Imprinted genes expressed in the brain are numerous, and they play important roles in nervous system development (Keverne et al. Reference Keverne, Fundele, Narasimha, Barton and Surani1996). Elucidating their influence on psychological function will require greater understanding of epigenetic processes among scientists trained to study behavioural phenomena. For this reason, the comprehensive work presented in Crespi & Badcock's (C&B's) article is an important one. The article is exhaustively researched and provides direction for further investigation. Unfortunately, the authors try to explain too much with too little, outstripping the explanatory capability of the model. A wide variety of psychiatric diagnoses are included in the target article, and their dichotomous categorization seems forced at times. Furthermore, the core “balanced see-saw” model presented is insufficiently precise.

Phenotypes in psychiatric research

The target article attempts to integrate a variety of psychological and psychiatric phenomena, many of which have not previously been conceptualized as related. Phenotypic commonalities can be beneficial guides in the search for etiological factors, particularly in the case of complex psychiatric disorders. However, the most narrowly defined traits are likely to be of greatest utility. The broad phenotypes used in target article, alternating between symptoms such as psychosis or depression and diagnostic categories such as autism, schizophrenia, and obsessive-compulsive disorder (OCD), are problematic in this regard.

The power to identify relevant genetic factors is directly influenced by the “purity” of the relationship between the trait under study and the underlying genes. Failure to identify appropriate phenotypes for genetic study is exacerbated in the field of psychiatric genetics where diagnostic classification is often used as the phenotype of interest (Feinstein & Singh Reference Feinstein and Singh2007). This is a very serious mistake, as heterogeneity is rampant within diagnostic categories, and individuals with the same diagnosis may vary significantly in phenotype and etiology, even in the presence of high heritability.

Endophenotypes are traits that function as indices of underlying genetic liability to disorder: they are phenotypic manifestations intermediate to the genes and the overt clinical syndrome (Gottesman & Gould Reference Gottesman and Gould2003). Consequently, they are more likely to be associated with specific susceptibility genes, and may show higher heritability than the complex clinical endpoints that are the usual focus of psychiatric research.

The use of endophenotypes in the study of complex psychiatric disorders is increasing, and has been shown to facilitate the identification of genetic risk factors (Castellanos & Tannock Reference Castellanos and Tannock2002; Kruglyak & Lander Reference Kruglyak and Lander1995). This approach has advanced the study of attention-deficit/hyperactivity disorder (ADHD; Crosbie et al. Reference Crosbie, Perusse, Barr and Schachar2008; Schachar et al. Reference Schachar, Crosbie, Barr, Ornstein, Kennedy, Malone, Roberts, Ickowicz, Tannock, Chen and Pathare2005); and evidence that different symptoms in autism are independently influenced by different genes (Ronald et al. Reference Ronald, Happé, Bolton, Butcher, Price, Wheelwright, Baron-Cohen and Plomin2006; Shao et al. Reference Shao, Cuccaro, Hauser, Raiford, Menold, Wolpert, Ravan, Elston, Decena, Donnelly, Abramson, Wright, DeLong, Gilbert and Pericak-Vance2003) suggests the same may be true for this complex disorder. To investigate potential imprinting effects, grouping on the basis of behavioural or psychiatric commonalities with known imprinted disorders (such as Prader-Willi syndrome [PWS]), and the study of parent-of-origin effects, are also effective ways to proceed (Goos et al. Reference Goos, Ezzatian and Schachar2007; Goos & Ragsdale Reference Goos, Ragsdale and Wilkins2008).

Imprinted genes and social cognition

The authors' imprinted brain model of autism is based on the enhanced influence of paternal genes relative to maternal. The “spectrum” of psychosis is presently suggested as the outcome of increased relative influence of maternal genes over paternal. C&B posit no significant phenotypic divergence based on the genetic cause of the relative over-expression (i.e., duplication vs. deletion).

This view is consistent with evidence of functional complementarity in a number of imprinted genes, such as insulin-like growth factor 2 and its receptor, which form a tightly linked physiological unit (Haig & Graham Reference Haig and Graham1991). In this case, deletion or duplication of either gene produces predictable outcomes of over- or undergrowth. This logic is unlikely to apply, however, in the case of highly variable behavioural phenomena.

The evolutionary forces driving imprinting (Haig & Westoby Reference Haig and Westoby1989) are compounded and exacerbated by those due to sociality (Haig Reference Haig, LeCroy and Moller2000a). In primates, social skills directly influence both survival and reproductive success (Silk Reference Silk2007). Furthermore, primate social interactions occur primarily with maternal kin (Cheney et al. Reference Cheney, Seyfarth and Smuts1986). In this context, the adaptive resolution of conflict over resource transfer, for maternal and paternal genes, would involve social skills, in order to maximize resource acquisition, minimize individual risk due to conflict, and restrain direct and indirect maternal fitness costs.

Uniparental cell deposition patterns in chimeric mice support this conjecture. Maternally expressed genes contribute primarily to the development of cortical areas mediating complex socio-cognitive skills, such as memory, planning, and language (Allen et al. Reference Allen, Logan, Lally, Drage, Norris and Keverne1995). The paternal contribution is primary in the mediobasal forebrain and hypothalamus (Keverne et al. Reference Keverne, Fundele, Narasimha, Barton and Surani1996). These areas are crucial for adaptive sociality, largely mediated via oxytocin and vasopressin. These hormones reduce social fear, anxiety, and aggression; moderate stress reactivity; facilitate social memory and behavioural flexibility; and coordinate behavioural and emotional responses to social stimuli (Carter & Altemus Reference Carter and Altemus1997; Kirsch et al. Reference Kirsch, Esslinger, Chen, Mier, Lis, Siddhanti, Gruppe, Mattay, Gallhofer and Meyer-Lindenberg2005; Winslow & Insel Reference Winslow and Insel2004).

If both maternal and paternal genes were selected to maximize fitness in a social milieu, it is likely that imbalance in either direction would result in social deficits. Moreover, different suites of social impairments may legitimately attract the same diagnosis of autism. Relative paternal over-expression due to mutation(s) in maternally expressed genes could produce the cognitive and language deficits characteristic of autism, as suggested by C&B. However, relative maternal over-expression due to mutation(s) in paternally expressed genes could lead to deficits in socio-emotional processing, such as heightened reactivity, anxiety and fear, and stereotypy and rigidity, also characteristic of autism (Corbett et al. Reference Corbett, Mendoza, Abdullah, Wegelin and Levine2006; Hollander et al. Reference Hollander, Novotny, Hanratty, Yaffe, DeCaria, Aronowitz and Mosovich2003; Jansen et al. Reference Jansen, Gispen-de Wied, van der Gaag and van Engeland2003). Indeed, this postulation is consistent with evidence in the literature documenting cortical or neuroendocrine dysfunction in autism, but not necessarily both (Belmonte & Carper Reference Belmonte and Carper2006; Chandana et al. Reference Chandana, Behen, Juhasz, Muzik, Rothermel, Mangner, Chakraborty, Chugani and Chugani2005; Green et al. Reference Green, Fein, Modahl, Feinstein, Waterhouse and Morris2001; Jacob et al. Reference Jacob, Brune, Carter, Leventhal, Lord and Cook2007; Modahl et al. Reference Modahl, Green, Fein, Waterhouse, Feinstein, Morris and Levin1998; Wu et al. Reference Wu, Jia, Ruan, Liu, Guo, Shuang, Gong, Zhang, Yang and Zhang2005).

The target article represents an important step in the integration of genomic imprinting and the study of psychopathology. It would, however, be much more compelling and influential if it were more finely parsed with respect to the imprinted genotypes included in it, as well as the behavioural phenotypes it seeks to explain.