millimetresin standardMNI space ((cid:6)LPI). up
.c
o
m
/s
c
a
(V1.7b, Eicknoff et al., 2005) was used to localize activation of neuroticism and extraversion, each subject’s neuroticism n/a
clusters; however, where there were no probabilistic cyto- and extraversion scores were correlated against the number rtic
architectoniclabelsavailable,aBrodmannarea(BA)isgiven of images they rated as positive and negative, and high and le-a
b
in the results table instead. low in arousal. In all cases, the results were not statistically s
significant (P(cid:7)0.38 in all cases), indicating that personality trac
RESULTS scores were not associated with the ratings of the images. t/7
/7
Behavioural results /8
5
8
Image ratings fMRI results /1
6
7
The participants’ individual arousal and valence ratings of The constant and modulated BOLD responses 6
3
thestimuliwereusedtocreateregressorswhichwereunique independent of personality variables 00
b
for each subject. Average ratings were calculated to give an The constant BOLD response during image viewing, as well y g
idea of the distribution of the scores; however, they are asthemodulations duetoarousalandpositivevalencewere ue
s
included here for guidance only. On average, 53.87 first examined independent of neuroticism and extraversion t o
n
((cid:2)26.91)imageswereratedaspositive(i.e.havingavalence levels. A conjunction analysis was performed to examine 0
4
of7,8or9),113.04((cid:2)37.19)asneutral(i.e.havingavalence which regions of the brain were activated in the constant M
of 4, 5 or 6), and 18.43 ((cid:2)13.96) as negative (i.e. having a response only, which regions were modulated only by ay
2
valence of 1, 2 or 3). The participants rated 99.57 ((cid:2)13.76) increasinglevelsofarousalandpositivevalenceandwhether 01
9
images as having an arousal level of between 1 and 5, while there were regions which were active in both the constant
67.35 were rated as having an arousal level of 6 or more. responseandwhichwerealsomodulatedbythearousaland
All negative images were excluded from the parametric valence of the stimuli. The results of this conjunction ana-
modulation analysis and were included in the GLM as a lysisareshowninFigure2,andtheinterestedreadercanfind
regressor of no interest. a table summarizing the significant activation clusters in
Supplementary Table S3 in the Supplementary Data.
EPQ scores Regions which were significantly active compared with
The mean and standard deviation of the neuroticism and baseline in the constant response only, and which were not
extraversionscoreswere4.8(cid:2)2.6and8.0(cid:2)3.6,respectively. modulated by the arousal or valence level of the stimuli are
Thesescoreswerenotcorrelated,althoughtherewasatrend showninpurpleinFigure2.Theyincludeaverylargecluster
towards an inverse correlation (Pearson’s r¼(cid:6)0.397, spanning widespread areas of the occipital and temporal
P¼0.061). In order to investigate whether there was a lobes and which included the hippocampus, and clusters in
linear relationship between the emotional ratings and levels the right superior temporal gyrus, the left middle frontal
Personalityinfluencesonemotionalperception SCAN(2012) 863
gyrus, the left and right cerebellum and the right middle Table1 LinearassociationsbetweenneuroticismlevelsandtheBOLDsignal
occipital gyrus. during image viewing
Voxels t-valuea MNI Area: prob
co-ordinates or BA
Overlap between the constant and arousal-dependent
BOLD responses
ConstantBOLD response
There was a large degree of overlap between regions which Frontallobe
were activated in the constant BOLD response and also Leftrectal gyrus 161 (cid:6)4.96 (cid:6)4, 38, (cid:6)26 BA 11
Leftrectal gyrus (cid:6)3.62 (cid:6)6, 30, (cid:6)22 BA 11
modulated by increasing levels of emotional arousal, espe-
Arousalmodulation
cially in the occipital and temporal visual processing areas.
FrontalLobe
These regions are shown in red in Figure 2 and include a Right middle orbitalgyrus 211 5.65 12, 52, (cid:6)8 BA 11
large cluster that spans the fusiform gyrus and primary and Right middle orbitalgyrus 4.72 2, 48, (cid:6)6 BA 10 D
o
Right middle orbitalgyrus 3.47 2, 58, (cid:6)14 BA 11 w
secondary visual cortex (BA 17 and 18) bilaterally, and fur- n
Valence modulation lo
ther clusters in the left temporal pole, the right superior a
TemporalLobe d
e
temporal gyrus, the right and left inferior frontal gyrus, the Right middle temporal gyrus 102 (cid:6)4.43 70, (cid:6)38, (cid:6)14 BA 21 d
right thalamus and the right precentral gyrus. Right middle temporal gyrus (cid:6)4.3 70, (cid:6)34, (cid:6)16 BA 21 fro
m
Right inferiortemporal gyrus (cid:6)4.16 64, (cid:6)26, (cid:6)18 BA 21 h
Right rolandic operculum 98 (cid:6)3.92 54, (cid:6)18, 18 OP 1: 50%; ttp
s
The arousal-dependent BOLD response OP 4: 30% ://a
c
Several brain regions in the temporal andfrontal lobes were at(cid:7)2.82, P<0.01; t(cid:7)3.79, P<0.001, corrected. OP= opercular cortex. ad
not activated in the constant BOLD response but only pref- LabelsweregeneratedusingSPM’sanatomytoolbox;however,wheretherewasno em
erentiallyrespondedtoincreasinglevelsofarousal.Theseare cytoarchitectonicinformationavailable,aBrodmannareaisgiven.MNIco-ordinates ic.o
shown in yellow in Figure 2, and included clusters in the are(cid:2)LPI (i.e. negative indicates left, posterior andinferior directions). up
.c
middle temporal gyri bilaterally, the left inferior temporal o
m
gyrus, the right superior temporal gyrus, the left superior /s
c
a
medial frontal gyrus, the left and right inferior frontal The relationship between extraversion and the n
/a
gyrus, the precuneus, the left middle cingulate cortex, the BOLD signal rtic
right thalamus and the left middle occipital gyrus. le
Extraversion was negatively associated with the constant -a
b
BOLD response in a region of BA 6 spanning the middle s
The valence-dependent BOLD response cingulatecortexandsupplementalmotorarea(SMA).There trac
The valence-dependent BOLD response was much less ex- was a negative linear relationship between extraversion and t/7/7
the arousal-dependent BOLD response in the right cerebel- /8
tensive than the arousal-dependent response. There was a 5
8
single cluster located in the left inferior temporal gyrus lum and a positive linear association between extraversion /16
and the arousal-dependent BOLD response in the right 7
which was not activated in the constant BOLD response 6
3
insula. Extraversion was negatively associated with the 0
but whose BOLD signal was significantly modulated by 0
valence-dependentBOLDsignalintherightsuperiorparietal b
increasing levels of positive valence. This cluster is shown y
in blue in Figure 2. There was also a cluster in the left fusi- lobule, the right postcentral gyrus and the left posterior gue
form gyrus that was both activated in the constant BOLD hippocampus (Table 2 and Figure 4). st o
n
response and which also showed a positive modulation of 0
4
theBOLDsignalinresponsetoincreasinglypositivevalence. M
DISCUSSION a
This is shown in green in Figure 2. y
2
Theresultsofthepresentstudydemonstrateseveralrelation- 0
1
shipsbetween personality andemotionalprocessing thatare 9
The influence of personality on emotional processing novel findings, in particular the relationships between neur-
The relationship between neuroticism and the oticismandthreedifferentelementsofemotionalprocessing.
BOLD signal First, neuroticism levels were negatively associated with the
There was a strong negative linear association between constant BOLD signal in the OFC, regardless of the valence
neuroticism and the constant BOLD response in the orarousallevelofthestimuli.Althoughthisrelationshipwas
middle orbitofrontal cortex (OFC). There was a negative notvalencespecific astheimagesincludedbothneutral and
linear association between neuroticism and the valence- positive exemplars, it is of particular interest given that the
dependent BOLD modulation in the right middle temporal OFC is known to play a pivotal role in the processing of
gyrusandtherightrolandicoperculum.Therewasapositive emotion and reward (Schultz et al., 2000; O’Doherty, 2004;
linear association between neuroticism and the degree of Rolls,2004;Kringelbach,2005),andinalsointhesuccessful
arousal-dependent BOLD modulation in the right medial down-regulationofnegativeemotionalstates(Ochsneretal.,
prefrontal cortex (mPFC; Table 1 and Figure 3). 2004; Eippert et al., 2007).
864 SCAN(2012) E.G.Kehoeetal.
widespread activation in the temporal lobes associated with
emotional image viewing. Previous neuroimaging studies
haveidentifiedthistraittobestronglyassociatedwithnega-
tive, rather than positive emotional processing (Canli et al.,
2001; Haas et al., 2007, 2008; Chan et al., 2008, 2009;
Cremers et al., 2010). We found in the current study that
theconverseisalsoapparent(cid:2)neuroticismisassociatedwith
atendencyforseveralregionsinthebraintobelessrespon-
sive to positive emotional stimuli.
The attenuated neural response to the stimuli in the OFC
and reduced activation to positive emotional valence in the
D
rightmiddletemporalgyrusmighttentativelybeinterpreted o
w
as a predisposition in cases of high neuroticism for under nlo
a
activationinreward-processingoremotionalregulation.We d
e
believe that this is the first demonstration of such a finding d fro
inanon-clinicalpopulation,anditmayrepresentabiologic- m
al vulnerability marker or putative phenotypic characteristic http
ataneuralnetworklevel(Leschetal.,1996;Senetal.,2003; s
Strobel et al., 2003), which may help to further explain the ://ac
a
relationship between this trait and the increased risk of de- de
m
velopingdepressionandanxietydisorders.However,further ic
elucidation of these types of functional phenotypes will be .ou
p
needed to understand the complex relationship between .c
o
personality, emotional processing and the risk for affective m
/s
disorders. ca
n
Last, we found that neuroticism was associated with an /a
increased BOLD response to emotional arousal in the left rtic
le
mPFC. This is the first time, to our knowledge, that a link -a
b
has been identified between neuroticism and increased re- stra
activity to emotional arousal, rather than valence, in the ct/7
brain. Eysenck (1967, 1994) proposed in his biological /7
Fig.3 NeuroticismlevelswerenegativelyassociatedwiththeconstantBOLDsignal /8
theory of personality that high neuroticism is associated 5
inthe(a)orbitofrontalcortex(OFC),positivelyassociatedwiththearousal-dependent 8
BOLDsignalinthe(b)medialprefrontalcortex(mPFC)andnegativelyassociatedwith with increased reactivity of the limbic system, which predis- /16
7
thevalence-dependent BOLD signalin the(c)rightmiddle temporal gyrus. poses individuals high in neuroticism to react strongly to 6
3
emotionally arousing experiences and take longer to return 00
b
to pre-arousal states. This suggests that neuroticism influ- y
g
ences emotional processing by enhancing neural sensitivity u
e
s
Second,wefoundthatinaccordancewithourprediction, tohighlevelsofemotionalarousal,makinghighneuroticism t o
neuroticism was negatively associated with valence process- individualspredisposedorprimedtoreactstronglytoarous- n 0
4
ing, in two regions of the right temporal lobe. It has previ- ing experiences (Eysenck, 1967). Indeed this trait is asso- M
ously been reported that patients with major depression ciated with greater sensitivity to negative mood inductions ay
2
show reduced brain activation in response to positive emo- (Larsen and Ketelaar, 1991); and it has been found to cor- 0
1
tional words on an emotional working memory task relate with larger electrodermal responses (EDR) to both 9
(Shestyuk et al., 2005). Given that higher levels of neuroti- arousing negative and positive emotional images, as well as
cism are associated with an increased risk of developing af- with a longer period of recovery in the EDR signal (Norris
fective disorders such as depression and anxiety (Kendler et al., 2007). The results of the present study support
et al., 2004), we hypothesized that a similar pattern may be Eysenck’stheory,withindividualshighinneuroticismshow-
evident in the case of high neuroticism individuals. Indeed, ing an enhanced response to emotional arousal in the left
we found a negative relationship between neuroticism and mPFC.Thiscorticalregionhaspreviouslybeenimplicatedto
the neural response to positive valence in the right middle playacentralroleinself-referentialprocessingandemotion-
temporal gyrus and the right rolandic operculum. The right al attribution and appraisal (Eysenck, 1967; Fossati et al.,
middle temporal gyrus has been found to be activated in 2003; Mitchell et al., 2005, 2006; Ochsner et al., 2005).
both emotional processing and encoding tasks (Critchley Furthermore, constant activation in the mPFC in response
et al., 2000; Dolcos et al., 2004; Olson et al., 2007), and to sad faces has been found to correlate with neuroticism
the group results from the current study also indicate (Haas et al., 2008). Thus, the enhancement of activation in
Personalityinfluencesonemotionalperception SCAN(2012) 865
Table 2 Linear associations between extraversion levels and the BOLD signal during image viewing.
Voxels t-valuea MNICo-ordinates Area: prob orBA
Constant BOLDresponse
Frontal Lobe
Leftmiddle cingulate cortex 189 (cid:6)4.01 (cid:6)6, (cid:6)0, 40 Area 6: 10%
Right SMA (cid:6)3.79 10, 0, 46 Area 6: 40%
Arousal modulation
Cerebellum
Right cerebellum 96 (cid:6)4.17 24, (cid:6)68, (cid:6)26 Lobule VI:75%
Right cerebellum (cid:6)3.85 26, (cid:6)72, (cid:6)38 Lobule VIIa CrusI:57%
Right cerebellum 91 (cid:6)3.85 30, (cid:6)84, (cid:6)42 Lobule VIIa CrusII: 98%
Right cerebellum (cid:6)3.84 22, (cid:6)86, (cid:6)38 Lobule VIIa CrusII: 93% D
o
Frontal Lobe w
n
Right insulaLobe 130 5.46 34, 14, 2 BA 13 lo
a
Right insulaLobe 4.23 32, 10, 14 BA 13 d
e
Valence modulation d
Parietal Lobe fro
m
Right superiorparietal lobule 102 (cid:6)4.72 24, (cid:6)58, 72 SPL (7A):50% h
Right postcentral gyrus (cid:6)3.99 30, (cid:6)48, 74 BA 2 ttp
RRiigghhtt psuopstecreionrtrpaalrgieytraulslobule 132 (cid:6)(cid:6)33..268 3104,, (cid:6)(cid:6)3526,, 4628 ASPreLa(53La)::4600%% s://a
c
Temporal Lobe ad
e
Lefthippocampus 109 (cid:6)4.47 (cid:6)18, (cid:6)36, 8 Hipp (CA): 50% m
Lefthippocampus (cid:6)4.19 (cid:6)24, (cid:6)34, 2 Hipp (CA): 20% ic.o
u
p
at(cid:7)2.82, P<0.01; t(cid:7)3.79, P<0.001, corrected. SPL¼superiorparietal lobule. Referto Table1 legend. .c
o
m
/s
c
a
n
/a
this key emotional appraisal area in high neuroticism indi- rtic
vidualssuggestsaheightenedresponsetohighlevelsofemo- le
-a
tional arousal which may go some way to explaining why b
s
arousing experiences might have a more intense and lasting tra
c
effect. t/7
/7
The relationship we identified between extraversion and /8
5
the neural response to emotional arousal in the right cere- 8
/1
bellum is consistent with Eysenck’s arousal hypothesis of 67
6
extraversion (1967, 1994), which proposes that extraverts 3
0
0
demonstrate both lower baseline levels of cortical arousal b
y
and under-arousability of the cortex, and can tolerate g
u
higher levels of arousal better than introverts (Eysenck, es
1967; Kumari et al., 2004; O’Gorman et al., 2006; t on
Hagemann et al., 2009). This is postulated to explain why 04
M
extravertstypicallyengageinmorerisk-takingandimpulsive a
y
behaviours (Costa and McCrae, 1980; Eysenck, 1994), as 2
0
they endeavour to enhance their intrinsic low arousal 19
levels. Our results provide some supporting evidence for
the under-arousability component of this hypothesis, with
individuals high in extraversion showing less activation in
response to increasing levels of arousal in areas Crus I and
Crus II in the right cerebellum.
Althoughonceconsideredapurelymotorregion,itisnow
known that outputs from the cerebellum inFuence more
widespread regions of the cerebral cortex than previously
Fig. 4 Extraversion levels were positively associated with the arousal-dependent recognized, and there is converging evidence from primate
BOLD signal in the (a) right insula lobe and negatively associated with the connectivitystudiesandhumanresting-statefMRIdatathat
valence-dependent BOLD signalin the(b) lefthippocampus.
there are several anatomical circuits between the cerebellum
and PFC (Kelly and Strick, 2003; Habas et al., 2009;
866 SCAN(2012) E.G.Kehoeetal.
O’Reilly et al., 2010). For example, there are reciprocal con- or positive stimuli (For a review, see Carretie et al., 2009).
nections between Crus II and BA 46 (Strick et al., 2009), Thecontrastbetweenpositiveversusnegativeimageviewing
supportingtheideathatthecerebellummayplayanimport- may also reveal greater differences in brain activation than
ant role in attention and working memory among other di- the contrast between positive and neutral image viewing.
verse cognitive and emotional functions (Dolan, 1998; Further,ourpositivestimulihadonlymoderatelyhigharou-
Rapoport et al., 2000; Stoodley and Schamahmann, 2009; sal levels (few were rated as 8 or 9 on a 9-point scale) as we
Leiner, 2010). Therefore, it is not entirely surprising to did not include any erotic stimuli. This may indicate that
find a relationship between personality and arousal process- highextraversionisassociatedwithgreaterneuralsensitivity
ingintheseregions,andinareviewofneuroimagingstudies to highly arousing, highly positive stimuli rather than all
which have reported cerebellar activations, Stoodley and positive stimuli. The amplitude modulation analysis and
Schamahmann (2009) localized the topography of use of subjective rather than group ratings may also have
D
emotion-related activation to Crus I in particular. In the contributed to these different results as previous neuroima- o
w
context of our study, the fact that highly extraverted indi- ging studies have tended to rely on standard ratings and nlo
a
vidualsshowedlessarousalmodulationinCrusIandIImay compare across image categories rather than using a para- d
e
indicateunder-sensitivitytotheeffectsofemotionalarousal. metric modulation approach. d fro
However,wealsofoundtheretobeastrongpositivelinear A general observation of our results is that, whereas the m
association between extraversion and arousal processing in linear associations between neuroticism and the neural re- http
therightinsulalobe,whichcontradictsEysenck’spredictions sponse to the emotional stimuli were found in areas which s
aboutthistrait.Theinsulaisthoughttoplayacentralrolein have previously been linked to emotional processing, e.g. ://ac
a
the perception of bodily sensations and in particular the the OFC and the mPFC (Phan et al., 2002; O’Doherty, de
m
perception of pain (Bornhovd et al., 2002); and there is 2004), the relationships we found between extraversion ic
also evidence that this structure has a role in the mainten- and the fMRI results were in regions which would not be .ou
p
ance of drug addiction (Naqvi and Bechara, 2009; Garavan, considered purely ‘emotional’, e.g. the SMA, cerebellum .c
o
2010),aswellasmoregeneralevidenceforitsinvolvementin and parietal cortex. The evidence for a link between extra- m
/s
a myriad different tasks involving emotional processing version levels and emotional processing is not as strong as ca
n
(Lamm and Singer, 2010) and attention (Nelson et al., the body of literature supporting the influence of neuroti- /a
2010). The fact that highly extraverted individuals in our cism on emotional processing at a behavioural (Bolger and rtic
le
study displayed increased insula activity to high levels of Schilling, 1991; Rusting and Larsen, 1997; Kendler et al., -a
b
arousal does not support the notion that extraverts are less 2004; Denissen and Penke, 2008) and neural level (Canli, stra
sensitivetotheeffectsofarousal.However,Kumariandcol- 2004; Haas et al., 2007, 2008; Kumari et al., 2007; Cremers ct/7
leagues (2004) found that high extraversion was associated et al., 2010). It may well be the case that extraversion does /7
/8
with increased activation in the dorsolateral PFC and anter- not play a large modulatory role in affective processing, 5
8
ior cingulate in response to increasing levels of cognitive which is why we do not see strong evidence for variation /1
6
7
demand on an n-back working memory task, indicating in this trait being related to neural activation in emotional 6
3
thatextravertsweremoresensitivetotheeffectsofcognitive processing regions. Given that the associations we observed 00
b
arousal.Itmaywellbethecasethattherelationshipbetween were mostly in motor and parietal regions, it seems likely y
g
extraversion and arousal is a more complex story than that that extraversion may not play a prominent mediating role u
e
s
originally proposed by Eysenck, with further research in the experience of emotion, but rather may affect other t o
required to elucidate if and how arousal processing differ- elements of the task performance captured by these differ- n 0
4
ences are a central feature of this personality trait, and ences in the BOLD signal. The negative association between M
whether there are effects related to emotional arousal, cog- extraversion and the mean BOLD signal for example may ay
2
nitive arousal or both. indicate that those low in this trait were more susceptible 0
1
The negative relationship we identified between extraver- to motor priming effects in readiness to press the button to 9
sion and activation due to positive valence was somewhat respond to the prompt following the image. It is known,
unexpected, as previous studies have found this trait to be for example, that highly emotional stimuli capture atten-
associated with increased activation in response to positive tion faster than neutral stimuli (Bradley, 2009); therefore, it
emotional stimuli (Canli et al., 2001; Haas et al., 2006). It is possible that the associations with arousal and/or valence
may be, however, that our results contrast with previous are due to attentional differences, e.g. high extraversion
neuroimaging studies of personality (e.g. Canli et al., 2001; may be linked to increased susceptibly to the effects of
Haas et al., 2006; Chan et al., 2008) due to the slightly dif- attention. Obviously, there is no way using the current
ferentdesignofourstudy.Forexample,weusedonlyneutral data to disentangle the effects of emotion and other factors
andpositivestimuli,whereastheotherstudiesalsoemployed such as attention, however, it would be interesting in future
negativestimuli(Canlietal.,2001;Canli,2004),whichhave studies to investigate whether extraversion levels influence
been found to activate the key emotional processing centres cognitive rather than affective aspects of emotional
such as the amygdala more substantially than either neutral perception.
Personalityinfluencesonemotionalperception SCAN(2012) 867
The amplitude modulation analysis we employed in this Watson et al., 1988) or POMS (Profile of Mood States;
studyhasbeenusedinotherneuroimagingstudiestoexam- McNair et al., 1971).
ine single trial dynamics, such as in motor learning (Friston There has been some discussion and controversy in the
et al., 1992) and stimulus–response pairings (Buchel et al., past few years about the reliability of studies in social and
1998);however,ithasonlybeenusedtoexaminehowemo- personalityneurosciencewhichcorrelatebrainactivationsas
tionaljudgementsmodulatetheBOLDsignalinafewstudies measured by fMRI with measures of personality (Vul et al.,
(Phan et al., 2004; Heinzel et al., 2005; Anders et al., 2008; 2009; Yarkoni, 2009). Some authors have argued that the
Northoffetal.,2009),andhasnotbeenappliedinthestudy results of such studies may be inflated due to lack of statis-
of personality and emotional processing. It proved highly tical power in small sample sizes (Vul et al., 2009; Yarkoni,
effective at disentangling the effects of emotional arousal 2009), while others have criticized the methods they claim
andvalenceontheBOLDsignal,andtheresultsarebroadly are used by certain experimenters, whereby, region of inter-
D
similar to studies which have used different analysis meth- est analyses are carried out on the basis of whole-brain cor- o
w
ods, such as contrasting the activation maps associated with relations, giving exaggeratedly high correlation coefficients nlo
viewing emotional versus neutral images (Lang et al., 1998; (Vuletal.,2009).Weareconfident,however,thattheresults ad
e
eBtraadl.l,e2y0e0t4)a.l.T,h2e00e1xat,enMsiovuermaoo-dMuilraatinodnaseint athl.,e2o0c0c3ip;iAtanldaenrds ofufntchteiocnurrreelnattesdtutdoypaerresornefalleictyti,vaenodfarerealndoitffespreunrcioesusinrebsrualitns d from
temporal lobes associated with increasing arousal corrobor- or inflated correlations. First, we employed a whole-brain http
ate many neuroimaging studies which have found increases correlation approach which has been validated as a reliable, s
in neural activation associated with increasing emotional independent measure to detect brain–behaviour relationships ://ac
a
arousal in the occipital cortex and inferotemporal regions (Lieberman et al., 2009; Poldrack and Mumford, 2009). de
of the ventral visual pathway (Bradley et al., 2003; Second,weemployedarobustmultiplecomparisonproced- mic
Sabatinelli et al., 2005; Junghofer et al., 2006; Sabatinelli ure which protects against the possibility of false positives. .ou
p
et al., 2007). Furthermore, the arousal-associated increases Furthermore, given the anatomical specificity of our results, .c
o
intheBOLD signalthat we observedin theprefrontal gyrus for example, reduced activation in the OFC associated with m
/s
have been found in others’ studies of emotional processing neuroticism,weareconfidentthatourresultsarenotsimply ca
and evaluation (Grimm et al., 2006). The amplitude modu- spuriouscorrelationsinrandombrainregions.Finally,byin- n/a
lation due to valence during image viewing revealed a less cludingonlywomeninthecurrentstudy,wegreatlyreduced rtic
le
extensive and distinct different pattern of neural activation, the heterogeneity of our sample given that there have been -a
b
inMnocotliuucrdeaaionb-glMysiigrcanonifndiccaaennetttracalltu.esd(t2e0r0sa3rin)ouathnlsedovriestuphoaelrtperfduosciinefsocsrriemnagserdeggyairocutnis-s. ocreofanecmstiidvoeittriyoabn(Ble(rBdaedifclfekeyreeetntacale.l,s.,1f2o909u06n1;dbC)ba,ehitniwllteehetenangl.ee,un2rd0ae0lr1rs;eipWnreeasmgeenortteaitotinaolan.l, stract/7/7
2003; Wrase et al., 2003) and in levels of neuroticism (Lynn /8
vation in the occipitotemporal visual processing areas when 5
8
theycomparedresponseswithpositiveversusneutralimages and Martin, 1997). /1
6
Insummary,ourfindingsofferauniqueinsightintohow 7
with the same arousal level. Lang et al. (1998) found that 6
3
only positive images, not neutral or negative valences, extraversion andneuroticism interactwith theneural repre- 00
sentationofemotionalarousalandvalence,twocriticalcom- b
increased the BOLD signal in the left fusiform gyrus. These y
ponentsofemotionalprocessing.Weidentifiedseveralnovel g
patternsofactivationareverysimilartothemodulationdue u
e
relationshipsbetweenpersonalityandtheemotionalprocess- s
tovalenceinthecurrentstudy,andsuggestthatarousaland t o
ing, notably reduced sustained activation in the OFC and n
valence are represented by quite distinct neural networks. 0
attenuated valence processing associated with high levels 4
The analysis of the fMRI results with respect to the sub- M
ject’s individual ratings is also a relatively novel approach, of neuroticism. These results provide further evidence for ay
the important role that this trait plays in individual re- 2
havingbeenusedonlyrarely(Andersetal.,2008)andweare 0
1
confident that this is preferable to using standard or group sponsestoaffectivestimuli,andtheyalsosuggestsimilarities 9
betweenemotionalprocessinginindividualswithhighlevels
averageratingsasactualemotionalratingsarequitevariable.
of neuroticism and in depressed patients which may help to
We believe that this design approach allowed us the oppor-
elucidate the role that this trait plays in the development of
tunity to explore the interaction of personality with emo-
depression and other affective disorders. Furthering our
tional processing in ways that have not yet been examined.
understanding of individual differences in neural reactivity
Although beyond the scope of the current study, it may be
toemotionalstimuliispivotaltoincreasingourunderstand-
interesting in future studies to compare this approach with
ingoftherolethatpersonalityplaysinemotionalprocessing,
thestandardapproachofusingaverageemotionalratings,in
and may help to elucidate how personality influences the
ordertoseewhateffectthishasonthefMRIresults.Itwould
development of affective disorders.
alsobeinterestingtoinvestigatewhetherthereareanyinter-
actions between mood, personality and emotional percep-
tion, with the administration of a mood scale such as SUPPLEMENTARY DATA
the PANAS (Positive and Negative Affective Scales; Supplementary data are available at SCAN online.