Sex differences in semantic language processing: A functional MRI study

Abstract

Predictions based on two models of sex differences in cerebral organization of language were compared by examining fMRI patterns of 10 females and 9 males during a semantic processing task. Both groups displayed activation of left inferior frontal gyrus (IFG), left superior temporal gyrus (STG), and cingulate. Females, but not males, showed bilateral IFG and STG activation. Further analyses revealed females had less diffuse left activation and greater right posterior temporal and insula region activation than males. Results support both an interhemispheric and an intrahemispheric model of sex differences in language, suggesting that the models may not be mutually exclusive.

Introduction

While there is some debate as to whether sex differences in language exist, a small but significant behavioral sex difference favoring females has been observed for some types of language-based abilities (see Hampson & Kimura, 1992 for review). Two models of differences in neural representation of language, both based on language deficits in males and females following stroke, have been proposed to explain the differences observed on behavioral tasks. McGlone (1977) suggested that females have greater bihemispheric representation of language than males because aphasia was more common in males with left hemisphere damage (48% in males vs. 13% in females). Furthermore, she found that for females only, language deficits were observed after both right and left hemisphere damage. Alternately, Kimura (1983) proposed that males have diffuse left hemisphere representation of language while females have more focal anterior left hemisphere language representation, an explanation suggesting intrahemispheric, rather than interhemispheric, sex differences in language representation. In a group of over 200 stroke patients, there were no sex differences in aphasia rates after right hemisphere damage (2% males, 1% females); however, the majority of the females with aphasia (80%) had anterior left hemisphere lesions compared to males who showed no relationship between left hemisphere lesion location and aphasia. The current study was designed to determine whether functional magnetic resonance (fMRI) activation patterns of males and females during a language task are different, and if so, whether the observed patterns support either of these models.

Functional magnetic resonance imaging provides a noninvasive opportunity to examine potential differences between males and females in brain activation patterns during language tasks. To date, the few published studies in this area have yielded conflicting results. Several studies have found greater bihemispheric activation during language tasks in females compared to males, supporting the “interhemispheric” model proposed by McGlone. For example, Shaywitz et al. (1995) reported significant sex differences in activation patterns associated with phonological processing. Participants determined if visually-presented words matched on a phonological basis. Females showed bilateral inferior frontal activation, but males showed only left inferior frontal activation. Similar findings of lateralized sex differences in anterior regions have been reported by Pugh et al. (1996). Schlosser et al. (1998) reported primarily left frontal activation for both males and females during a silent verbal fluency task; however, a small but significant region of activation in the right orbitofrontal region was observed only in the group of females. Evidence for sex differences in posterior language areas was found by Kansaku, Yamaura, and Kitazawa (2000). Their task was designed to evaluate higher order language processing by having participants listen to short stories vs. hearing narratives in reverse. Females showed bilateral middle temporal gyrus activation, while activation was primarily in the left superior and middle temporal region for males. Jaeger et al. (1998) found greater bilateral PET cerebral blood flow in the temporal and inferior frontal regions for females when generating past tense forms of verbs. To our knowledge, the only published fMRI study to date that has reported findings consistent with the Kimura model is that of Pugh et al. (1996). They reported that males and females showed different patterns of activation during semantic and phonological tasks. Females showed greater overlap of activation within the left hemisphere during semantic and phonological language tasks compared to males. This was interpreted to be consistent with Kimura’s model because the overlap reflected more circumscribed language-relevant regions in females. Not all neuroimaging studies have found significant sex differences. For example, Frost et al. (1999) found no sex differences during a language comprehension task in a large group of subjects and in the study by Schlosser et al., the authors note that only 2 of the 6 females showed right frontal activation.

We examined sex differences in neural organization among males and females using a “category-exemplar” task emphasizing understanding of semantically meaningful words that we predicted would produce different activation patterns among males and females. In this task, described by Saykin et al. (1999), participants decide if word pairs match on a semantic level. Saykin et al. reported that healthy participants showed left frontal and left superior temporal activity consistent with prior reports on semantic memory tasks. In the current study, we tested specific hypotheses based on the two models of sex differences in cerebral organization discussed above. Greater bilateral activation in females than males would lend support to McGlone’s interhemispheric model, whereas more diffuse left hemisphere activation in males compared to females would favor Kimura’s intrahemispheric model.