The representation of self and other may be fundamentally altered in autism spectrum disorders 
. While previous work has largely emphasized deficits in interpersonal interaction in ASD, current empirical work emphasizes the need for understanding differences with respect to both interpersonal
(social) and intrapersonal
(self-referential) cognition 
. Previously we and others have shown that in normal adults, self-recognition is accompanied by signal changes in a right hemisphere network including the inferior frontal gyrus and inferior parietal lobule 
, consistent with earlier work implicating preferential right hemisphere involvement in self-recognition 
. Here we used this paradigm to test whether children with ASD recruit the same brain areas during self- and other-face processing as do TD children. We showed that while self- and other-face processing involve virtually overlapping right frontal activity in TD children, children with ASD only exhibit such activation when viewing their own faces. ROI analyses confirm that ASD and TD children differentially activate the rIFG (specifically BA44 and BA45) during viewing of other faces. Specifically, children with ASD activate the rIFG less than do TD children during viewing of other faces. In every other ROI within the broader face-processing networks we examined, we observed no significant group differences during viewing of self or other faces. Thus, consistent with clinical observations of higher levels of self-focused behavior in autism, these children show decreased neural response to viewing faces of others compared to viewing faces of themselves. While hypoactivation of the fusiform gyrus to faces has been previously reported 
, this region does not appear to be involved in self-other distinction as examined in the current study. One can speculate that normal engagement of several brain regions by self faces and reduced activity to other faces in the rIFG might be related to the social impairments characteristic of children with ASD.
Previous work on the neural basis of self-face recognition in adults suggests that while the right inferior frontal gyrus responds most strongly to self-faces, it also shows signal increases (although weaker) above baseline to personally familiar faces 
. The fact that the TD children examined in this study do not yet show self-other differentiation in the right IFG suggests that our cross-sectional design captures a period in cognitive development where this distinction is not robustly represented, and a great deal of self-other overlap exists. One may speculate that as the representation of self becomes more differentiated in post-adolescent development, the rIFG activation to self relative to others becomes more selective. In contrast, children with ASD appear to not exhibit this neural overlap. As our study did not use personally familiar faces as controls, we cannot address the interesting question of whether the response of the rIFG of children with ASD is modulated by personal familiarity, as has been suggested for the fusiform face area 
Our paradigm required participants to explicitly evaluate facial identity to decide whether the image presented resembled themselves. Both groups of children behaviorally demonstrated a diminishing “self” response as the image presented contained a smaller percentage of the self face, indicating successful self-other discrimination. This result is consistent with behavioral work demonstrating successful visual self-recognition in most children with ASD 
. Our previous work has implicated the right inferior parietal lobule in this type of discrimination, as virtual lesions induced with rTMS reduced subjects sensitivity to detecting self faces 
. The right IPL is thought to be part of a circuit mediating complex own-body perception 
, and contributes to the sense of agency, or the feeling of generation of action 
. We saw no significant group differences during viewing of self-faces in this region, suggesting that the perception of the self face as part of one's own body is not altered in children with ASD. Rather, it appears that the mechanisms for detecting self-other similarity, likely dependent on the right inferior frontal gyrus, are dysfunctional in these children. The mechanisms implemented by the rIFG likely mediate differences between ASD and TD children in social interactions that are more complex than those tapped by our self-other discrimination task.
Social cognition researchers have previously suggested that understanding of others' experiences may involve the activation of shared affective neural networks that enable us to “feel the emotions of others as if they were our own” 
. There is now a fairly substantial literature documenting the existence of shared neural representations that bridge the gap between the self and other in various domains, including the experience of pain 
, touch 
, and emotion 
. Such “embodied simulation” accounts of action and emotion understanding go by many names (e.g. “shared manifold of intersubjectivity” or “intentional attunement” 
, “shared representations” 
, “shared circuits” 
), but they all support the notion that one of the ways by which an individual makes sense of the social world is by using the same brain systems that are used for self-related experiences in order to understand others. This overlap in self- and other-representation may breakdown in autism, which is characterized by decreased empathy 
and theory of mind impairments 
. The restricted social interests of individuals with autism may thus reflect a fundamental lack of appreciation of self-other similarities, which may be the result of altered mirroring mechanisms in the brains of such individuals. There is also mounting evidence for dysfunction of the so-called mirror neuron system, which is implicated in social cognition 
, in individuals with autism 
One possible explanation for the lack of rIFG response to others observed in this group of children with ASD may be the fact that from a young age autistic individuals seem to lack the motivation for orienting to social cues that their typically developing counterparts demonstrate 
. Indeed, some have theorized that the lack of interest in social stimuli evident in autistic children may result from the fact that ASD children do not find these stimuli to be rewarding 
. Previous reports have shown that when children with ASD are instructed to pay attention to specific aspects of social stimuli (e.g. a person's facial expression or tone of voice), neural responses more closely resemble those of TD children 
. Thus, while socially relevant stimuli such as other's faces may not automatically engage the attention of children with ASD who are less driven by social motivation, explicit instruction to attend can ameliorate this effect. In the present study, however, the task instructions focused the children's attention to detecting the self, perhaps further decreasing the amount of attention directed toward others. Future work should examine whether simply modifying task instruction (i.e. asking participants to detect the presence of other faces in the morphs) may “normalize” neural activity to other faces in children with ASD.
In conclusion, we find that children with ASD do not activate shared regions for self- and other-face processing, as do TD children. Reduction of activity in right prefrontal areas during other-face processing may be a neural signature of reduced social engagement and understanding in these individuals.