This study used two frontally mediated language tasks to measure the functional organization of language in ASD with fMRI. The present study showed reduced hemispheric differentiation for verbal fluency tasks in individuals with autism spectrum disorders. At the group level, robust activation in left prefrontal cortex was observed in the ASD group during both verbal fluency conditions. However, a direct group comparison of the letter fluency task showed that the ASD group had greater right hemisphere activation than controls in frontal (BA 44, 45), insular, and temporal (BA 20, 21, 22, 37) regions. Between-group differences were not observed in left prefrontal cortex. A follow-up analysis conducted to further characterize the abnormal functional lateralization in prefrontal cortex for letter fluency found significantly greater leftward asymmetry in controls than in the ASD group, a difference which persisted when groups were matched on handedness. In fact, all controls showed leftward asymmetry with 10 of 14 exclusively activating the left frontal lobe, whereas 12 of the 14 ASD individuals evidenced right, bilateral, absent, or weak left lateralized activation patterns. These findings indicate that impairments in verbally mediated executive functioning tasks may be due to abnormal functional organization of language in prefrontal cortex.
Unlike the current study, decreased activation in the left inferior frontal gyrus and increased left temporal lobe activation during sentence comprehension (Just et al., 2004
; Kana et al., 2006
), semantic processing (Gaffrey et al., 2007
; Harris et al., 2006
) has been previously reported in FMRI studies of language in ASD. Increased temporal lobe activation during language processing was interpreted to reflect an unusual strength in single word processing (Just et al., 2004
). It is possible that reduced left frontal activation was not found in the current study because we used a task that required generating single words, a skill that in this paradigm is mediated by the frontal (instead of temporal) lobes. The differences between the current results and previous studies may also be due to differences in language ability. The participants with ASD in the current study were considerably more language impaired than those in Just et al, Kana et al. or Harris et al. Thus, it is possible that atypical language dominance is related to language impairment in autism, rather than specific to individuals with autism across all language abilities (Herbert et al., 2005
). Furthermore, in the Just et al study, individuals who did not have strongly left lateralized activation were excluded, which may also account for the lack of increased right hemisphere activation reported in their study. Interestingly, increased right frontal activation was reported in two studies that investigated language comprehension tasks with a social attribution component in children (mean age 11) with ASD (Takeuchi et al., 2004
; Wang et al., 2006
). The children with ASD in these studies were high functioning. Although speculative, one possible interpretation of the current findings in adults and children with ASD is that high functioning children may eventually acquire typical language dominance while individuals with greater language impairments do not. Therefore, the atypical lateralization findings by Wang et al and Takeuchi et al may reflect a transient state in ASD, reflecting perhaps delayed maturation of the frontal lobes (Zilbovicius et al., 1995
) and language lateralization processes (see below).
Task-Related Differences in Activation: Letter Fluency vs. Category Fluency
The between-group differences in lateralization of activation present in the letter fluency condition were not observed in the category fluency condition. In contrast to the strong left-lateralized activation present in controls during letter fluency (M LI = 1.62), the control group recruited right prefrontal cortex to a significantly greater extent during category fluency, resulting in weaker lateralization (M LI = .66). The direct comparison of letter fluency to category fluency in the control group did not yield a significant cluster of activation in right prefrontal cortex related to the category fluency condition, indicating the loci of right hemisphere activation(s) were variable in the control participants.
The presence of stronger left lateralization in letter fluency compared to category fluency in typically developing controls was reported by one group (Billingsley et al., 2004
) but not others (Gourovitch et al., 2000
; Mummery et al., 1996
) (Paulesu et al., 1997
). It is possible that previous studies did not find this effect because group results, rather than individual-based analyses of lateralization, were provided. Right hemisphere activation was present in all controls during the category fluency task, compared to only 4 controls in the letter fluency task. This is consistent with previous studies that have reported right hemisphere involvement in tasks which require semantic processing (Bookheimer, 2002
; Gold and Kertesz, 2000
; Kang et al., 1999
; Seger et al., 2000
; Seghier et al., 2004
Greater activation in BA 8 was also found in category fluency compared to letter fluency in the control group in the direct comparison of the fluency tasks. Activation in left superior frontal gyrus has been reported in other studies involving lexical-semantic processes (Binder et al., 1997
; Braver et al., 1997
; Demonet et al., 1992
; Muller et al., 2003
) and has been suggested to reflect working memory demands (Muller et al., 2003
). Greater working memory demands were likely present in the category fluency task compared to the letter fluency task given the significantly greater number of words produced by participants during category fluency. When more words are generated, the process of monitoring responses in order to avoid repeating a previously stated word increases working memory demands. Thus, the significantly greater activation in BA 8 in category fluency in this study likely reflects task-related differences in working memory.
Lateralization and Behavioral Impairment in ASD
In individuals with ASD, absence of strongly, left-lateralizing activation during the letter fluency task may underlie performance difficulties. Task-related differences in lateralization of activation were not present in the ASD group (M
letter = .383, M
category = .323), indicating reduced hemispheric specialization for linguistically driven tasks compared to typically developing individuals. It is also notable that between-group differences in activation in the letter fluency task were more striking than in the category fluency task, which corresponds to behavioral performance on these tasks (i.e., letter fluency behavioral performance was clinically impaired while category fluency was not). Thus, it is possible that individuals with ASD may have more difficulty performing tasks in which unilateral hemispheric specialization confers an advantage. Lateralization of functions is potentially advantageous, because greater efficiency may result from transferring information within, as opposed to across, cerebral hemispheres (Toga and Thompson, 2003
). Another possible hypothesis is that lack of hemispheric specialization may contribute to reduced abilities in right hemisphere-specialized tasks as well, which include processing the prosodic, emotional, and melodic aspects of language, and interpreting figurative meanings in language, humor, and metaphor (Toga and Thompson, 2003
). Behavioral studies report deficits in these skills in autism (e.g., McCann and Peppe, 2003
), which suggests that further study may be warranted in this area.
Early Brain Growth Abnormalities and Lateralization of Language
Atypical organization of language functions may result from early neural insults as demonstrated by experiments with perinatal focal lesions and early onset epilepsy (Adcock et al., 2003
; Muller et al., 1999b
; Muller et al., 1998b
) and has been reported in a preliminary study of five individuals with autism (Muller et al., 1999a
). As such, the atypical functional organization of language found in the current study may be due to the aberrant neurodevelopmental processes which have been recently identified in ASD. Recent studies demonstrated that rapid, excessive brain growth occurs in the first years of life in autism (Courchesne et al., 2003
; Dementieva et al., 2005
; Webb et al., 2007
). In the Courchesne et al. study, head circumference measures showed that autistic newborns were in the 25th percentile for head size at birth, which was significantly smaller than the normative sample. However, head circumference measurements increased dramatically beginning at 2 to 3 months of age, and by 6 to 14 months of age, had reached the 86th percentile relative to normative databases. From age 2 to 4 years, brain growth appears to slow down in autism, but absolute brain size is still enlarged compared to typically developing controls (Courchesne et al., 2001
). In this 2 to 4-year-old age period, increased grey matter volume in the cerebrum and increased white matter volume in the cerebrum and cerebellum is present (Courchesne et al., 2001
; Hazlett et al., 2005
). Early brain overgrowth is followed by premature arrested development, with maximum brain size being attained at approximately 4-5 years of age (Courchesne et al., 2001
). Thus, brain growth in autism appears to stop approximately 9 years earlier than brain growth in typically developing children, who do not reach maximum brain size until 12-16 years of age (Courchesne et al., 2000
; Giedd et al., 1999
The course of early brain growth and development in autism contrasts sharply with typically developing children. In typically developing children, brain growth is a slow, experience modulated process (Huttenlocher, 2002
) with regional variation in the timing of the development of cerebral cortex (Huttenlocher and Dabholkar, 1997
). The right hemisphere develops faster than in the left hemisphere in the first year of life (Toga and Thompson, 2003
) and remains dominant until approximately 3 years of age (Chiron et al., 1997
). Putative language areas (BA 44/45) exhibit adult-like cytoarchitectonic asymmetry at an even later point in development than the emergence of generalized left hemisphere dominance (Amunts et al., 2003
). Left greater than right structural asymmetry is not reached in BA 45 until 5 years of age and in BA 44 until 11 years of age in typically developing children (Amunts et al., 2003
). Amunts and colleagues suggested that delayed maturation is the microstructural reflection of the development of language abilities and that language experience may in turn influence brain anatomy.
Prematurely arrested brain growth in autism may have specific consequences for language development. Notably, brain growth in autism ceases before the mature pattern of left hemisphere dominance is reached in typically developing children (Chiron et al., 1997
), and in the majority of children with autism spectrum disorders, prior to the emergence of complex language (Charman et al., 2003
). Thus, unlike typical development where extended periods of brain growth coincide with language acquisition and mastery, in autism, language acquisition typically occurs subsequent to brain growth, without the benefits of neuroplasticity to promote the development of adaptive neural connections. Courchesne and colleagues (2003)
suggested that such aberrantly rapid and disordered growth may lead to an excessive amount of connections that may be maladaptive. Consistent with that theory, the widespread bilateral language activation in ASD found in the current study suggests that excessive, maladaptive connections and failure to develop typical lateralization may underlie disordered language in ASD. However, because volumetric analysis was not performed, we cannot determine whether reduced lateralization is also present at the morphological level in this group.
The inclusion of a diagnostically mixed group covering a wide age range was both a strength and weakness. The generalizability of the current findings to individuals across the broader, higher functioning autism spectrum is strengthened. Although the individuals in the current study differed in early developmental history and level of autistic symptomatology, as a group, consistent functional abnormalities were observed. However, given the evidence that language ability may be linked to neural abnormalities, it is possible that differences in degree or type of neural abnormality may lead to these diagnostic outcomes. Future studies that look specifically at differences in early language history and current language functioning may help determine the role these factors play in neurofunctional language organization. In addition, developmental changes in language organization throughout adolescence and adulthood in ASD were not addressed in the current study. Given the wide age-range of our sample, it is important to note that the lateralization differences reported here may not be observed to the same extent at all ages; such a possibility should also be addressed in future studies. Finally, because our ASD group was not matched on verbal ability to the control group, the neural abnormalities identified in this study may reflect language impairment generally rather than an abnormality that is specific to individuals on the autism spectrum. Further, it should be noted that reduced cognitive ability may underlie language impairment in ASD. Thus, both language impairment and general intelligence may be associated with reduced lateralization of cognitive functions.
In summary, the ASD group had significantly greater activity than controls in right frontal and temporal lobes in the letter fluency task. Between-group differences were not observed in left prefrontal cortex. A lateralization analysis of prefrontal activation found significantly greater leftward asymmetry in controls than in the ASD group in the letter fluency task. In fact, all controls showed leftward asymmetry whereas 12 of the 14 ASD individuals evidenced right, bilateral, absent, or weak left lateralized activation patterns. Between-group differences in lateralization were not found in the category fluency task, due to the significantly greater right hemisphere activation present in controls on this task. The lack of between-group differences in lateralization corresponds to the lack of clinical impairment found on the category fluency test in ASD. Overall, these data indicate reduced hemispheric differentiation for verbal fluency tasks in ASD. While lack of hemispheric specialization for higher-order tasks may have broad implications across many language abilities in ASD, greater behavioral impact may be observed in tasks which are strongly lateralized in typically developing controls (e.g., letter fluency). Abnormal functional organization may be related to early, rapid overgrowth of frontal lobes and subsequent arrested brain development recently reported in autism. Such growth dysregulation may disrupt the protracted developmental progression by which the left hemisphere becomes dominant for language, and in turn contribute to the language impairment seen in autism.