At the conceptual core of Crespi & Badcock's (C&B's) case are two developmental syndromes in humans attributed to imprinted genes on chromosome 15q11–13: Angelman syndrome (AS), caused by mutations abolishing expression of the maternally transcribed UBE3A gene (Lalande & Calciano Reference Lalande and Calciano2007), and Prader-Willi syndrome (PWS), caused by deficits in expression of paternal genes in the same imprinting cluster (Bittel & Butler Reference Bittel and Butler2005). Given this, and the dominance of the conflict theory for the evolution of imprinting (Haig & Westoby Reference Haig and Westoby1989; Moore & Haig Reference Moore and Haig1991), the effects of intragenomic conflict have been inferred from several phenotypes manifested in these conditions (Brown & Consedine Reference Brown and Consedine2004; Haig & Wharton 2003). The genetic causes of autism spectrum disorder (ASD) and schizophrenia are more complex than those of AS and PWS, and manifestly polygenic in nature. This should make one cautious of the authors' proposal, though not dismissive.

Because of their complex epigenetic regulation, imprinted genes are vulnerable to dysregulation (target article, sect. 3), though they are not unique in this respect (e.g., maternal behaviour regulates promoter methylation of the glucocorticoid receptor gene in rat pups; Weaver et al. Reference Weaver, Cervoni, Champagne, D'Alessio, Sharma, Seckl, Dymov, Szyf and Meaney2004). However, many imprinted genes are expressed in the mammalian brain (Davies et al. Reference Davies, Isles and Wilkinson2005), thereby presenting a large mutational target and increasing the prior probability of imprinted gene involvement in ASD and schizophrenia. Classic work with mouse embryos chimeric for wildtype and androgenetic (Ag) or parthenogenetic (Pg) cells (Allen et al. Reference Allen, Logan, Lally, Drage, Norris and Keverne1995; Keverne et al. Reference Keverne, Fundele, Narasimha, Barton and Surani1996) also suggests a role for imprinting in brain development, but some evidence presented by C&B seems contrary to neuroanatomical predictions one might derive from this work. For example, increased and decreased hippocampal size in autism and schizophrenia (target article, sect. 6.1.2) is not consistent with Pg cell accumulation in the hippocampus (Allen et al. Reference Allen, Logan, Lally, Drage, Norris and Keverne1995), and overall brain size is decreased in chimeras with a high contribution of Ag cells (Keverne et al. Reference Keverne, Fundele, Narasimha, Barton and Surani1996), contrary to brain size increases in autism (sect. 6.1.1).

These concerns aside, C&B's theory is impressive in terms of the wealth of phenomena it endeavours to embrace, and several features described in Table 1 are plausibly supportive. Even here, though, the authors' exclusive reliance on the conflict theory may be misleading. For example, in utero growth restriction is associated with paternal over-expression in transient neonatal diabetes (Temple & Shield Reference Temple and Shield2002), against the predictions of this and some other theories for the evolution of imprinting. Comparisons with existing theories or data are post hoc, and C&B know they need to propose falsifiable hypotheses. While they make some interesting predictions, we do not believe their model sufficiently specifies how imprinted genes are involved and in what phenotypes.

The behavioural phenotypes of ASD and schizophrenia are complex. In ASD the trio of “impaired social interaction, impaired communication and restricted and repetitive interests and activities” are linked conceptually by jointly providing the inclusive definition of ASD rather than biologically by any strong associations in their occurrence in psychological tests of the general population (Happé et al. Reference Happé, Ronald and Plomin2006) or in genetic twin studies (Ronald et al. Reference Ronald, Happé, Bolton, Butcher, Price, Wheelwright, Baron-Cohen and Plomin2006). Therefore, comparative studies between groups of ASD versus other individuals could produce artefactual associations between the separate components of this triad (Happé et al. Reference Happé, Ronald and Plomin2006). The same is likely to be true a fortiori for any umbrella concept of schizophrenia (Bentall Reference Bentall2003), let alone for an opposing cluster of psychoticism which also includes bipolar disorder and major depression.

These considerations raise the question, “What constitutes a continuum in biology?” Is a nominal scale sufficient? We see this conspectus of phenotypic features as more idiographic than nomothetic. But even on a nominal scale there is the problem of co-morbidity. The authors cite evidence for co-morbidity of obsessive-compulsive disorder (OCD) and attention-deficit/hyperactivity disorder (ADHD) with both autism and schizophrenia. From this they predict the existence of different types of OCD and ADHD in autism and schizophrenia. But why do they not consider co-morbidity a major problem for their position? Are there any kinds of co-morbidity that would constitute a disconfirmation of the hypothesis?

Insofar as C&B's thesis depends on conflict theory, we note the need for hypotheses to be developed regarding differential manipulation of parents in autism and schizophrenia. We remain to be persuaded that the mechanizing/mentalizing dichotomy will map onto more manipulation in autism and less in schizophrenia. It seems reasonable to suppose the existence of Machiavellian manipulators of maternal care possessed of good mentalizing abilities. We must wait for data to be collected to settle this question.

More generally, although we believe that conflict theory has considerable explanatory utility, an alternative model of imprinting evolution under sexually antagonistic selection might help C&B elucidate the links between parental gene effects and sexual differences (sect. 7). For example, a gene may tend to show expression limited to the paternal allele when alleles of that gene benefit males more than they cost females (Day & Bonduriansky Reference Day and Bonduriansky2004). Such a mechanism could provide a specific explanation for why autism appears simultaneously to be caused by an excess of paternal gene expression and manifests as an “extreme male brain” phenotype. In this context the authors' observations about the relations between sex and severity in autism and schizophrenia (e.g., their Fig. 6) seem to hint at such a selective regime.

Finally we emphasize the importance of considering alternative genetic models that explain the prevalence of ASD and schizophrenia or their components. We will mention just one here. Some alleles (of one or more genes) may show benefits when inherited alone, but cause mental dysfunction when inherited together such that selection maintains them in the population in a balanced polymorphism. For instance, Nettle and Clegg (Reference Nettle and Clegg2006) have noted that schizotypy is strongly related to creativity, which in turn has been linked to reproductive success (Miller Reference Miller2001), at least in terms of number of sexual partners over a lifetime. They confirmed this by showing that two out of four component dimensions of schizotypy were positively correlated with mating success in a large sample of British adults that included amateur and professional artists and poets.