Autism is a neurodevelopmental condition characterized by marked impairments in the domains of social interaction, language/communication, and pronounced repetitive stereotyped behaviors and/or restricted interests. The etiology of autism is complex as many genetic syndromes characterized by single-gene mutations (e.g., fragile X syndrome, Rett syndrome, tuberous sclerosis, Timothy syndrome) can result in a phenotype that meets the diagnostic criteria of autism (i.e. ‘syndromic autism’) . Furthermore, research on non-syndromic forms of autism show that many simplex cases (up to 10% to 20%, but possibly more) possess de novo mutations in segments of DNA (i.e. deletions and/or duplications) presented as copy number variations [2–4]. This has resulted in the idea that there may be many types of autism arising from a complex combination of many different mechanisms.
In order to decompose some of this complexity, there is a need for clear models at a variety of levels of analysis. Autism genetics has already provided several useful models for understanding how genetic mechanisms linked to autism may have pathophysiological impact [5–8]. However, we currently lack clear models at higher levels of analysis such as at the level of neuroanatomy. In this paper, we focus on agenesis of the corpus callosum as one model at the level of neuroanatomy for dissecting aspects of self-referential and social-cognitive difficulties in autism. More specifically, we focus on areas within the self-referential domain related to the cognitive benefits of self-referential information processing (i.e. the self-reference effect in memory) as well as difficulties in emotional awareness (i.e. alexithymia), which are known to be more pronounced in higher-functioning adults with ASC. Within the social-cognitive domain, we focus on higher-level social-cognitive understanding of others (i.e. mentalizing, empathy), as well as memory for social agents; each of which have been shown to be sensitive to deficits in higher-functioning adults with ASC.
Agenesis of the corpus callosum (AgCC) is a congenital condition manifested in the failure to completely develop a corpus callosum; the largest white matter tract connecting the two cerebral hemispheres [9, 10]. AgCC can present as a complete or partial absence of the CC. Complex AgCC occurs with the presence of other confounding brain abnormalities such as polymicrogyria, heterotopia, or schizencephaly, whereas isolated AgCC does not occur with such abnormalities. However, isolated AgCC commonly co-occurs with colpocephaly (i.e. dilation of the posterior section of the lateral ventricles and reduction of ipsilateral cortical association tracts) and Probst bundles . The etiologies of AgCC are complex and likely reflect a multitude of routes important for fully establishing callosal connections (i.e. cellular proliferation and migration, axon guidance and growth, glial development, and midline patterning). Many genetic syndromes are associated with AgCC (e.g., X-linked lissencephaly, Mowat-Wilson syndrome, CRASH syndrome), and some of these syndromes have been associated with autism (i.e. ARX mutations leading to X-linked lissencephaly ). However, like autism, the majority of AgCC cases do not have an identifiable single genetic cause . There are important clinical and neuroanatomical parallels between AgCC and autism that may suggest the role of atypical callosal development in understanding the emergence of phenotypes associated with autism.
First, one of the more replicable findings in neuroscience research on autism is marked atypicalities in the corpus callosum (CC). In a meta-analysis of volumetric MRI studies on the CC in autism, Frazier and Hardan  found that all subsections are consistently observed as reduced in volume. This result is further bolstered by longitudinal evidence for persistence of CC volumetric reduction at a 2-year follow-up examination . In diffusion MRI work, several atypicalities in CC integrity, particularly with respect to reduced fractional anisotropy and increased mean diffusivity are reported [14–20]. Finally, a recent study using magnetization transfer imaging as an index of processes relevant to myelination found evidence for atypical myelination of the CC in autism . The mounting evidence for atypical CC development in autism, particularly with respect to reduced size, potentially indicates that mechanisms involved in AgCC may also be important in autism.
Second, there are important clinical aspects that overlap between AgCC and autism. One of the most well known cases of such overlap is the case of the real ‘Rainman’, Kim Peek, who was diagnosed with autism well before later MRI scans determined he also had complete AgCC. Approximately 3% to 5% of individuals with neurodevelopmental conditions also have AgCC or CC hypoplasia [22, 23], and this is of particular relevance for autism, where CC volume is notably reduced . In addition, 8.5% of individuals with AgCC also have a specific diagnosis of autism . However, this estimate may be conservative [9, 25]. Badaruddin et al.  reported that in a sample of 61 two- to eleven-year-old children with AgCC, as many as 34% met criteria for at least one item in the domain of social impairment, 25% met criteria for at least one item in the communicative domain (with 51% showing difficulty in sustaining conversation), and up to 28% had a preoccupation with a specific interest. More recently, in 106 patients with AgCC, Lau et al.  found that 43% of children, 35% of adolescents, and 18% of adults score above cut-offs on the Autism Spectrum Quotient (AQ) that are typically used for screening for clinical diagnoses of autism spectrum conditions (ASCs).
Among the most prominent findings in the neuropsychological profile of AgCC are impairments in self-referential and social cognition . Within the social-cognitive domain, many atypical findings in AgCC overlap with those seen in autism. Theory of mind or mentalizing impairments have long been known in autism [28, 29], and these impairments can be helpful for explaining difficulties in pragmatic aspects of social communication . Deficits in cognitive aspects of empathy [31–33] and on self-report measures of empathy [34–38] have been consistently reported in ASC. Affective components of empathy are spared in ASC [31, 33], although self-report measures of this component are somewhat inconsistent. Some studies show decreases on subscales such as empathic concern [35, 39] but increases on subscales such as personal distress . The latter subscale is likely linked to comorbid anxiety traits in ASC . Still other reports of self-reported affective empathy show no differences [36, 38]. Given difficulties in ASC with self-referential cognitive processing, it is likely that the experimental studies on the topic comparing ASC to other groups with dissociable empathy deficits, such as individuals with psychopathic tendencies or conduct disorder [31, 33], give the clearest indication of the empathy profile in ASC rather than relying solely on self-report measures.
Several studies now show that individuals with AgCC have marked impairments in pragmatic aspects of language processing, including comprehension of idioms, proverbs, vocal prosody, and non-literal interpretation of humorous statements [41–44]. Outside of pragmatic social communication difficulties, AgCC also shares social-cognitive deficits with those seen in autism. These difficulties extend into difficulty providing narratives that demonstrate understanding of social-emotional aspects of stories [45–47]. There are mixed reports regarding AgCC and ToM deficits. One study found deficits on the Reading the Mind in the Eyes test in some, but not all, AgCC patients . However, other work on adolescents and adult patients did not reveal deficits on other advanced ToM stories and faux pas tests  known to be sensitive to deficits within autism [49, 50]. While Symington et al.  did not find such advanced ToM deficits, they found difficulties in AgCC when asked to interpret videotaped social vignettes which indicated subtle deficits in emotion recognition, understanding sarcasm, and difficulty interpreting textual cues . Finally, AgCC patients tend to underestimate emotional valence and intensity from negatively valenced pictures . However, no studies exist in AgCC that comprehensively test components of empathy, making comparison with ASC difficult.
Prior work in the domain of self-referential cognition also indicates that difficulties in autism [35, 52–54] may be shared by patients with AgCC. First, individuals with ASC show deficits across a wide range of areas where self-relevant information processing is critical, such as emotional understanding (e.g., alexithymia) , memory [35, 56], introspection , pronoun usage [35, 58], orienting to name [59, 60], monitoring own intentions , remembering own false beliefs , attributing privileged access to self over close others  (for review see [52, 53, 56, 58, 64, 65]). Within AgCC, one case study highlighted the possibility that alexithymia may be more pronounced . This observation is similar to other reports on alexithymia hypothesizing that one aspect of its neurophysiological basis is reduced interhemispheric transfer [66–69]. Parental reports indicate that patients with AgCC show poor personal insight . Also indicative of a potential lack of insight is the effect that parental ratings tend to be higher than self-reported ratings on the AQ . Analysis of narrative content indicates that patients with AgCC tend to use more first-person pronouns than comparison groups . Finally, Brown and Paul  suggest that high scores on the Minnesota Multiphasic Personality Inventory Lie scale (without elevations in other scales) may be indicative of poor self-awareness in the two AgCC cases they tested.
Despite the evidence reviewed here with regard to self-referential and social cognition, much work is still needed. At present, very few studies on AgCC exist (e.g., ) that assess social-cognitive abilities known to be clearly atypical in autism. In addition, even less research in AgCC specifically tests known self-referential atypicalities in autism. In this study, we add to this literature by using a battery that measures processes such as empathy, mentalizing, alexithymia, self-referential memory biases, private self-consciousness, and implicit self-focused attention. We also attempt to go further by assessing a case with both partial AgCC and autism in comparison to both neurotypical controls and cases of autism without AgCC. By comparing this single case to the two groups, we examined the extent to which aspects of his self-referential and social-cognitive profile overlap with the pattern found in autism and also identified aspects which may be even more atypical than those seen in autism. From the viewpoint that individuals with autism typically have smaller CCs than controls, AgCC could be construed as an extreme of this profile. Therefore, highlighting areas in AgCC that are even more extreme compared to the autism group may provide some indication that the CC is critical for such aspects of self-referential and social cognition.