Corresponding: Sex Differences In Amygdala
NEUROREPORT
MOTIVATION, EMOTION, FEEDING, DRINKING
Sex differences in amygdala activation during the perception of facial affect
William D. S. KillgoreCA and Deborah A. Yurgelun-Todd
Cognitive Neuroimaging Laboratory, Brain Imaging Center, McLean Hospital/Harvard Medical School, 115 Mill Street, Belmont,
MA 02478, USA
CA
Corresponding Author
Received 11 April 2001; accepted 5 June 2001
The cognitive and affective systems of the cerebral cortex are often more lateralized in males than females, but it is unclear whether these differences extend to subcortical systems. We used fMRI to examine sex differences in lateralized amygdala activity during happy and fearful face perception. Amygdala activation differed for men …show more content…
and women depending on the valence of the expression. Overall, males were more lateralized than females, but the direction differed between valence
conditions. Happy faces produced greater right than left amygdala activation for males but not females. Both sexes showed greater left amygdala activation for fearful faces. These
®ndings suggest that the lateralization of affective function may extend beyond the cortex to subcortical regions such as the amygdala. NeuroReport 12:2543±2547 & 2001 Lippincott Williams & Wilkins.
Key words: Affect; Amygdala; Cerebral laterality; Emotion; Faces; Fear; fMRI; Functional neuroimaging; Happy; Sex differences; Visual perception
INTRODUCTION
Men and women appear to differ in the manner in which they perceive, process, express, and experience emotion.
On the whole, females tend to be more emotionally expressive than males and the two sexes show differences in their responsiveness to affectively charged stimuli [1].
Females tend to rate their emotions more intensely [2], show stronger evoked potential responses to emotional faces [3] and demonstrate greater facility at decoding nonverbal messages than males [4]. Unfortunately, women also have a rate of depression that is estimated to be 2±3 times higher than that of men [5]. Although the existence of emotional processing differences between the sexes is becoming better recognized, the factors that produce these differences are still not well understood. While many of the differences between men and women in the processing of emotion may be attributable to social factors and learned patterns of behavior [2], it is likely that emotional differences also re¯ect genotypic differences in the sexual dimorphism of the nervous system.
One of the most obvious manifestations of sex differences in the structure and function of the nervous system is exempli®ed by the ®nding that men and women show different patterns of cerebral lateralization of function [6].
Sex differences in cerebral laterality have been documented in the neuropsychological literature for decades [6]. Across a number of studies, a consistent pattern of sex-speci®c lateralization of function has emerged, with males typically demonstrating greater lateralization of language and visuospatial abilities, whereas females tend to be less lateralized, showing a greater frequency of bilateral representation of verbal abilities across the two hemi-
0959-4965 & Lippincott Williams & Wilkins
spheres [7]. The sexually dimorphic structure of the nervous system appears to produce consistent asymmetries in affective function, such as the lateralized perception of the affective features of facial expressions of emotion [8±10], and recent studies have suggested that males may also demonstrate a greater extent of lateralization of emotional perception than females [9,10].
At the cortical level, emotional processing is associated with consistent patterns of lateralized activity as measured by EEG methods, with positive affect or approach-related emotions associated with relatively greater left than right prefrontal activation, and negative affect or withdrawalrelated emotions associated with the opposite pattern [11].
While lateralization of cognitive abilities is often observed within the neocortex, the extent of functional lateralization within subcortical structures is uncertain.
There is some evidence of a lateralized pattern of amygdala activity during affective processing, with several studies ®nding greater left than right amygdala activation during negative emotional facial expressions such as fear [12], and sex differences in this lateralized activity may be evident as early as adolescence [13]. In contrast to negative emotions, there have been few studies examining the functional responsiveness of the amygdala to positive emotions such as happiness. The laterality of amygdala activation during the perception of happy faces has been less consistent than the ®ndings for fearful faces [12,14] and no functional imaging studies have speci®cally examined the in¯uence of sex differences on the activation of the amygdala during the perception of either happy or fearful faces. We chose to examine whether the theory of sex-dependent lateralization of function that accounts for ®ndings within the neo-cortex
Vol 12 No 11 8 August 2001
2543
NEUROREPORT could be extended to also account for subcortical processing of affective …show more content…
stimuli.
While undergoing blood oxygen level dependent functional magnetic resonance imaging (fMRI), male and female subjects were presented with static photographs of affective faces depicting either fearful or happy expressions alternating with periods of resting visual ®xation. We hypothesized that males perceiving affective faces would show greater lateralization of activation within the amygdala relative to females, and further hypothesized that this activation would differ between the sexes as a function of the valence of the perceived emotion.
MATERIALS AND METHODS
Subjects: Thirteen healthy adult volunteers participated
(seven male, six female). The mean age of the subjects was
24.2 (range 20±31 years). Subjects were right-handed by self-report. All subjects were unremarkable for neurological or psychiatric illness as measured by a structured clinical interview and none were taking psychotropic medications.
Each subject had normal or corrected normal visual acuity.
All conventional MR images were interpreted by a neuroradiologist, and no clinical abnormalities were detected for any study subject. The study was approved by the Institutional Review Board at McLean Hospital and all subjects provided written informed consent.
Imaging methods: Prior to the scan, subjects were briefed about the face perception protocol and were instructed to view each photograph and silently identify the emotion displayed. Scanning was performed on a 1.5 T scanner using a quadrature head coil. A T1-weighted sagittal image wis used to localize a plane perpendicular to the AC-PC line. Twelve to fourteen high-resolution T1 coronal images were acquired for each subject and covered an area from the anterior portion of the frontal pole to the post-central gyrus. T2Ã -weighted echo planar images were acquired in the same coronal plane during the face perception tasks.
Functional images were collected every 3 s using a gradient echo pulse sequence (TE 40 ms, ¯ip angle 758). An image matrix of 64 3 128 was used with a 3 3 3 mm inplane resolution and a 6 mm slice thickness. Each stimulus sequence resulted in a total study length of 150 s. A series of 50 sequential images was obtained during each cognitive task condition.
Face activation paradigms: Stimuli were generated by a
Macintosh computer and were projected with a magnetically shielded LCD video projector onto a translucent screen viewed via a mirror placed above the head in the scanner. Each scan sequence began with a 30 baseline condition and was followed by two alternating task testing-control epochs, resulting in a total run of 150 s for each stimulation paradigm. During baseline and resting control periods, subjects maintained visual ®xation on a small circle located at the center of the projection screen. Six faces, all displaying a similar emotional expression were selected from the stimulus set of Ekman and Friesen [15].
Subjects viewed three face photographs during each 30 s stimulus period (9.5 s on, 0.5 s ISI).
Each subject participated in a happy and a fearful affect perception task. Both tasks were identical in timing and
2544
Vol 12 No 11 8 August 2001
W. D. S. KILLGORE AND D. A. YURGELUN-TODD
design, except that the emotion displayed in the photographs differed for each task. The order of presentation for the happy and fearful conditions was counterbalanced across subjects. To assess participation, subjects were immediately asked to report the affect displayed after the completion of each task, while still within the magnet.
Image processing and Analysis: All images were corrected for in-plane and translational motion using the
DART program [16]. Data for which motion exceeded
1 mm in any direction or 18 of rotation were not included in analyses. Cortical activation was measured within neuroamatomically de®ned regions of interest (ROI), which were selected with reference to an anatomic atlas and placements were made based on gyral boundaries and structural landmarks visible on MR images. Each amygdala
ROI was comprised of four pixels, 3 3 3 mm sampled from one coronal slice (see Fig. 1). Measures of signal intensity were derived by averaging the MR signal in the four pixels of each ROI over the task activation periods. These data were converted to an index of the percentage change in
MR signal relative to each subject 's average signal during the ®rst seven baseline images. The signal intensity data were analyzed using a 2 (sex) 32 (affect valence) 3 2
(laterality) split-plot ANOVA, with amygdala laterality and affect valence condition as a within-subjects factors.
RESULTS
Amygdala activation: We restricted our analyses to changes in BOLD response within two a priori de®ned regions of interest (ROIs) that included the left and right amygdala (Fig. 1). The mean signal intensities for each ROI for both affect conditions are presented in Fig. 2. Males and females differed in the pattern of lateralized activation within the amygdala during the perception of fearful and happy faces. This difference was re¯ected as a signi®cant sex 3 laterality 3 valence interaction (F(1,11) 6.98, p
0.023). Separate analyses were undertaken to examine sexrelated effects for the perception of fearful and happy affect. Fearful affect perception: Males did not differ signi®cantly from females in overall MR signal change when viewing fearful faces and there was no signi®cant interaction between sex and lateralized amygdala activation. As predicted, there was a main effect of laterality, with greater activation in the left relative to the right amygdala while subjects viewed faces expressing fear (F(1,11) 8.30, p
0.015; Fig. 2a). Analysis of simple effects, however, revealed that the lateralized differences in activation favoring the left amygdala were only signi®cant within the sample of males (F(1,6) 6.83, p 0.04). In contrast, the difference between left and light amygdala activation was not signi®cant within the sample of females (F(1,5) 2.15, p 0.203).
Happy affect perception: As illustrated in Fig. 2b, males and females demonstrated noticeably different patterns of lateralized activation of the amygdala while viewing happy faces, as re¯ected in a signi®cant sex 3 laterality interaction
(F(1,11) 10.56, p 0.008), with no signi®cant main effects of sex or laterality. Within-subject analyses of simple effects revealed that males viewing happy faces demonstrated signi®cantly greater activation within the right relative to
NEUROREPORT
AMYGDALA SEX DIFFERENCES
Fig. 1. Example of the placement of regions of interest (ROIs) and representative patterns of amygdala activation in a male (top) and female (bottom) subject during the fearful (left) and happy (right) face perception conditions. For purposes of illustration, areas of activation represent local maxima within the amygdala following placement of a mask to exclude all non-amygdala activity.
the left amygdala (F(1,6) 15.41, p 0.008). Females, in contrast, showed a non-signi®cant trend in the opposite direction (F(1,5) 1.57, p 0.265), re¯ecting essentially no difference in MR signal change between the left and right amygdala during the viewing of happy faces.
Differences in activation asymmetry: In addition to the comparisons of relative changes in the magnitude of MR signal from baseline, we also compared males and females directly on an index of asymmetry of amygdala activation during the affect perception tasks. The asymmetry index was calculated by determining the proportion of the mean
MR signal above or below baseline during the affect perception condition (i.e. mean task/mean baseline) for the right and left amygdala individually and then entering these proportions into the following formula: [(rightÀleft)/
(right left)] 3 100. Thus, a positive value re¯ected an asymmetry of activation within the amygdala that was relatively greater on the right than the left, while a negative
value re¯ected asymmetry of activation favoring the left relative to the right amygdala. As evident in Fig. 3, males demonstrated signi®cant asymmetry of activation within the amygdala, with a left-lateralized asymmetry index for the perception of fearful faces À0.21 Æ 0.21; t(6) À2.61, p 0.04), and a signi®cantly right-lateralized asymmetry index for the perception of happy faces (0.21 Æ 0.14; t(6) 3.92, p 0.008). In contrast, females demonstrated a trend toward left-lateralized activation that did not differ signi®cantly from zero for either affect condition, suggesting that the activation of each amygdala was essentiality bilateral during the perception of fearful (À.010 Æ 0.17; t(5) 1.46, p 0.20) and happy faces (À11 Æ 0.21; t(5) 1.25, p 0.27).
DISCUSSION
These data suggest that males and females differ in the direction and magnitude of lateralized activation within the amygdala during the visual perception of facial affect.
Vol 12 No 11 8 August 2001
2545
NEUROREPORT
Mean percentage signal change
(a)
W. D. S. KILLGORE AND D. A. YURGELUN-TODD
0.3
Male
Female
0.2
0.1
0
Female
Happy
Ϫ0.1
Fear
Ϫ0.2
Ϫ0.3
Ϫ0.3
Left
Ϫ0.2
Ϫ0.1
0
0.1
Asymmetry index
0.2
0.3
Fig. 3. Plot of the mean asymmetry index scores re¯ecting amygdala activation for males and females when viewing happy and fearful faces.
The asymmetry index re¯ects the relative difference in the proportion of right amygdala vs. left amygdala activation [(right Àleft)/(right left)] 3 100. Overall, males showed signi®cant right lateralized activation for happy ( p , 0.01) and left lateralized activation for fearful ( p , 0.05) faces, whereas the asymmetry of amygdala activation within the female sample did not differ signi®cantly from zero for either affect condition.
Right
(b) 0.3
Mean percentage signal change
Male
0.2
0.1
0
Ϫ0.1
Ϫ0.2
Male
Female
Ϫ0.3
Left
Right
Fig. 2. Plots of the mean percent signal intensity change for males and females during the perception of affective faces. Male (blue) and female
(red) groups showed a trend toward increased signal intensity in the left versus the right amygdala when viewing fearful faces, although the difference only reached statistical signi®cance ( p , 0.05) for the males
(a). Males and females showed opposite patterns of lateralized amygdala signal intensity changes in response to happy faces, with males showing signi®cantly greater right than left amygdala activation ( p , 0.01), whereas females showed a non-signi®cant trend toward greater left than right amygdala activation (b).
While perceiving happy and fearful faces, males demonstrated signi®cant asymmetry of activation within the amygdala, whereas females failed to show signi®cant differences between the activation of the left and the right amygdala, regardless of the affective valence of the expressions. These ®ndings are consistent with earlier research suggesting that males tend to show greater lateralization of function within the cerebral cortex for cognitive [17] and affective processes [9,10], but further extend the scope of these sex differences in lateralization to include subcortical structures within the limbic system involved in affective processing. These ®ndings are important in that they suggest that sex differences in cerebral lateralization are not limited to cortical regions but are evident even in phylogenetically older structures of the brain. Further, our results suggest that affective processing, particularly for
2546
Vol 12 No 11 8 August 2001
positive emotions, may differ in a fundamental way between men and women, which may have important implications for understanding the way that the sexes differ in their social and emotional behaviors.
Previous neuroimaging studies have demonstrated reliable activation of the left amygdala during the perception of facial displays of fearful affect [12,18,19]. Our ®ndings replicate previous functional imaging studies of fear perception, with our male and female groups showing greater activation within the left amygdala during the perception of fearful faces, although the difference only reached statistical signi®cance for the males. Greater activation within the left amygdala during fear perception is consistent with its purported role in the acquisition of fear responses and aversive conditioning [20]. Furthermore, lesions restricted to the amygdala are known to reduce emotional responsiveness and impair the ability to recognize fear within facial expressions [21], and increased cerebral blood ¯ow to the left amygdala has been reported during the induction of fear with intravenous injections of procaine [22]. While the left amygdala appears to be particularly important for the processing of external stimuli related to withdrawal behaviors, our data suggest that this specialized function of the left amygdala is most prominent and clearly observed in males.
In contrast to the growing research literature on amygdala activation during fearful affect perception, few investigators have examined the functional responsiveness of the amygdala to the perception of happy faces, and none have systematically explored the potential in¯uence of sex differences on these responses. We found that males showed signi®cant asymmetry of activation of the amygdala when presented with happy faces, demonstrating increased MR signal within the right amygdala and slightly reduced activation in the left. In contrast, females showed the opposite pattern of activation, demonstrating a nonsigni®cant trend toward greater activation of the left relative to the right amygdala. The laterality data for the
NEUROREPORT
AMYGDALA SEX DIFFERENCES
males are consistent with ®ndings reported by Morris and colleagues using PET [23], who reported decreases in left amygdala activation during the perception of happy faces, although they found no signi®cant change in the right amygdala. We present the ®rst fMRI data speci®cally comparing the asymmetry of amygdala responses of males and females during the perception of both happy and fearful faces in a free-viewing presentation format.
For some cognitive domains, such as language and visuospatial processing, males appear to show greater cortical lateralization than females [6,7], and it appears that the same pattern of lateralized sex differences may also be evident for facial perception [9,10]. Support for sex dependent cortical lateralization has also been demonstrated using PET technology, with one study showing leftward lateralization of glucose metabolic activity within the inferior frontal and orbitofrontal cortices of men during recall of sad memories, whereas women showed bilateral activation [24]. Our current ®ndings suggest that the lateralization of affective processing seen in cortical regions may be extended to subcortical systems, as proposed recently by other investigators [25].
The present ®ndings further suggest that the asymmetry of amygdala activation may follow a pattern of sex dependent lateralization analogous to that seen in higher regions of the cerebral cortex, with males typically showing greater lateralized function relative to females, who tend to show a more bilateral pattern of function.
The present study represents the ®rst attempt to use fMRI to characterize sex differences in amygdala responsiveness to faces expressing valence speci®c affect. Obviously, the limited number of males and females in each group temper these results and additional studies will be needed with larger samples in order to establish the generalizability of these ®ndings. Furthermore, in the present study, we employed a simple visual ®xation point as the control baseline condition rather than the use of a neutral face. While this strategy maximized our possibility of detecting signal changes for each valence condition and permitted the direct comparison of fearful and happy activation within the ROIs, we were not able to make the tighter task comparison that could be accomplished
by subtracting the level of activation produced by neutral faces from the activation produced by the affective faces.
As a preliminary investigation, the present strategy provided us with the greatest power for detecting sex and valence related signal changes. While it is unlikely that the use of a tighter control task using neutral faces would substantially in¯uence the direction of lateralized ®ndings, such a study is obviously the next logical step for future research. With the aforementioned points in mind, we believe that the present results provide compelling data and the ®rst fMRI evidence demonstrating sex differences
in the limbic processing of effectively valenced facial stimuli. CONCLUSION
We found that males and females demonstrate signi®cantly different patterns of lateralized activation within the amygdala during the perception of happy and fearful faces.
Consistent with previous studies that have shown greater lateralization of cortical processing within males, we found that subcortical activation during affect processing was also signi®cantly more lateralized in males than females, and these differences were most apparent for the processing of happy expressions. Such ®ndings suggest that sex differences should be considered as a critical factor in functional neuroimaging studies of affect. Furthermore, these ®ndings raise the possibility that some of the differences in the manner in which men and women interpret and react to social and emotional information may be due, at least in part, to sexually divergent patterns of limbic responsiveness that may be neurobiologically rather than socially determined.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
Kring AM and Gordon AH. J Pers Soc Psychol 74, 686±703 (1998).
Grossman M and Wood W. J Pers Soc Psychol 65, 1010±1022 (1993).
Orozco S and Ehlers CL. Biol Psychiatry 44, 281±289 (1998).
Hall JA. Nonverbal Sex Differences, Communication Accuracy and Expressive
Style. Baltimore: Johns Hopkins University Press; 1984.
Kessler RC, McGonagle KA, Zhao S et al. Arch Gen Psychiatry 51, 8±19
(1994).
McGlone J. Cortex 14, 122±128 (1978).
Halpern DF. Sex Differences in Cognitive Abilities. Hillsdale, NJ: Lawrence
Erlbaum Associates, 1992.
Burton LA and Levy J. Brain Cogn 11, 210±228 (189).
Killgore WD and Gangestad SW. Percept Mot Skills 89, 311±314 (1999).
Killgore WD. Percept Motor Skills 91, 525±530 (2000).
Davidson RJ. Am Psychol 55, 1196±1214 (2000).
Breiter HC, Etcoff NL, Whalen PJ et al. Neuron 17, 875±887 (1996).
Killgore WDS, Oki M and Yurgelun-Todd DA. Neuroreport 12, 427±433
(2001).
Whalen PJ, Rauch SL, Etcoff NL et al. J Neurosci 18, 411±418 (1998).
Ekman P and Friesen WV. Pictures of Facial Affect. Palo Alto, CA:
Consulting Psychologists Press; 1976.
Maas LC, Frederick BD and Renshaw PF. Magn Reson Med 37, 131±139
(1997).
McGlone J. Behav Brain Sci 3, 215±263 (1980).
Phillips ML, Young AW, Scott SK et al. Proc R Soc Lond B Biol Sci 265,
1809±1817 (1998).
Wright CI, Fischer H, Whalen PJ et al. Neuroreport 12, 379±383 (2001).
LeDoux J. Biol Psychiatry 44, 1229±1238 (1998).
Adolphs R, Tranel D, Damasio H and Damasio A. Nature 372, 669±672
(1994).
Ketter TA, Andreason PJ, George MS et al. Arch Gen Psychiatry 53,
59±69 (1996).
Morris JS, Friston KJ, Buchel C, et al. Brain 121, 47±57 (1998).
Pardo JV, Pardo PJ, Raichle ME. Am J Psychiatry 150, 713±719 (1993).
Schneider F, Habel U, Kessler C et al. Hum Brain Mapp 9, 226±238
(2000).
Vol 12 No 11 8 August 2001
2547
MOTIVATION, EMOTION, FEEDING, DRINKING
Sex differences in amygdala activation during the perception of facial affect
William D. S. KillgoreCA and Deborah A. Yurgelun-Todd
Cognitive Neuroimaging Laboratory, Brain Imaging Center, McLean Hospital/Harvard Medical School, 115 Mill Street, Belmont,
MA 02478, USA
CA
Corresponding Author
Received 11 April 2001; accepted 5 June 2001
The cognitive and affective systems of the cerebral cortex are often more lateralized in males than females, but it is unclear whether these differences extend to subcortical systems. We used fMRI to examine sex differences in lateralized amygdala activity during happy and fearful face perception. Amygdala activation differed for men …show more content…
and women depending on the valence of the expression. Overall, males were more lateralized than females, but the direction differed between valence
conditions. Happy faces produced greater right than left amygdala activation for males but not females. Both sexes showed greater left amygdala activation for fearful faces. These
®ndings suggest that the lateralization of affective function may extend beyond the cortex to subcortical regions such as the amygdala. NeuroReport 12:2543±2547 & 2001 Lippincott Williams & Wilkins.
Key words: Affect; Amygdala; Cerebral laterality; Emotion; Faces; Fear; fMRI; Functional neuroimaging; Happy; Sex differences; Visual perception
INTRODUCTION
Men and women appear to differ in the manner in which they perceive, process, express, and experience emotion.
On the whole, females tend to be more emotionally expressive than males and the two sexes show differences in their responsiveness to affectively charged stimuli [1].
Females tend to rate their emotions more intensely [2], show stronger evoked potential responses to emotional faces [3] and demonstrate greater facility at decoding nonverbal messages than males [4]. Unfortunately, women also have a rate of depression that is estimated to be 2±3 times higher than that of men [5]. Although the existence of emotional processing differences between the sexes is becoming better recognized, the factors that produce these differences are still not well understood. While many of the differences between men and women in the processing of emotion may be attributable to social factors and learned patterns of behavior [2], it is likely that emotional differences also re¯ect genotypic differences in the sexual dimorphism of the nervous system.
One of the most obvious manifestations of sex differences in the structure and function of the nervous system is exempli®ed by the ®nding that men and women show different patterns of cerebral lateralization of function [6].
Sex differences in cerebral laterality have been documented in the neuropsychological literature for decades [6]. Across a number of studies, a consistent pattern of sex-speci®c lateralization of function has emerged, with males typically demonstrating greater lateralization of language and visuospatial abilities, whereas females tend to be less lateralized, showing a greater frequency of bilateral representation of verbal abilities across the two hemi-
0959-4965 & Lippincott Williams & Wilkins
spheres [7]. The sexually dimorphic structure of the nervous system appears to produce consistent asymmetries in affective function, such as the lateralized perception of the affective features of facial expressions of emotion [8±10], and recent studies have suggested that males may also demonstrate a greater extent of lateralization of emotional perception than females [9,10].
At the cortical level, emotional processing is associated with consistent patterns of lateralized activity as measured by EEG methods, with positive affect or approach-related emotions associated with relatively greater left than right prefrontal activation, and negative affect or withdrawalrelated emotions associated with the opposite pattern [11].
While lateralization of cognitive abilities is often observed within the neocortex, the extent of functional lateralization within subcortical structures is uncertain.
There is some evidence of a lateralized pattern of amygdala activity during affective processing, with several studies ®nding greater left than right amygdala activation during negative emotional facial expressions such as fear [12], and sex differences in this lateralized activity may be evident as early as adolescence [13]. In contrast to negative emotions, there have been few studies examining the functional responsiveness of the amygdala to positive emotions such as happiness. The laterality of amygdala activation during the perception of happy faces has been less consistent than the ®ndings for fearful faces [12,14] and no functional imaging studies have speci®cally examined the in¯uence of sex differences on the activation of the amygdala during the perception of either happy or fearful faces. We chose to examine whether the theory of sex-dependent lateralization of function that accounts for ®ndings within the neo-cortex
Vol 12 No 11 8 August 2001
2543
NEUROREPORT could be extended to also account for subcortical processing of affective …show more content…
stimuli.
While undergoing blood oxygen level dependent functional magnetic resonance imaging (fMRI), male and female subjects were presented with static photographs of affective faces depicting either fearful or happy expressions alternating with periods of resting visual ®xation. We hypothesized that males perceiving affective faces would show greater lateralization of activation within the amygdala relative to females, and further hypothesized that this activation would differ between the sexes as a function of the valence of the perceived emotion.
MATERIALS AND METHODS
Subjects: Thirteen healthy adult volunteers participated
(seven male, six female). The mean age of the subjects was
24.2 (range 20±31 years). Subjects were right-handed by self-report. All subjects were unremarkable for neurological or psychiatric illness as measured by a structured clinical interview and none were taking psychotropic medications.
Each subject had normal or corrected normal visual acuity.
All conventional MR images were interpreted by a neuroradiologist, and no clinical abnormalities were detected for any study subject. The study was approved by the Institutional Review Board at McLean Hospital and all subjects provided written informed consent.
Imaging methods: Prior to the scan, subjects were briefed about the face perception protocol and were instructed to view each photograph and silently identify the emotion displayed. Scanning was performed on a 1.5 T scanner using a quadrature head coil. A T1-weighted sagittal image wis used to localize a plane perpendicular to the AC-PC line. Twelve to fourteen high-resolution T1 coronal images were acquired for each subject and covered an area from the anterior portion of the frontal pole to the post-central gyrus. T2Ã -weighted echo planar images were acquired in the same coronal plane during the face perception tasks.
Functional images were collected every 3 s using a gradient echo pulse sequence (TE 40 ms, ¯ip angle 758). An image matrix of 64 3 128 was used with a 3 3 3 mm inplane resolution and a 6 mm slice thickness. Each stimulus sequence resulted in a total study length of 150 s. A series of 50 sequential images was obtained during each cognitive task condition.
Face activation paradigms: Stimuli were generated by a
Macintosh computer and were projected with a magnetically shielded LCD video projector onto a translucent screen viewed via a mirror placed above the head in the scanner. Each scan sequence began with a 30 baseline condition and was followed by two alternating task testing-control epochs, resulting in a total run of 150 s for each stimulation paradigm. During baseline and resting control periods, subjects maintained visual ®xation on a small circle located at the center of the projection screen. Six faces, all displaying a similar emotional expression were selected from the stimulus set of Ekman and Friesen [15].
Subjects viewed three face photographs during each 30 s stimulus period (9.5 s on, 0.5 s ISI).
Each subject participated in a happy and a fearful affect perception task. Both tasks were identical in timing and
2544
Vol 12 No 11 8 August 2001
W. D. S. KILLGORE AND D. A. YURGELUN-TODD
design, except that the emotion displayed in the photographs differed for each task. The order of presentation for the happy and fearful conditions was counterbalanced across subjects. To assess participation, subjects were immediately asked to report the affect displayed after the completion of each task, while still within the magnet.
Image processing and Analysis: All images were corrected for in-plane and translational motion using the
DART program [16]. Data for which motion exceeded
1 mm in any direction or 18 of rotation were not included in analyses. Cortical activation was measured within neuroamatomically de®ned regions of interest (ROI), which were selected with reference to an anatomic atlas and placements were made based on gyral boundaries and structural landmarks visible on MR images. Each amygdala
ROI was comprised of four pixels, 3 3 3 mm sampled from one coronal slice (see Fig. 1). Measures of signal intensity were derived by averaging the MR signal in the four pixels of each ROI over the task activation periods. These data were converted to an index of the percentage change in
MR signal relative to each subject 's average signal during the ®rst seven baseline images. The signal intensity data were analyzed using a 2 (sex) 32 (affect valence) 3 2
(laterality) split-plot ANOVA, with amygdala laterality and affect valence condition as a within-subjects factors.
RESULTS
Amygdala activation: We restricted our analyses to changes in BOLD response within two a priori de®ned regions of interest (ROIs) that included the left and right amygdala (Fig. 1). The mean signal intensities for each ROI for both affect conditions are presented in Fig. 2. Males and females differed in the pattern of lateralized activation within the amygdala during the perception of fearful and happy faces. This difference was re¯ected as a signi®cant sex 3 laterality 3 valence interaction (F(1,11) 6.98, p
0.023). Separate analyses were undertaken to examine sexrelated effects for the perception of fearful and happy affect. Fearful affect perception: Males did not differ signi®cantly from females in overall MR signal change when viewing fearful faces and there was no signi®cant interaction between sex and lateralized amygdala activation. As predicted, there was a main effect of laterality, with greater activation in the left relative to the right amygdala while subjects viewed faces expressing fear (F(1,11) 8.30, p
0.015; Fig. 2a). Analysis of simple effects, however, revealed that the lateralized differences in activation favoring the left amygdala were only signi®cant within the sample of males (F(1,6) 6.83, p 0.04). In contrast, the difference between left and light amygdala activation was not signi®cant within the sample of females (F(1,5) 2.15, p 0.203).
Happy affect perception: As illustrated in Fig. 2b, males and females demonstrated noticeably different patterns of lateralized activation of the amygdala while viewing happy faces, as re¯ected in a signi®cant sex 3 laterality interaction
(F(1,11) 10.56, p 0.008), with no signi®cant main effects of sex or laterality. Within-subject analyses of simple effects revealed that males viewing happy faces demonstrated signi®cantly greater activation within the right relative to
NEUROREPORT
AMYGDALA SEX DIFFERENCES
Fig. 1. Example of the placement of regions of interest (ROIs) and representative patterns of amygdala activation in a male (top) and female (bottom) subject during the fearful (left) and happy (right) face perception conditions. For purposes of illustration, areas of activation represent local maxima within the amygdala following placement of a mask to exclude all non-amygdala activity.
the left amygdala (F(1,6) 15.41, p 0.008). Females, in contrast, showed a non-signi®cant trend in the opposite direction (F(1,5) 1.57, p 0.265), re¯ecting essentially no difference in MR signal change between the left and right amygdala during the viewing of happy faces.
Differences in activation asymmetry: In addition to the comparisons of relative changes in the magnitude of MR signal from baseline, we also compared males and females directly on an index of asymmetry of amygdala activation during the affect perception tasks. The asymmetry index was calculated by determining the proportion of the mean
MR signal above or below baseline during the affect perception condition (i.e. mean task/mean baseline) for the right and left amygdala individually and then entering these proportions into the following formula: [(rightÀleft)/
(right left)] 3 100. Thus, a positive value re¯ected an asymmetry of activation within the amygdala that was relatively greater on the right than the left, while a negative
value re¯ected asymmetry of activation favoring the left relative to the right amygdala. As evident in Fig. 3, males demonstrated signi®cant asymmetry of activation within the amygdala, with a left-lateralized asymmetry index for the perception of fearful faces À0.21 Æ 0.21; t(6) À2.61, p 0.04), and a signi®cantly right-lateralized asymmetry index for the perception of happy faces (0.21 Æ 0.14; t(6) 3.92, p 0.008). In contrast, females demonstrated a trend toward left-lateralized activation that did not differ signi®cantly from zero for either affect condition, suggesting that the activation of each amygdala was essentiality bilateral during the perception of fearful (À.010 Æ 0.17; t(5) 1.46, p 0.20) and happy faces (À11 Æ 0.21; t(5) 1.25, p 0.27).
DISCUSSION
These data suggest that males and females differ in the direction and magnitude of lateralized activation within the amygdala during the visual perception of facial affect.
Vol 12 No 11 8 August 2001
2545
NEUROREPORT
Mean percentage signal change
(a)
W. D. S. KILLGORE AND D. A. YURGELUN-TODD
0.3
Male
Female
0.2
0.1
0
Female
Happy
Ϫ0.1
Fear
Ϫ0.2
Ϫ0.3
Ϫ0.3
Left
Ϫ0.2
Ϫ0.1
0
0.1
Asymmetry index
0.2
0.3
Fig. 3. Plot of the mean asymmetry index scores re¯ecting amygdala activation for males and females when viewing happy and fearful faces.
The asymmetry index re¯ects the relative difference in the proportion of right amygdala vs. left amygdala activation [(right Àleft)/(right left)] 3 100. Overall, males showed signi®cant right lateralized activation for happy ( p , 0.01) and left lateralized activation for fearful ( p , 0.05) faces, whereas the asymmetry of amygdala activation within the female sample did not differ signi®cantly from zero for either affect condition.
Right
(b) 0.3
Mean percentage signal change
Male
0.2
0.1
0
Ϫ0.1
Ϫ0.2
Male
Female
Ϫ0.3
Left
Right
Fig. 2. Plots of the mean percent signal intensity change for males and females during the perception of affective faces. Male (blue) and female
(red) groups showed a trend toward increased signal intensity in the left versus the right amygdala when viewing fearful faces, although the difference only reached statistical signi®cance ( p , 0.05) for the males
(a). Males and females showed opposite patterns of lateralized amygdala signal intensity changes in response to happy faces, with males showing signi®cantly greater right than left amygdala activation ( p , 0.01), whereas females showed a non-signi®cant trend toward greater left than right amygdala activation (b).
While perceiving happy and fearful faces, males demonstrated signi®cant asymmetry of activation within the amygdala, whereas females failed to show signi®cant differences between the activation of the left and the right amygdala, regardless of the affective valence of the expressions. These ®ndings are consistent with earlier research suggesting that males tend to show greater lateralization of function within the cerebral cortex for cognitive [17] and affective processes [9,10], but further extend the scope of these sex differences in lateralization to include subcortical structures within the limbic system involved in affective processing. These ®ndings are important in that they suggest that sex differences in cerebral lateralization are not limited to cortical regions but are evident even in phylogenetically older structures of the brain. Further, our results suggest that affective processing, particularly for
2546
Vol 12 No 11 8 August 2001
positive emotions, may differ in a fundamental way between men and women, which may have important implications for understanding the way that the sexes differ in their social and emotional behaviors.
Previous neuroimaging studies have demonstrated reliable activation of the left amygdala during the perception of facial displays of fearful affect [12,18,19]. Our ®ndings replicate previous functional imaging studies of fear perception, with our male and female groups showing greater activation within the left amygdala during the perception of fearful faces, although the difference only reached statistical signi®cance for the males. Greater activation within the left amygdala during fear perception is consistent with its purported role in the acquisition of fear responses and aversive conditioning [20]. Furthermore, lesions restricted to the amygdala are known to reduce emotional responsiveness and impair the ability to recognize fear within facial expressions [21], and increased cerebral blood ¯ow to the left amygdala has been reported during the induction of fear with intravenous injections of procaine [22]. While the left amygdala appears to be particularly important for the processing of external stimuli related to withdrawal behaviors, our data suggest that this specialized function of the left amygdala is most prominent and clearly observed in males.
In contrast to the growing research literature on amygdala activation during fearful affect perception, few investigators have examined the functional responsiveness of the amygdala to the perception of happy faces, and none have systematically explored the potential in¯uence of sex differences on these responses. We found that males showed signi®cant asymmetry of activation of the amygdala when presented with happy faces, demonstrating increased MR signal within the right amygdala and slightly reduced activation in the left. In contrast, females showed the opposite pattern of activation, demonstrating a nonsigni®cant trend toward greater activation of the left relative to the right amygdala. The laterality data for the
NEUROREPORT
AMYGDALA SEX DIFFERENCES
males are consistent with ®ndings reported by Morris and colleagues using PET [23], who reported decreases in left amygdala activation during the perception of happy faces, although they found no signi®cant change in the right amygdala. We present the ®rst fMRI data speci®cally comparing the asymmetry of amygdala responses of males and females during the perception of both happy and fearful faces in a free-viewing presentation format.
For some cognitive domains, such as language and visuospatial processing, males appear to show greater cortical lateralization than females [6,7], and it appears that the same pattern of lateralized sex differences may also be evident for facial perception [9,10]. Support for sex dependent cortical lateralization has also been demonstrated using PET technology, with one study showing leftward lateralization of glucose metabolic activity within the inferior frontal and orbitofrontal cortices of men during recall of sad memories, whereas women showed bilateral activation [24]. Our current ®ndings suggest that the lateralization of affective processing seen in cortical regions may be extended to subcortical systems, as proposed recently by other investigators [25].
The present ®ndings further suggest that the asymmetry of amygdala activation may follow a pattern of sex dependent lateralization analogous to that seen in higher regions of the cerebral cortex, with males typically showing greater lateralized function relative to females, who tend to show a more bilateral pattern of function.
The present study represents the ®rst attempt to use fMRI to characterize sex differences in amygdala responsiveness to faces expressing valence speci®c affect. Obviously, the limited number of males and females in each group temper these results and additional studies will be needed with larger samples in order to establish the generalizability of these ®ndings. Furthermore, in the present study, we employed a simple visual ®xation point as the control baseline condition rather than the use of a neutral face. While this strategy maximized our possibility of detecting signal changes for each valence condition and permitted the direct comparison of fearful and happy activation within the ROIs, we were not able to make the tighter task comparison that could be accomplished
by subtracting the level of activation produced by neutral faces from the activation produced by the affective faces.
As a preliminary investigation, the present strategy provided us with the greatest power for detecting sex and valence related signal changes. While it is unlikely that the use of a tighter control task using neutral faces would substantially in¯uence the direction of lateralized ®ndings, such a study is obviously the next logical step for future research. With the aforementioned points in mind, we believe that the present results provide compelling data and the ®rst fMRI evidence demonstrating sex differences
in the limbic processing of effectively valenced facial stimuli. CONCLUSION
We found that males and females demonstrate signi®cantly different patterns of lateralized activation within the amygdala during the perception of happy and fearful faces.
Consistent with previous studies that have shown greater lateralization of cortical processing within males, we found that subcortical activation during affect processing was also signi®cantly more lateralized in males than females, and these differences were most apparent for the processing of happy expressions. Such ®ndings suggest that sex differences should be considered as a critical factor in functional neuroimaging studies of affect. Furthermore, these ®ndings raise the possibility that some of the differences in the manner in which men and women interpret and react to social and emotional information may be due, at least in part, to sexually divergent patterns of limbic responsiveness that may be neurobiologically rather than socially determined.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
Kring AM and Gordon AH. J Pers Soc Psychol 74, 686±703 (1998).
Grossman M and Wood W. J Pers Soc Psychol 65, 1010±1022 (1993).
Orozco S and Ehlers CL. Biol Psychiatry 44, 281±289 (1998).
Hall JA. Nonverbal Sex Differences, Communication Accuracy and Expressive
Style. Baltimore: Johns Hopkins University Press; 1984.
Kessler RC, McGonagle KA, Zhao S et al. Arch Gen Psychiatry 51, 8±19
(1994).
McGlone J. Cortex 14, 122±128 (1978).
Halpern DF. Sex Differences in Cognitive Abilities. Hillsdale, NJ: Lawrence
Erlbaum Associates, 1992.
Burton LA and Levy J. Brain Cogn 11, 210±228 (189).
Killgore WD and Gangestad SW. Percept Mot Skills 89, 311±314 (1999).
Killgore WD. Percept Motor Skills 91, 525±530 (2000).
Davidson RJ. Am Psychol 55, 1196±1214 (2000).
Breiter HC, Etcoff NL, Whalen PJ et al. Neuron 17, 875±887 (1996).
Killgore WDS, Oki M and Yurgelun-Todd DA. Neuroreport 12, 427±433
(2001).
Whalen PJ, Rauch SL, Etcoff NL et al. J Neurosci 18, 411±418 (1998).
Ekman P and Friesen WV. Pictures of Facial Affect. Palo Alto, CA:
Consulting Psychologists Press; 1976.
Maas LC, Frederick BD and Renshaw PF. Magn Reson Med 37, 131±139
(1997).
McGlone J. Behav Brain Sci 3, 215±263 (1980).
Phillips ML, Young AW, Scott SK et al. Proc R Soc Lond B Biol Sci 265,
1809±1817 (1998).
Wright CI, Fischer H, Whalen PJ et al. Neuroreport 12, 379±383 (2001).
LeDoux J. Biol Psychiatry 44, 1229±1238 (1998).
Adolphs R, Tranel D, Damasio H and Damasio A. Nature 372, 669±672
(1994).
Ketter TA, Andreason PJ, George MS et al. Arch Gen Psychiatry 53,
59±69 (1996).
Morris JS, Friston KJ, Buchel C, et al. Brain 121, 47±57 (1998).
Pardo JV, Pardo PJ, Raichle ME. Am J Psychiatry 150, 713±719 (1993).
Schneider F, Habel U, Kessler C et al. Hum Brain Mapp 9, 226±238
(2000).
Vol 12 No 11 8 August 2001
2547