Results
Experiment 1—FEPT Performance
Performance on the FEPT in Experiment 1 (n = 75 females, n = 63 males) revealed that females were significantly more accurate for fear stimuli, relative to males, t = 2.37, P < 0.05, Cohen's d = 0.40, but females and males performed equivalently for anger, happy and sad stimuli (all ps > 0.17). Table 1 provides the descriptive statistics, t statistics and effect sizes for the comparisons. d′ was also higher in females for fear stimuli, indicating greater sensitivity to fear, relative to males, t = −2.53, P = 0.01, but not for anger, happy or sad stimuli (ps > 0.20). For the smaller sample (part of Experiment 2), in-scanner behavioral results were not significantly different between the two groups for accuracy or d′ for any of the emotions (though were in the same direction and of similar effect size as observed in the larger sample for fear). Please refer to Table 2 for descriptive statistics, t statistics and effect sizes for the comparisons.
Experiment 2—fMRI Areas of Activation for Females and Males
Table 3 and Figure 1 display areas of significant activation assessed for females and males separately. For the anger-neutral contrast, females activated far fewer areas than males, in general, with lesser extent of activation (females = 6408 mm, males = 15 360 mm). Two of four regions activated among females were exclusive to females. For the fear-neutral contrast, both females and males activated frontal and parietal regions, though the majority of these regions were exclusive to females or males. Females demonstrated greater extent of activation than males (females = 19 728 mm, males = 11 160 mm). Females additionally activated right insula and superior temporal gyrus, as well as cerebellum (declive), while males additionally activated right parahippocampal gyrus. For the happy-neutral contrast, females activated only right precuneus/cuneus and left caudate body. Males displayed much more widespread activation for the happy-neutral contrast, including frontal, temporal, parietal, occipital and subcortical regions (females = 6432 mm, males = 45 304 mm), with the majority of regions being exclusive to males. For the sad-neutral contrast, females activated only right posterior cingulate, while males activated bilateral mid-cingulate, left middle frontal gyrus, precuneus, insula and caudate. Again, males displayed more widespread activation than did females (females = 832 mm, males = 4160 mm).
(Enlarge Image)
Figure 1.
Areas of activation for females and males. Panels (A–D) illustrate statistically significant activation for anger in areas that are exclusive to females (red), areas that are exclusive to males (green) and in regions inclusive to females and males (blue). Panels (E–H) illustrate statistically significant activation for fear in areas that are exclusive to females (red), areas that are exclusive to males (green) and in regions inclusive to females and males (blue). Panels (I–L) illustrate statistically significant activation for happy in areas that are exclusive to females (red), areas that are exclusive to males (green) and in regions inclusive to females and males (blue). Panels (M–P) depict statistically significant areas for sad in areas exclusive to females (red), areas that are exclusive to males (green) and in regions inclusive to females and males (blue). Panels A, E, I and M are located at coordinates −4 +10 +32. Panels B, F and J are located at coordinates +8, +7 +32. Panel N is located at coordinates 0 −18 13. Panels C, G, K and O are located at coordinates +62 −24 +18. Panels D, H, L and P are located at coordinates −42 −36 +58.
For the amygdala ROI analyses, males demonstrated nominally greater lateralization to the left for fear and anger, but bilateral activation for happy. Males also demonstrated indiscriminate levels of activation for positive and negative emotions, whereas females activated amygdala to a greater extent in response to anger and fear, relative to happy. Activation of amygdala for sad was not significant for females or males (see Table 3 and Figure 2). To specifically test whether there is an interaction of gender × emotion × hemisphere, we ran a repeated measures ANOVA. For anger, there were two regions of activation in each hemisphere, so we created an average for each hemisphere, weighted by extent of activation. Here we observed a significant interaction of gender × emotion, F = 6.03, P < 0.05, primarily driven by happy, but the interaction of gender × emotion × hemisphere was not significant (P = 0.15).
(Enlarge Image)
Figure 2.
Emotion × hemisphere × gender amygdala activation. Weighted averages (weighted by extent of activation) of right and left amygdala activation for anger, fear and happy in females and males.
Sex Differences in Emotion Activation
The independent sex effect was significant (F = 11.29, P < 0.05 corrected). Activation was significantly different between females and males in a wide range of frontal, temporal, parietal, occipital and subcortical regions, in addition to cerebellum and midbrain (see Table 4).
Post hoc t-tests were conducted in SPM5 to assess the direction of activation differences for females and males for each emotion-neutral contrast. No area of activation was significantly greater in females than in males for any of the four emotions (i.e. happy, fear, anger, sad). Several areas for each emotion were significantly more active in males than in females, however. For anger, males showed significantly greater activation in a large region encompassing right amygdala/hippocampus/medial geniculum body of the thalamus and right mid-cingulate and fusiform gyri, left middle frontal gyrus and a number of regions in bilateral cerebellum. For fear, males demonstrated greater activation relative to females in right postcentral gyrus. For happy, males demonstrated greater activation than females in left superior and middle frontal and middle temporal gyri, left precuneus and right cerebellum (culmen); the sex difference for happy was especially striking. For sad, males displayed greater activation than females in left posterior cingulate and caudate body. Table 5 and Figure 3 display specific areas of activation for each emotion activated more in males, relative to females. For ROI analyses of the amygdala, males exhibited greater left amygdala activation, relative to females, for anger and happy and greater right amygdala activation for sad (see Table 5).
(Enlarge Image)
Figure 3.
Effect of sex on emotion activation. Areas of activation that were greater for males, relative to females are depicted. Anger is displayed in magenta, fear in green, happy in blue and sad in red.
In a post hoc manner, we also sought to understand whether there are more specific differences in activation during identification of facial emotions between females and males, when activation of specific emotions is compared with one another. In prior (unpublished) analyses and in the broader literature, we have observed that activation for each of the negative emotions is quite similar to one another when directly compared (relative to neutral). Therefore, before modeling each emotion separately in the overall model, we compared each of the negative emotions to one another in the entire group of females and males. There was just one region that was significantly different in any combination of comparing the negative emotions to one another (fear-neutral vs anger-neutral), and this was a small cluster in the prefrontal cortex (34 −16 66, mm = 248). Thus, we combined all negative emotions to include in All Negative-Happy and Happy-All Negative contrasts. There were a number of regions that were significantly different using these contrasts in the whole group (reported in Table 6). We then tested the interaction of gender by emotion for regions that were significant in the prior analyses, and here, there were significant interactions detected in two regions of the culmen of the cerebellum, as well as left lingual gyrus, left insula and left middle frontal gyrus. These regions are denoted by asterisks in Table 6. Follow-up t-tests demonstrated that for happy (relative to all negative emotions), males activated one region of the cerebellum culmen (12 −38 −6) significantly more than females [for whom this region was actually deactivated, t = 2.7, P = 0.01]. For All Negative (relative to happy), males deactivated lingual gyrus significantly more so than females, t = −2.2, P = 0.04. Males also deactivated one region of the cerebellum culmen (24 −54 −16) significantly more than did females, t = −2.4, P = 0.02.
Relationship of Performance to Regions with Significant Sex Effects
Finally, Pearson product-moment correlations were conducted to investigate associations between FEPT performance (i.e. accuracy) and extracted fMRI activation in areas that were significantly different between females and males. This set of analyses was exploratory and descriptive; therefore, no correction for multiple comparisons was employed. In males, there were no significant relationships between extracted mean BOLD signal for any of the four emotions and performance. In females, only activation in left precuneus (happy extracted mean from Table 3; 2, −50, 58) was related to accuracy for happy faces (r = 0.48, P < 0.05).