We begin by stating that the theoretical framework of Crespi & Badcock (C&B) is an important contribution to the unraveling of the nature of disorders of the social brain. However, although it is not our intention to question the consensus that autism and schizophrenia are two different psychiatric disorders, we argue that the described differences in brain functioning and cognition between autism and psychosis should be considered within a developmental context. The comparison of social-cognitive dysfunctions in autism and psychosis as C&B have presented, implies in most cases a comparison of two different developmental windows. Autism is by definition a childhood-onset disorder, whereas psychosis is diagnosed primarily in (early) adulthood. This may have important implications, as findings from developmental neuropsychology suggest that the same cognitive dysfunctions may present differently at different ages.

First, depending on the maturation of the brain, specific cognitive functions become “on-line” as development progresses. For example, maturation of the frontal cortex continues into early adulthood, and executive functions will therefore be quantitatively and qualitatively different in early adulthood as compared to childhood. Hence, certain cognitive impairments will become more salient as development progresses. Considering this “growing into deficit,” a concept proposed by Rourke et al. Reference Rourke, Bakker, Fisk and Strang(1983), the profile of cognitive impairments may change over time in individuals with a neurodevelopmental disorder. Hence, the pattern of behavioral difficulties may also change over time in some individuals possibly resulting in a shift in psychopathology. Recent reports suggest that shifts from the autism spectrum to the psychosis spectrum are mainly seen when considering the broad autism spectrum, that is, children who have severe social impairments but who may not meet full criteria for the core autism phenotype. Mouridsen et al. Reference Mouridsen, Rich and Isager(2008) have followed up 89 children with atypical autism, of which 30.4% developed a psychotic disorder in adulthood, whereas only 1.9% of the 258 non-clinical controls did so. Second, Karmiloff-Smith Reference Karmiloff-Smith(1997) has stressed the importance of considering the interactive nature of cognitive impairments associated with neurodevelopmental disorders. Thus, the neurodevelopmental origins of two distinct phenotypical outcomes, such as autism and psychosis, may show similarities. However, the exact phenotypical outcome is shaped by interactions with other neurodevelopmental and environmental factors during development.

Although autism and schizophrenia can quite reliably be separated as two distinct syndromes by DSM-IV or IDC-10 criteria, at the level of individual symptoms some of the behavioral characteristics are observed in both disorders. This is not new, as Bleuler Reference Bleuler, Bleuler, Dover, Bleuler and Dover(1911) already delineated “autism” as a cardinal feature of schizophrenia. C&B attribute this overlap to superficial behavioral similarities and argue that autistic features may also be expressed in some negative schizophrenia symptoms. However, there is also evidence for increased levels of paranoid traits in individuals with Asperger's syndrome (Blackshaw et al. Reference Blackshaw, Kinderman, Hare and Hatton2001; Craig et al. Reference Craig, Hatton, Craig and Bentall2004), suggesting that behavioral overlap is not limited to negative schizophrenia symptoms and extends to psychotic traits. Taking this together with the dynamics of cognitive development, we argue that although social-cognitive impairments may seem different in (adults with) psychosis as compared to (children with) autism, this does not necessarily point to two different or opposite social brain dysfunctions.

Indeed, reports of similarities in social-cognitive performance in both disorders support this suggestion. Besides mentalizing deficiencies in autism, a recent meta-analysis has pointed to deficits in theory of mind (ToM) in different subtypes of schizophrenia patients (Sprong et al. Reference Sprong, Schothorst, Vos, Hox and van Engeland2007). Not only the schizophrenia patients characterized by disorganized or negative symptoms, but also those with predominantly persecutory delusions scored significantly lower in each study, on average 1.2 standard deviations lower than non-clinical controls, in various tasks measuring mentalizing and ToM capacities. Moreover, in a study by Harrington and colleagues (Harrington et al. Reference Harrington, Langdon, Siegert and McClure2005a), the severity of persecutory delusions in schizophrenia patients was related (r=−.48) to more severely impaired mentalizing abilities. Also, patients with paranoid delusions could not be discriminated from individuals with Asperger syndrome on several tasks measuring ToM skills (Craig et al. Reference Craig, Hatton, Craig and Bentall2004), as both groups performed comparably worse as compared to non-clinical controls. Not only in clinical populations, but also in the general population, “positive” schizotypal traits (analogous to psychotic clinical symptoms) seem to predict poorer mentalizing performance (Pickup Reference Pickup2006). One possible explanation for the relation between psychosis or paranoia and ToM impairments has been put forward by Bentall and Kinderman Reference Bentall, Kinderman, Lewis, Tarrier and Wykcs(1998). They propose that the following two factors contribute to persecutory delusions: (1) an external attributional bias, and (2) an impairment in incorporating contextual information needed for making attributions. An impaired ability to understand what other persons are thinking and feeling (ToM) represents one type of deficit in taking into account important contextual information.

Interestingly, similarities between the autism and psychotic spectrum have also been observed with regard to the neurobiological underpinnings of impaired theory of mind. When asking individuals to judge faces with regard to trustworthiness, that is, attribute intentions to others, functional magnetic resonance imaging (fMRI) shows very similar patterns of brain activation in individuals with an autism-spectrum disorder and paranoid schizophrenia patients (Pinkham et al. Reference Pinkham, Hopfinger, Pelphrey, Piven and Penn2008).

Finally, our studies on Klinefelter syndrome also suggest an overlap in social-cognitive dysfunctions associated with autistic and psychotic behavior. Men with Klinefelter syndrome display various social-cognitive deficits and have increased levels of both autistic and psychotic traits (van Rijn et al. Reference van Rijn, Aleman, Swaab and Kahn2006a; Reference van Rijn, Swaab, Aleman and Kahn2006b; Reference van Rijn, Aleman, Swaab, Krijn, Vingerhoets and Kahn2007; in press).

In sum, we feel that C&B's comparison of the cognitive mechanisms characteristic of the autism spectrum and psychotic spectrum should take into account the dynamics of cognitive development from childhood into adolescence and adulthood. Also, the observations of similarities in social-cognitive dysfunctions in the autism and schizophrenia spectrum should not be ignored. We hope to have provided convincing evidence showing that mentalizing abilities, which are among the core domains of cognitive functions that dissociate autism from psychosis according to C&B, are hypodeveloped in both the autism and psychotic spectrum. As also suggested by Pinkham et al. Reference Pinkham, Hopfinger, Pelphrey, Piven and Penn(2008), social brain dysfunctions in the autism and psychotic spectrum may be deficit-specific, rather than disorder-specific.