A

ncssfta©

K

1

senc(M

ec

Vf

0d

Neuropsychologia 45 (2007) 2369–2377

Amygdala activation and facial expressions: Explicit emotiondiscrimination versus implicit emotion processing

Ute Habel a,b, Christian Windischberger b,c, Birgit Derntl a,b,d, Simon Robinson b,e,Ilse Kryspin-Exner c,d, Ruben C. Gur f, Ewald Moser b,c,f,g,∗

a Department of Psychiatry and Psychotherapy, University of Aachen, Germanyb MR Centre of Excellence, Medical University of Vienna, Austria

c Center of Biomedical Engineering and Physics, Medical University of Vienna, Austriad Institute for Clinical, Biological and Differential Psychology, Faculty of Psychology, University of Vienna, Austria

e Center of Mind/Brain Sciences, University of Trento, Italyf Department of Psychiatry, University of Pennsylvania, Philadelphia, USA

g Department of Radiology, Medical University of Vienna, Austria

Received 10 August 2006; received in revised form 23 January 2007; accepted 24 January 2007Available online 6 February 2007

bstract

Emotion recognition is essential for social interaction and communication and is a capacity in which the amygdala plays a central role. So far,euroimaging results have been inconsistent as to whether the amygdala is more active during explicit or incidental facial emotion processing. Inonsideration of its functionality in fast automatic evaluation of stimuli and involvement in higher-order conscious processing, we hypothesize aimilar response to the emotional faces presented regardless of attentional focus. Using high field functional magnetic resonance imaging (fMRI)

pecifically optimized for ventral brain regions we show strong and robust amygdala activation for explicit and implicit processing of emotionalacial expressions in 29 healthy subjects. Bilateral amygdala activation was, however, significantly greater when subjects were asked to recognizehe emotion (explicit condition) than when required to discern the age (implicit condition). A significant correlation between amygdala activationnd emotion recognition, but not age discrimination performance, emphasizes the amygdala’s enhanced role during conscious emotion processing.

2007 Published by Elsevier Ltd.

fMRI

eiadGcTef

eywords: Facial emotion recognition; Task modulation; Attention; High-field

. Introduction

The ability to correctly recognize emotions in facial expres-ions plays an essential role in social communication and,volutionarily, is important for survival (Darwin, 1872). Theeural basis of facial emotion processing includes a network ofortical and subcortical structures, centering on the amygdalaAdolphs, 2002; Adolphs, Gosselin et al., 2005; LeDoux, 1995;

orris, Ohman, & Dolan, 1999).

Growing evidence supports the notion that the amygdala is

ssential for several domains of emotional behavior, such as fearonditioning (Blair, Sotres-Bayon, Moita, & LeDoux, 2005),

∗ Corresponding author at: MR Centre of Excellence, Medical University ofienna, Lazarettgasse 14, 1090 Vienna, Austria. Tel.: +43 1 40400 1771;

ax: +43 1 40400 7631.E-mail address: ewald.moser@meduniwien.ac.at (E. Moser).

achas1uf

028-3932/$ – see front matter © 2007 Published by Elsevier Ltd.oi:10.1016/j.neuropsychologia.2007.01.023

motional memory (Adolphs, Tranel, & Buchanan, 2005), moodnduction (Habel, Klein, Kellermann, Shah, & Schneider, 2005)nd emotion discrimination (Gur, Schroeder et al., 2002). Amyg-ala dysfunction has been investigated in brain lesion (Adolphs,osselin et al., 2005; Adolphs, Tranel et al., 2005) and psy-

hiatric patients (Kohler et al., 2003; Surguladze et al., 2004;ownshend & Duka, 2003). The amygdala’s functionality in thearly phases of processing is suggested by rich cortical afferentsrom sensory cortices and fast input routes via the thalamus,s well as extensive output routes to prefrontal and other corti-al and subcortical areas. The amygdala’s input provides bothighly processed and raw information sufficient to prompt fastutomatic responses. Accordingly, the amygdala has been con-

idered “the gateway to the emotions” (Aggleton & Mishkin,986). Its role has been implicated in evaluating whether a stim-lus is pleasant or unpleasant, harmless or dangerous, with aocus on facial expressions of emotions as particularly relevant

2 cholo

sW

tneasudrd22ffirajmrweFa2ertseFat(aaiwtpcsrstaMtfiytTfpr

mrameet

2

2

rSwc((a

2

tpebGi

gsrb

ttdddtteo

2

rt1CfKr(pWi

370 U. Habel et al. / Neuropsy

ources of information (see Hariri, Tessitore, Mattay, Fera, &einberger, 2002).Given this functionality, the amygdala should be responsive

o all facial expressions of emotion, regardless of whether orot attention is directed toward emotional aspects. This hypoth-sis has elicited only limited testing with neuroimaging toolsnd with quite contradictory results. Several studies reportedtronger activation of the amygdala-hippocampal area duringnattended emotion processing, that is, implicit or passive (gen-er or age discrimination), compared to explicit tasks (emotionecognition or positive/negative discrimination), in which theepicted emotion was the focus of attention (Critchley et al.,000; Hariri, Bookheimer, & Mazziotta, 2000; Keightley et al.,003; Lange et al., 2003). The opposite, however, has also beenound: no activation during implicit processing of disgustedaces (Gorno-Tempini et al., 2001) or less activation duringncidental emotional processing compared to explicit emotionecognition (Gur, Schroeder et al., 2002). Winston, O’Doherty,nd Dolan (2003) presented two faces at a time and asked sub-ects to either identify the more emotional (explicit condition) or

ore male face (implicit condition). Task independent amygdalaesponses were found when high- and low-intensity expressionsere compared across four emotions. This study did not, how-

ver, require subjects to actually perform emotion identification.urthermore, an interaction between valence and attention haslso been reported (Williams, McGlone, Abbott, & Mattingley,005), with stronger activation to unattended, potentially threat-ning facial stimuli, in contrast to attended stimuli, whilstesponses to happy faces were greater when attended. Some ofhis divergence may be attributed to methodological issues. Mosttudies applied low-resolution measurement methods, whichxacerbate susceptibility-related signal dropout (Merboldt,ransson, Bruhn, & Frahm, 2001)—particularly affecting themygdala. Furthermore, most studies used a block presenta-ion design, which is particularly vulnerable to habituationi.e. Buchel, Dolan, Armony, & Friston, 1999) and movementrtifacts (Robinson & Moser, 2004). The choice of analysispproaches may also affect results given the small signal changesn fMRI. Finally, implicit tasks of gender or age discriminationith a binary division of stimuli (male/female and younger/older

han 30) and positive/negative decisions are usually much sim-ler than tasks assessing emotion recognition ability, and taskomplexity might disengage emotion processing. The presenttudy was designed to address these issues. We used a high-esolution data acquisition (Echo-Planar-Imaging, EPI) protocolpecifically developed for imaging the amygdala region at 3 T,o reliably detect BOLD (blood-oxygen-level-dependent) basedctivation changes (Robinson, Windischberger, Rauscher, &oser, 2004; Robinson et al., 2005). In particular spatial resolu-

ion, slice orientation and echo time were optimized for the higheld strength and ventral brain region, giving measurements thatield 60 percent higher time-series signal-to-noise ratio (SNR) inhe amygdala than measurements with standard EPI parameters.

he effects of physiological artifacts have been compensated

or in post-processing, increasing sensitivity. An event-relatedresentation design was used, which is more flexible and moreobust against scanner drifts and head motion. The stimulus

uisp

gia 45 (2007) 2369–2377

aterial has been pre-validated, and the implicit task has beene-designed to be more demanding and engaging. Whole-brainnalysis has been supplemented by ROI approaches. With theseethodological improvements we attempted to examine the

xtent to which the attentional focus of the task – implicit orxplicit emotional processing – influences neural activation inhe amygdala.

. Methods

.1. Subjects

Fourteen right-handed healthy females (mean age 24.2 years, SD = 4.09,ange 20–35) and 15 right-handed healthy males (mean age 26.9 years,D = 2.85, range 23–33), all Caucasians, participated in the study. All gaveritten informed consent according to procedures approved by the local ethics

ommittee and the Helsinki declaration. Exclusion of psychiatric disordersaccording to DSM IV) was ascertained by the Structured Clinical InterviewGerman Version of the SCID). The usual exclusion criteria for MRI werepplied.

.2. Tasks and stimuli

In total 42 colored Caucasian facial identities depicting five basic emo-ions (happiness, sadness, anger, fear and disgust) and neutral expressions wereresented—30 for explicit and 12 for implicit emotion recognition. The facialxpressions used for the study (for age as well as emotion discrimination) wereoth selected from a standardized stimulus set (Gunning-Dixon et al., 2003;ur, Sara et al., 2002) that was comparable with respect to gender, age, emotion

ntensity, emotional valence, and brightness.Stimulus presentation was randomized with regard to task, emotion and

ender but kept constant for all subjects. Emotional facial expressions were pre-ented for a maximum of 5 s with a randomized, variable inter-stimulus intervalanging from 12 to 18 s. A scrambled face with a central crosshair served asaseline. Responses triggered immediate progression to the next baseline period.

In the explicit task subjects had to select which emotion was portrayed fromwo alternatives presented verbally on the right and left of the face using awo-button response device. The implicit task was to judge which of two ageecades was closer to the poser’s age (see Fig. 1 for examples of tasks andiscrimination accuracy). To prevent possible habituation effects, different facesepicting emotional expressions were used for age discrimination. However,hey were taken from the same stimulus pool to maximize the similarity of theasks. Explicit and implicit trials were randomly distributed and presented in anvent-related paradigm in one single run. Furthermore, each actor appeared onlynce, hence order effects for identities or emotions were controlled.

.3. MRI acquisition and data analysis

Functional imaging was performed at 3 T using high-resolution gradient-ecalled echo planar imaging (EPI). Ten oblique axial slices centered onhe amygdala were acquired using asymmetric k-space sampling (matrix size28 × 91, slice thickness 2 mm, slice gap 0.5 mm, TR = 1,000 ms, TE = 31 ms).ardiac action and breathing were digitally recorded to allow physiological arti-

act correction in post-processing (Windischberger et al., 2002; Windischberger,ilzer, & Moser, 2004) using software developed in-house, based on the algo-

ithms PHYSIOFIX (Hu, Le, Parrish, & Erhard, 1995) and RETROICORGlover, Li, & Ress, 2000). In a recent study we quantitatively compared theerformance of these algorithms investigating a variety of “flavours” (Kilzer,indischberger, & Moser 2003) and the variant that performed best was applied

n this study.

Functional data were preprocessed using SPM2 (http://www.fil.ion.

cl.ac.uk/spm/). Images were slice timing corrected, realigned to the meanmage, normalized into the standardized stereotactic space, and spatiallymoothed (9 mm isotropic Gaussian kernel). Functional data were corrected forhysiological artifacts using the physiofix and retroicor algorithms as described

U. Habel et al. / Neuropsychologia 45 (2007) 2369–2377 2371

F recognr

arsttcapta

wnsrNbs(esLvat(dwc

ptd

ct

3

3

mftRm2heptTatdAt(ttrp

ig. 1. Stimulus material and behavioral performance. (a) Example for emotionecognition and age discrimination for female and male subjects.

bove. Each stimulus was modeled separately with the canonical hemodynamicesponse function and its temporal derivative. An additional box-car regres-or without hemodynamic delay was used to account for signal changes dueo head motion during stimulus presentation. Trials were modeled based onhe individual stimulus presentation periods (“trial length”) convolved with theanonical hemodynamic response function as implemented in the SPM pack-ge, that is the model was different for each subject. Statistical analysis waserformed at the individual and group levels. To detect group differences, con-rast images of all subjects were included in a second level random effectsnalysis.

Since our main hypothesis focused on the amygdala and results of the voxel-ise whole brain analysis might be affected by misplacements due to imperfectormalization, we performed a ROI analysis with the aim of maximizing theensitivity and specificity of amygdala results. A further aim was to consolidateesults and analyze possible hemispheric lateralization effects in greater detail.ote that the data sets were spatially smoothed by a Gaussian kernel of 9 mm,efore whole-brain and ROI analysis, which corresponds approximately to theize of the amygdala. ROI definition was based on the MRIcro amygdala templatehttp://www.psychology.nottingham.ac.uk/staff/cr1/mricro.html). Mean param-ter estimates were extracted for left and right amygdala in each condition andubject using IDL (Interactive Data Language, RSI Inc., Boulder, CO, USA).evene tests for homogeneity of variances established homoscedasticity for allariables, right (p = 0.16) and left amygdala (p = 0.22) activation during emotions well as age discrimination (right: p = 0.28; left: p = 0.27). These values werehen entered in a three-way analysis of variance with repeated factors conditionexplicit, implicit) and laterality (left, right) and the between-subject factor gen-er. Significant activation in each hemisphere and during each task (Table 1)as confirmed by one sample t-tests (one-tailed) (Table 1). Greenhouse-Geisser

orrected p-values are presented.Correlation analysis was performed between recognition accuracy (mean

ercent correct) and BOLD effect (voxel with highest parameter estimate within

he amygdala as shown by SPM random effects analysis) for emotion and ageiscrimination.

Results for the whole brain analysis of explicit and implicit emotion pro-essing separately, and direct comparisons are presented at an FWE correctedhreshold of p < 0.05.

ttae

ition (left) and age discrimination task (right). (b) Mean accuracies for emotion

. Results

.1. Behavioral performance

Emotion recognition accuracy was 90.3 percent (±5.9,ean ± SD) on average for male and 90.6 percent (±5.8) for

emale subjects. The mean percent correct for age discrimina-ion was 39.1 percent (±10.1) and 45.1 (±9.8), respectively.eaction time for emotion recognition was 2.5 s (±0.3) inales and 2.3 s (±0.4) in females, for age discrimination

.5 s (±0.3) and 2.6 s (±0.3), respectively. Levene tests foromogeneity of variances established homoscedasticity formotion recognition (p = 0.94) and age discrimination (p = 0.90)erformance. Shapiro–Wilk tests demonstrated a normal dis-ribution for both variables (emotion p = 0.16; age p = 0.11).he two-way ANOVA with gender as between-subject factornd task as within-subject factor revealed only a significantask difference (F(1,23) = 618.9, p = 0.0001) without any gen-er differences (F(1,23) = 1.40; p = 0.25). In the equivalentNOVA for reaction time (Levene tests: emotion recogni-

ion (p = 0.68) and age discrimination (p = 0.83); Shapiro–Wilksemotion p = 0.57; age p = 0.99)) similarly a main effect forask emerged (F(1,23) = 5.74, p = 0.03) as well as a significantask-by-gender interaction (F(1,23) = 7.85, p = 0.01), reflectingelatively faster response times for women on emotion com-ared to age identification. However, post hoc tests decomposing

his interaction remained not significant. For age discrimina-ion reaction time was not significantly different between malesnd females (t(23) = −1.11, p = 0.28) as was reaction time formotion discrimination (t(23) = 1.83, p = 0.08).

2372 U. Habel et al. / Neuropsychologia 45 (2007) 2369–2377

Table 1Main activated brain regions, corresponding MNI coordinates and Z-scores for (a) explicit emotion recognition (EMO), (b) implicit face processing (AGE), and (c)direct comparison between explicit and implicit processing (EMO-AGE)

MNI coordinates Z-value p-Value (FWE corr.) Region

X Y Z

EMO−10 −54 −18 6.71 <0.001 Left fusiform gyrus−4 −24 −10 6.64 <0.001 Left brainstem

4 −26 −10 6.62 <0.001 Right brainstem14 −18 −8 5.82 <0.001 Right hippocampus

−18 −26 −8 5.52 <0.001 Left hippocampus20 4 −14 5.35 <0.001 Right amygdala

−52 16 −24 4.69 <0.008 Left inferior frontal gyrus−20 0 −20 4.56 <0.009 Left amygdala

40 20 −22 4.46 <0.019 Right superior temporal pole−44 20 −24 4.28 <0.038 Left superior temporal pole

AGE42 −58 −22 6.64 <0.001 Right fusiform gyrus

−10 −56 −18 6.54 <0.001 Left cerebellum−26 −56 −18 6.28 <0.001 Left fusiform gyrus−12 −24 −10 5.35 <0.001 Left hippocampus

20 −28 −8 4.93 <0.005 Right hippocampus−20 4 −18 4.84 <0.007 Left amygdala

18 4 −18 4.65 <0.009 Right amygdala46 18 −18 4.51 <0.012 Right superior temporal pole38 28 −28 4.16 <0.043

EMO-AGE−28 −58 −16 6.38 <0.001 Left fusiform gyrus

38 −40 −20 6.07 <0.001 Right fusiform gyrus−2 −24 −12 5.82 <0.001 Left brainstem48 18 −20 5.38 <0.001 Right superior temporal pole

−50 14 −24 4.65 <0.012 Left superior temporal pole−16 −2 −12 4.57 <0.017 Left amygdala

I at the

3

3

athtiattRasttcba

ua

Aotrda

3

wirpate

be

−40 20 −22 4.37

mplicit vs. explicit processing (AGE-EMO) revealed no significant difference

.2. Functional data

.2.1. Whole brain analysesThis study was specifically designed to clarify the role of the

mygdala in recognizing and processing basic emotions, inves-igating the influence of task instruction on its activation. Asypothesized, significant BOLD signal increases were found inhe amygdala bilaterally, during explicit and implicit process-ng of emotional faces (Figs. 2a and b and 3). There was alsoctivation in adjacent areas: the hippocampus, fusiform gyrus,emporal pole and brainstem. Differences between tasks wereested by paired t-tests (explicit–implicit and implicit–explicit).esults revealed significant differences showing stronger leftmygdala activation during explicit emotion recognition, as pre-ented in Fig. 3. A higher degree of fusiform, brainstem andemporal activation also emerged during emotion discrimina-ion, pointing to emotion-specific functions of these areas duringonscious processing. See Table 1 summarizing main activatedrain regions for explicit and implicit face processing separately

s well as for direct comparison.

To exclude the possibility that findings resulted from annequal number of stimuli in each condition, we performed andditional analysis pooling randomly 12 emotion task stimuli.

an

r

<0.036 Left superior temporal pole

threshold of p < 0.05 FWE corrected.

lthough the effect was somewhat attenuated and significantnly at a level of p = 0.001 uncorrected, the contrast of 12 emo-ion stimuli against the 12 age stimuli revealed nearly identicalesults: explicit emotional processing produced a stronger amyg-ala response than implicit processing of such stimuli during thege identification task.

.2.2. ROI analysesThe three-way repeated measures analysis of variance (Fig. 4)

ith gender as between-subject factor and condition (explicit,mplicit) and laterality (left, right) as within-subject factorsevealed a significant main effect for condition (F(1, 27) = 20.63;= 0.0001) without any further significant main effects or inter-ctions. Hence the ROI analysis confirmed the whole brain resulthat the amygdala response is significantly enhanced duringxplicit processing.

Fig. 5 shows the mean activation differences in the amygdalaetween the explicit and implicit emotion processing task, forach subject. Of the 29 participants 26 demonstrated reduced

mygdala response during age compared to emotion discrimi-ation.

To control for the different number of stimuli and in cor-espondance to the whole brain analysis, an ROI analysis

U. Habel et al. / Neuropsychologia 45 (2007) 2369–2377 2373

Fig. 2. Group results overlaid on mean echo-planar images revealing bilateral amygdala activation (p = 0.05, FWE corrected) as well as temporal and fusiformgyrus, cerebellum and brainstem involvement for explicit emotion processing (emotion recognition, a), implicit emotion processing (age discrimination, b) and thedifference between explicit and implicit emotion processing (c).

F processing overlaid on a surface rendered brain showing strong bilateral amygdalaa

(p2

vsmltfCir

Table 2Region of interest analysis for the amygdala

Amygdala parameter estimates

Left p-Value Right p-Value

Emotion recognition 2.92 ± 0.42 <0.0001 2.59 ± 0.37 <0.0001Age discrimination 2.16 ± 0.48 <0.0001 2.05 ± 0.43 <0.0001

ig. 3. Group activation results for explicit (left) and implicit (right) emotionctivation during both conditions.

ANOVA) using only 12 emotion and 12 age stimuli was alsoerformed and confirmed main analysis results (condition: F(1,7) = 7.24, p = 0.01), see Table 2.

A significant correlation between activation levels in theoxel with highest activation (located in the right amygdala ashown by SPM random effects analysis) and behavioral perfor-ance during emotion recognition (r = 0.42, p = 0.04, see Fig. 6

eft) further supports the role of the amygdala in conscious emo-ional processing. No corresponding correlation was observed

or age discrimination (right: r = 0.19, p = 0.36, see Fig. 6 right).orrelations for the left amygdala, however, remained not signif-

cant (emotion recognition r = 0.29, p = 0.16, age discrimination= 0.21, p = 0.31).

Paten

arameter estimates (±SEM) of activation during both tasks, separately for leftnd right amygdala ROI (12 emotion vs. 12 age stimuli). Values have beenested on their significance applying one-sample t-tests. No significant differ-nces between left and right amygdala were obtained for emotion recognition,or age discrimination.

2374 U. Habel et al. / Neuropsycholo

Fig. 4. ROI results for the amygdala. Mean parameter estimates (±SEM)extracted from the amygdala ROI for male and female subjects, and left andright amygdala regions. Only a main effect of condition emerged, pointing tohigher responsivity of the amygdala during explicit emotion processing, an effectwhich is similar in both genders and for both amygdalae.

Fig. 5. Scatterplot of individual differences of mean parameter estimates in theafm

4

tafi

oist2stia

udmtaowthtbfgfftartebmhtettegories may be less distinct and more dimensional in nature,

Fra

mygdala ROIs between explicit and implicit emotion processing for male andemale subjects indicating the stronger response during the explicit condition inost of the subjects.

. Discussion

Our results demonstrate that the amygdala responds to emo-ional facial stimuli irrespective of attentional focus, but strongerctivation is elicited during explicit emotion recognition. Thisnding was corroborated in single subject results, as 90 percent

mhl

ig. 6. Scatterplot of correlation analysis between voxel with highest activation takecognition (left) and implicit emotion processing (right) across all subjects indicatinmygdala response only.

gia 45 (2007) 2369–2377

f the study sample showed stronger amygdala activation dur-ng the explicit condition. The finding replicates earlier studieshowing greater amygdala activation for explicit emotion iden-ification tasks (e.g., Gur, Schroeder et al., 2002; Gur, Sara et al.,002). The correlation of behavioral data and amygdala responseupports the notion that the amygdala is not only involved inhe initial evaluation of emotional stimuli, but may also play anmportant role in more complex processing of these stimuli suchs their appraisal and labeling.

Several methodological issues need to be considered in eval-ating the results, and these include imaging approach and taskifficulty. The present study improves on earlier efforts by opti-izing the SNR from the amygdala. The fMRI signal from

he amygdala region is prone to poor SNR if standard fMRIcquisition parameters are applied (Robinson et al., 2004). Inrder to ensure reliable signal from the amygdala in this studye have employed an acquisition protocol that has been sys-

ematically optimized for fMRI in ventral brain regions, whichas been shown to yield 60 percent higher time-series SNR inhe amygdala than standard protocols despite higher receiverandwidth. Concerning task difficulty, activation was greateror the emotion than the age identification task in spite of thereater difficulty of the latter—as evident in the behavioral per-ormance. The greater difficulty might have prevented subjectsrom focusing on the emotional features. It seems reasonablehat healthy subjects have more difficulties in recognizing thege of a face than the emotion due to its greater evolutionaryelevance. Nonetheless, strong and reliable amygdala participa-ion was documented for both conditions. Task difficulty couldxplain differences in amygdala activation as recently discussedy Pessoa, Japee, Sturman, and Ungerleider (2005). Further-ore, a higher difficulty could lead to a loss of engagement and

ence reduced attention. However, since trials of age and emo-ion discrimination were randomized within one run, long-termffects of one task are minimized. Also, emotion discrimina-ion is a categorical task in which the subject can clearly definehe distinct emotion of a face. For age discrimination the cat-

aking it more difficult to correctly infer the age. Finally, per-aps stronger amygdala responses occurred for the easier task asess frontal influence on amygdala activation can be expected in

en from SPM random effects analysis and performance for explicit emotiong a significant positive correlation between emotion recognition accuracy and

cholo

twtdna

4

LPsswan(erea

iodrebitsEsa

4

e1wsdfMrmmctatsnpi

4

ccutte(tetgAt

ss(thipeafd

deisbfGtttcrme

4

aca–ea

U. Habel et al. / Neuropsy

hat case. For the harder task of age discrimination, interferenceith higher cortical prefrontal areas involved in the detec-

ion of age clues might have lead to less amygdala activationue to inhibition effects (e.g. Levesque et al., 2003). Unfortu-ately, our restricted brain coverage prevented analysis of frontalctivation.

.1. Hemispheric lateralization

In contrast to other results (Baas, Aleman, & Kahn, 2004;ieberman, Hariri, Jarcho, Eisenberger, & Bookheimer, 2005;hillips et al., 2001), our findings do not suggest a specific hemi-pheric lateralization of amygdala response. The stronger leftided activation during explicit processing detected in the SPMhole brain analysis has not been confirmed by the ROI-based

nalysis. This is in line with lesion studies where emotion recog-ition deficits occurred only when bilateral damage was presentAdolphs, Tranel, Damasio, and Damasio 1994, 1995; Adolphst al., 1999; Schmolck & Squire, 2001). In addition, bilateralesponse to the perception and evaluation of biologically rel-vant sensory stimuli seems strongly supported by interactionnalysis (Das et al., 2005).

Hemispheric lateralization of amygdala function may benfluenced by external as well as internal factors. A recent reviewf 54 fMRI and PET studies (Baas et al., 2004) suggests a pre-ominant left amygdala activation, which was, however, notelated to stimulus type, task instructions, habituation differ-nces or elaborate processing. Hence, lateralized findings maye influenced by methodological factors such as differentialmage signal dropout, measurement sensitivity and divergentask demands. Our results have been validated using two analysistrategies in addition to applying an optimized high resolutionPI protocol (Robinson et al., 2004). The ROI analysis pos-esses higher sensitivity to reveal the exact characterization ofmygdala response with respect to lateralization.

.2. The role of adjacent regions around the amygdala

The data acquisition protocol was specifically designed toxamine signal changes in the amygdala and hence limited to0 slices covering this region. However, several other areasere involved during both tasks, namely the fusiform gyrus, the

uperior temporal pole and the brainstem, with more activationuring explicit processing. The superior temporal cortex andusiform gyrus are known as face-sensitive areas (Kanwisher,

cDermott, & Chun, 1997; Pourtois et al., 2004), and henceelevant for emotion as well as age discrimination in a face. Theore pronounced response during explicit emotion processingay reflect the more detailed facial analysis. Alternatively, it

ould reflect the generally stronger brain response to basic emo-ions, due to their evolutionary greater relevance (compared toge information), even though the age discrimination task provedo be harder. These activations demonstrate that the intact con-

cious and incidental emotion processing relies on a widespreadetwork of participating regions in which the amygdala maylay a special role. The amygdala, however, is not sufficient forntact processing.

ict(

gia 45 (2007) 2369–2377 2375

.3. Methodological constraints

The study has several methodological limitations. Two verbalategories were used for emotion recognition and two numericalategories appeared next to the face for age discrimination. These of emotional words as labels could have possibly influencedhe neural activation. This procedure was however necessaryo effectively compare explicit recognition of the presentedmotion with implicit processing. Notably, Bleich-Cohen et al.2006) found no task effect on the amygdala response to emo-ional faces, when subjects were asked to identify the presentedmotion or gender of poser, respectively. However, with theseasks explicit recognition of the emotions could not be investi-ated because no overt behavioral performance was assessed.lso, verbal age decades would have been more comparable to

he emotion task.The number of stimuli was not equal for the two tasks. While a

ubsequent analysis showed that effects were similar when a sub-ample of the emotion-identification task stimuli was selectedalbeit at an uncorrected threshold, p < 0.001), it is more difficulto guarantee that different emotional-cognitive processes mightave not occurred. However, we used a long inter-stimulus-nterval of about 15 s, separating each trial from the other. Thisrocedure enabled modeling of the hemodynamic response forach stimulus separately. The regions of interest around themygdala revealed quite similar activation patterns, with dif-erences mainly in intensity, which points to similar processesespite divergent instructions.

Task difficulty poses another potential limitation. The ageiscrimination task was designed to be comparable to themotion task with respect to stimulus material with the tasknstruction being the exclusive difference. In former studies alightly modified version of the age discrimination task haseen applied, in which subjects had to decide whether theace is younger or older than 30 (Gur, Schroeder et al., 2002;ur, Sara et al., 2002; Schneider et al., 2006). This instruc-

ion makes the task easy and perhaps subjects can still directheir attention to the emotion after successful completion ofhe age discrimination task. Debriefing of our participants indi-ated that subjects were indeed attending to the age withoutegarding the poser’s emotion. Activation in the amygdalaay, therefore, be unavoidable during the processing of facial

motions.

.4. Summary and conclusion

Our experimental design was aimed at ensuring strongmygdala signal and high sensitivity to provide the basis foromputing robust activation maps. The results reveal amygdalactivation during both tasks, with less – but highly significantactivation in the implicit age discrimination compared to the

xplicit emotion identification task. Thus it can be concluded thatctivation in the amygdala is present during both conditions but

s greater in the explicit compared with the implicit emotion dis-rimination task. This effect replicates in an event-related designhe results of Gur, Schroeder et al. (2000) and Gur, Sara et al.2002) who reported, with a blocked design, greater amygdala

2 cholo

aa

psspaspvbi

A

(

R

A

A

A

A

A

A

A

B

B

B

B

C

D

D

G

G

G

G

G

H

H

H

H

K

K

K

K

L

L

L

L

M

M

P

P

376 U. Habel et al. / Neuropsy

ctivation for an emotion-identification condition compared ton age-identification condition.

Survival, especially in a hostile context, requires the ability toroduce fast and appropriate responses to emotionally relevanttimuli. Due to its multiple connections, the amygdala is welluited for this function. Our results indicate that the amygdalalays a more general role in emotion processing than previouslyssumed. Most studies on subliminal processing of emotionaltimuli did not compare amygdala activation to more consciousrocessing forms. We conclude that the extent of amygdala acti-ation is task-dependant, and stronger responses can be elicitedy directing attention towards emotional aspects of the stimulusncluding recognition of the expression.

cknowledgements

This work has been supported by the Austrian Science FundFWF grant P16669-B02 to E.M.) and grant MH60722 to R.C.G.

eferences

dolphs, R. (2002). Neural systems for recognizing emotion. Current Opinionin Neurobiology, 12, 169–177.

dolphs, R., Gosselin, F., Buchanan, T. W., Tranel, D., Schyns, P., & Damasio,A. R. (2005). A mechanism for impaired fear recognition after amygdaladamage. Nature, 433, 68–72.

dolphs, R., Tranel, D., & Buchanan, T. W. (2005). Amygdala damage impairsemotional memory for gist but not details of complex stimuli. Nature Neu-roscience, 8, 512–518.

dolphs, R., Tranel, D., Damasio, H., & Damasio, A. R. (1994). Impaired recog-nition of emotion in facial expressions following bilateral damage to thehuman amygdala. Nature, 372, 669–672.

dolphs, R., Tranel, D., Damasio, H., & Damasio, A. R. (1995). Fear and thehuman amygdala. Journal of Neuroscience, 15, 5879–5891.

dolphs, R., Tranel, D., Hamann, S., Young, A. W., Calder, A. J., Phelps, E. A.,et al. (1999). Recognition of facial emotion in nine individuals with bilateralamygdala damage. Neuropsychologia, 37, 1111–1117.

ggleton, J. P., & Mishkin, M. (1986). The amygdala sensory gateway to theemotions. In R. Plutchick & H. Kellerman (Eds.), Emotion: theory, researchand experience. biological foundations of emotion (pp. 281–299). Orlando,FL: Academic Press.

aas, D., Aleman, A., & Kahn, R. S. (2004). Lateralization of amygdalaactivation: A systematic review of functional neuroimaging studies. BrainResearch, Reviews, 45, 96–103.

lair, H. T., Sotres-Bayon, F., Moita, M. A., & LeDoux, J. E. (2005). The lateralamygdala processes the value of conditioned and unconditioned aversivestimuli. Neuroscience, 133, 561–569.

leich-Cohen, M., Mintz, M., Pianka, P., Andelman, F., Rotshtein, P., & Hendler,T. (2006). Differential stimuli and task effects in the amygdala and sensoryareas. Neuroreport, 17, 1391–1395.

uchel, C., Dolan, R. J., Armony, J. L., & Friston, K. J. (1999). Amygdala-hippocampal involvement in human aversive trace conditioning revealedthrough event-related functional magnetic resonance imaging. Journal ofNeuroscience, 19, 10869–10876.

ritchley, H., Daly, E., Phillips, M., Brammer, M., Bullmore, E., Williams,S., et al. (2000). Explicit and implicit neural mechanisms for processing ofsocial information from facial expressions: A functional magnetic resonanceimaging study. Human Brain Mapping, 9, 93–105.

arwin, C. (1872). The expression of emotions in man and animals. London:John Murray.

as, P., Kemp, A. H., Liddell, B. J., Brown, K. J., Olivieri, G., Peduto, A., et al.(2005). Pathways for fear perception: Modulation of amygdala activity bythalamo-cortical systems. NeuroImage, 26, 141–148.

P

gia 45 (2007) 2369–2377

lover, G. H., Li, T. Q., & Ress, D. (2000). Image-based method for retrospectivecorrection of physiological motion effects in fMRI: RETROICOR. MagneticResonance in Medicine, 44, 162–167.

orno-Tempini, M. L., Pradelli, S., Serafini, M., Pagnoni, G., Baraldi, P., Porro,C., et al. (2001). Explicit and incidental facial expressions processing: AnfMRI study. NeuroImage, 14, 465–473.

unning-Dixon, F. M., Gur, R. C., Perkins, A. C., Schroeder, L., Turner, T.,Turetsky, B. I., et al. (2003). Age-related differences in brain activationduring emotional face processing. Neurobiology of Aging, 24, 285–295.

ur, R. C., Schroeder, L., Turner, T., McGrath, C., Chan, R. M., Turetsky, B. I., etal. (2002). Brain activation during facial emotion processing. NeuroImage,16, 651–662.

ur, R. C., Sara, R., Hagendoorn, M., Marom, O., Hughett, P., Macy, L., etal. (2002). A method for obtaining 3-dimensional facial expressions and itsstandardization for use in neurocognitive studies. Journal of NeuroscienceMethods, 115, 137–143.

abel, U., Klein, M., Kellermann, T., Shah, N. J., & Schneider, F. (2005). Sameor different? Neural correlates of happy and sad mood in healthy males.NeuroImage, 26, 206–214.

ariri, A. R., Bookheimer, S. Y., & Mazziotta, J. C. (2000). Modulating emo-tional responses: Effects of a neocortical network on the limbic system.Neuroreport, 11, 43–48.

ariri, A., Tessitore, A., Mattay, V. S., Fera, F., & Weinberger, D. R. (2002). Theamygdala response to emotional stimuli: a comparison of faces and scenes.NeuroImage, 17, 317–323.

u, X., Le, T. H., Parrish, T., & Erhard, P. (1995). Retrospective estimationand correction of physiological fluctuation in functional MRI. MagneticResonance in Medicine, 34, 201–212.

anwisher, N., McDermott, J., & Chun, M. M. (1997). The fusiform face area: Amodule in human extrastriate cortex specialized for face perception. Journalof Neuroscience, 17, 4302–4311.

eightley, M. L., Winocur, G., Graham, S. J., Mayberg, H. S., Hevenor, S.J., & Grady, C. L. (2003). An fMRI study investigating cognitive modula-tion of brain regions associated with emotional processing of visual stimuli.Neuropsychologia, 41, 585–596.

ilzer, M., Windischberger, C., & Moser, E. (2003). A quantitative comparisonof algorithms for physiological artifacts correction. MAGMA, 16, S152.

ohler, C. G., Turner, T. H., Bilker, W. B., Brensinger, C. M., Siegel, S. J.,Kanes, S. J., et al. (2003). Facial emotion recognition in schizophrenia:Intensity effects and error pattern. American Journal of Psychiatry, 160,1768–1774.

ange, K., Williams, L. M., Young, A. W., Bullmore, E. T., Brammer, M. J.,Williams, S. C., et al. (2003). Task instructions modulate neural responsesto fearful facial expressions. Biological Psychiatry, 53, 226–232.

eDoux, J. E. (1995). Emotion: Clues from the brain. Annual Reviews in Psy-chology, 46, 209–235.

evesque, J., Eugene, F., Joanette, Y., Paquette, V., Mensour, B., Beaudoin, G.,et al. (2003). Neural circuitry underlying voluntary suppression of sadness.Biological Psychiatry, 53, 502–510.

ieberman, M. D., Hariri, A., Jarcho, J. M., Eisenberger, N. I., & Bookheimer,S. Y. (2005). An fMRI investigation of race-related amygdala activityin African-American and Caucasian-American individuals. Nature Neuro-science, 8, 720–722.

erboldt, K. D., Fransson, P., Bruhn, H., & Frahm, J. (2001). Functional MRIof the human amygdala? NeuroImage, 14, 253–257.

orris, J. S., Ohman, A., & Dolan, R. J. (1999). A subcortical pathway tothe right amygdala mediating “unseen” fear. Proceedings of the NationalAcademy of Sciences of the United States of America, 96, 1680–1685.

essoa, L., Japee, S., Sturman, D., & Ungerleider, L. G. (2005). Target visibilityand visual awareness modulate amygdala response to fearful faces. CerebralCortex, 16, 366–375.

hillips, M. L., Medford, N., Young, A. W., Williams, L., Williams, S. C., Bull-more, E. T., et al. (2001). Time courses of left and right amygdalar responses

to fearful facial expressions. Human Brain Mapping, 12, 193–202.

ourtois, G., Sander, D., Andres, M., Grandjean, D., Reveret, L., Olivier, E.,et al. (2004). Dissociable roles of the human somatosensory and superiortemporal cortices for processing social face signals. European Journal ofNeuroscience, 20, 3507–3515.

cholo

R

R

R

S

S

S

T

W

W

W

artifacts in BOLD-EPI of the human brain. Magnetic Resonance Imaging,

U. Habel et al. / Neuropsy

obinson, S., & Moser, E. (2004). Positive results in amygdala fMRI: Emotionor head motion. NeuroImage, 22, S47.

obinson, S., Hoheisel, B., Windischberger, C., Habel, U., Lanzenberger, R., &Moser, E. (2005). FMRI of the emotions: Towards an improved understand-ing of amygdala function. Current Medical Imaging Reviews, 1, 115–129.

obinson, S., Windischberger, C., Rauscher, A., & Moser, E. (2004). Optimized3T EPI of the amygdalae. NeuroImage, 22, 203–210.

chmolck, H., & Squire, L. R. (2001). Impaired perception of facial emotionsfollowing bilateral damage to the anterior temporal lobe. Neuropsychology,15, 30–38.

chneider, F., Gur, R. C., Koch, K., Backes, V., Amunts, K., Shah, N. J., et al.(2006). Impairment in the specificity of emotion processing in schizophrenia.

American Journal of Psychiatry, 163, 442–447.

urguladze, S. A., Young, A. W., Senior, C., Brebion, G., Travis, M. J., &Phillips, M. L. (2004). Recognition accuracy and response bias to happy andsad facial expressions in patients with major depression. Neuropsychology,18, 212–218.

W

gia 45 (2007) 2369–2377 2377

ownshend, J. M., & Duka, T. (2003). Mixed emotions: Alcoholics’ impairmentsin the recognition of specific emotional facial expressions. Neuropsycholo-gia, 41, 773–782.

illiams, M. A., McGlone, F., Abbott, D. F., & Mattingley, J. B. (2005). Dif-ferential amygdala responses to happy and fearful facial expressions dependon selective attention. NeuroImage, 24, 417–425.

indischberger, C., Kilzer, M., & Moser, E. (2004). The importance of correct-ing for physiological artifacts for functional MRI in deep brain structures.NeuroImage, 22, S28.

indischberger, C., Langenberger, H., Sycha, T., Tschernko, E. M., Fuchsjager-Mayerl, G., Schmetterer, L., et al. (2002). On the origin of respiratory

20, 575–582.inston, J. S., O’Doherty, J., & Dolan, R. J. (2003). Common and distinct

neural responses during direct and incidental processing of multiple facialemotions. NeuroImage, 20, 84–97.