The goal of the present investigation was to map differential recruitment of brain regions by individuals with ASD during a social target detection task. In-scanner behavioral results indicated that, within both diagnostic groups, accuracy and reaction times were decreased for both target categories, and that performance to shape targets was worse than to face targets. Further, the ASD group demonstrated faster but less accurate responses to both categories of targets than their neurotypical counterparts. Of central relevance in the present context, however, is that both groups demonstrated comparably slower and less accurate responses to target events, validating that this oddball target detection task required cognitive control in both diagnostic groups.
Imaging data from neurotypical participants in this study reported previously (Dichter et al.
, 2009) indicated that the dACC (Brodmann's Area 32) and supracalcarine cortex were preferentially activated to face relative to shape targets. Relatively greater supracalcarine cortex activation may be conceptualized within the role of this region more broadly in face processing. Preferential activation of dACC, a region that is typically implicated in standard oddball tasks, suggested that this region may play a critical role in processing cognitive control stimuli that contain social information, a finding that is consistent with conceptualizations of this region as an intermediary processing stream between ventral cortical and subcortical regions (Mayberg, 1997
) and as a region that integrates the emotional and motivational relevance of stimuli with attentional functions (Mesulam, 1981
; Papez, 1995
Direct comparisons between brain activation in the neurotypical and autism groups revealed a number of interesting findings. First, as a whole, the autism group showed generally greater activation to target events of both categories than did the neurotypical group. In fact, no brain regions were relatively more active in the neurotypical group to shape targets, and the only regions showing relatively greater neurotypical activation to the face target condition were small localized clusters outside of the classic cognitive control network (i.e. the dACC, the midfrontal gyrus, the inferior frontal gyrus).
We interpret DMPFC/dACC hyperactivation to face targets to reflect the impaired cognitive control processes (specifically flexible responding to social information and the inhibition of prepotent response sets) that are generally reported to characterize individuals with autism (Hill, 2004
; but see Geurts, Corbett, and Solomon, 2009
regarding inconsistencies in this literature). In other words, it may be the case that hyperactivation in ASD reflects a compensatory mechanism engaged to perform the target detection task (see also, Schmitz et al.
). Data from other disorders validate the possibility that psychopathological states may be associated with hyper
activation of relevant brain regions. Explanations proposed for this pattern of findings include cortical ‘inefficiency’ (e.g. Wagner et al.
; Buchsbaum et al.
) as well as a critical dependence between brain activation magnitude and task performance (Karlsgodt et al.
). Indeed, as suggested by Manoach (2003
) in a review of an analogous issue in the schizophrenia literature, it may well be that variability in behavior and brain activation may best be regarded as intrinsic to heterogeneous disorders. Thus, we interpret the present pattern of findings within the framework of dysregulated and inefficient frontostriatal recruitment in ASD during cognitive control tasks.
A pattern of hyperactivation during cognitive control of both social and nonsocial stimuli in autism is broadly consistent with findings of other groups: (i) Schmitz and colleagues (2006
) reported greater frontal activation during go/no-go and spatial Stroop tasks and greater parietal activation during a set-shifting task, and (ii) Gilbert and colleagues (2008
) reported greater cerebellar activation during a random response generation task. However, our research group has reported hypo
activation in autism during a standard oddball task (Shafritz et al.
) and a social flanker task (Dichter and Belger, 2007
), whereas Belmonte and Yurgelun-Todd (2003
) reported a pattern of mixed results using a bilateral visual spatial attention task (i.e. decreased activity in the left ventral occipital cortex, increased activity in the left intraparietal sulcus and variable patterns in the superior parietal lobe in the autism group). Finally, others have reported comparable activation in autism using a Tower of London task (Just et al.
) and to a standard flanker task (Dichter and Belger, 2007
These seemingly contradictory patterns of findings in the literature are summarized in . Though the precise reasons for the disparities are presently unclear, the heterogeneity with respect to fMRI tasks and analyses methods do not allow for a direct comparison between studies. Additionally, heterogeneity of patient samples and matching strategies likely contribute to the inconsistent findings. Despite these limitations, a generally consistent pattern emerges of aberrant recruitment of frontostriatal systems during cognitive control in autism, though the direction of effects is inconsistent. It may be that neurofunctional compensatory mechanisms result in hyperactivation of relevant brain regions [i.e. cortical ‘inefficiency’ (Wagner et al.
; Buchsbaum et al.
)], whereas differential task performance may result in reduced activation (Shafritz et al.
) during tasks of cognitive control.
Regions of activation for experimental contrasts
fMRI studies that have investigated cognitive control in ASD.
We also note that one motivating factor for the present study was to disambiguate a potential confound of the findings of Dichter and Belger (2007
) that indicated decreased functioning of cognitive control brain regions in ASD during a social flanker task. The design of the flanker task left unresolved whether results reflected differential processing of the central or peripheral stimuli in the task array. The social oddball task utilized herein presents a single stimulus in isolation, and thus results may be linked more directly to anomalous brain activation during processing of the attended social versus nonsocial stimuli.
Exploratory covariate analyses revealed an indirect association between dACC (Target Face >Novel) >(Target Shape >Novel) activation in the 10 autism participants who received the ADOS-G and ADOS-G reciprocal social interaction scores. This inverse relation, where greater activation of the dACC was associated with less symptom severity, may appear initially to contradict the primary findings of the present study, since hyperactivity of this region was observed in the ASD relative to the control group. However, given our conceptualization described earlier that dACC hyperactivation in the ASD group reflected a compensatory mechanism, we interpret these correlational findings to indicate that ASD participants with less severe symptoms were capable of engaging such compensatory mechanisms to a relatively greater degree than those with more severe symptoms. Stated another way, it may be that brain function in less severe autism does not necessarily mimic brain function observed in neurotypical individuals, but rather reflects relatively greater compensatory brain activation to cope with environmental demands.
We note a number of limitations of the present study. First, as noted in Dichter et al. (2009), the social oddball task did not include non-face target events that differed in stimulus features from the novel and standard events (i.e. that were not geometric shapes themselves), a design feature of future studies that would be needed to confirm that the present findings are due to the social nature of face targets rather than differential saliency. Additionally, the inclusion of faces as novel stimuli would be necessary to establish that aberrant responses to face targets in the ASD group was not indicative of aberrant responses to faces per se, although we note that the high overlap of frontostriatal activation to face targets, shape targets, and the contrast of these two conditions suggests that results reflect primarily activation during target detection of these classes of stimuli. Despite these limitations, the present study expands our knowledge of brain areas mediating cognitive control of social information in ASD, and may ultimately serve as a useful benchmark of ASD treatments designed to improve social cognitive functioning.