Crespi & Badcock (C&B) have done a wonderful job synthesizing a prodigious amount of information on schizophrenia and autism. In doing so, they have made a convincing case that these disorders are diametric malfunctions of the social brain. On a surprisingly large number of anatomical and cognitive dimensions, schizophrenia (or at least its positive symptoms; negative symptoms hold an awkward place in the target article) and autism appear to be diametrically opposed. Unfortunately, the evidence C&B bring to bear on their central imprinting hypothesis is less convincing. In the following paragraphs, I expand upon this and other reservations.

Much of the evidence C&B discuss is unrelated to their core hypothesis

C&B's core hypothesis is that if gene expression is paternally biased, offspring should attempt to extract more maternal resources (increasing autism risk as a side effect) than if gene expression is maternally biased (increasing schizophrenia risk). However, the bulk of C&B's evidence (lateralization, hemispheric dysfunction, mirror neuron dysregulation), and especially their behavioral evidence (gaze sensitivity, delusions, theory of mind, etc.) has little to do with the offspring's extraction of maternal resources. Possibly a mentalistic cognitive style (related to schizophrenia) could make a child easier to raise, but it also could easily do the opposite by enabling the child to be better at playing social chess. C&B state that autistic traits appear to impose fitness costs to mothers, but so too would traits related to virtually any childhood-onset psychiatric disorder. The evidence C&B present favoring schizophrenic/schizotypic behavioral traits being less costly to mothers has to do with fertility of schizophrenic relatives (of questionable validity; see further on). But this is a red herring: Of relevance to their hypothesis is whether schizotypic traits reduce maternal investment, not whether they increase fertility.

Similarly, C&B conflate two very different issues: maleness versus femaleness, on the one hand, and paternal versus maternal gene expression, on the other. Much of C&B's evidence (Table 1 of the target article) seems more directly related to extreme male versus female brains, rather than to extreme paternal versus maternal gene expression. C&B appreciate this but seem to hedge: “By this model, the two axes overlap partially but broadly with one another, such that the so-called male brain is relatively similar … to a brain biased towards increased influence of paternally expressed imprinted genes” (sect. 7, para. 5). No explanation is provided for this striking statement, despite the fact that it is not obvious why these two axes would be related. Paternally expressed genes are expressed in both male and female offspring, and to my knowledge have no inherent relationship to maleness versus femaleness. By analogy, it would be wrong to assume that IGF-II protein levels are higher in males than females simply because the Igf2 gene is paternally active (indeed, its levels are the same in males and females; Juul et al. Reference Juul, Dalgaard, Blum, Bang, Hall, Michaelsen, Muller and Skakkebaek1995). For the same reason, it is wrong for C&B to presume that a maleness–autism association is consistent with increased paternally expressed genes, or vice versa for a femaleness-schizophrenia association.

C&B's core hypothesis lacks a clearly explained mechanism

C&B do not provide a clear explanation for why the putative imbalance in paternally and maternally gene expression occurs. Other disorders related to such an imbalance (e.g., Prader-Willi and Angelman syndromes) occur due to deletions, duplications, or uniparental disomy of specific chromosomal areas – but these causes are rare, and the disorders caused by them are commensurately rare (about 1 in 20,000). Schizophrenia and autism are, on the other hand, 50 (autism) to 200 (schizophrenia) times more common than either Prader-Willi or Angelman syndrome (Jablensky et al. Reference Jablensky, Sartorius, Ernberg, Anker, Korten, Cooper, Day and Bertelsen1992; Rutter Reference Rutter2005). What specific mechanism is being proposed that causes an imbalance in imprinted gene expression, and why does the imbalance occur so often? Moreover, what explains the heritability of autism and schizophrenia? Without a clear description of the mechanism that causes the putative imbalance, these questions can be answered only vaguely, making it difficult to derive quantifiable predictions from the hypothesis.

C&B fail to discuss counter-evidence and overstate supportive evidence

First, C&B state “imprinted genes comprise only about 1% of the genome,” (sect. 3, para. 2) but in fact the percentage is much smaller. C&B presumably intended to say 1% of genes rather than 1% of the entire genome. However, even this 1% figure (around 250 genes) would be too high – 60 to 80 genes are known to be imprinted in humans (Imprinted Gene Catalog; http://igc.otago.ac.nz/home.html). Given that only a subset of these 60–80 genes is likely to be involved in autism and schizophrenia risk, it is, again, difficult to see how imprinting in such a small subset of genes could malfunction often enough to produce the rates and heritabilities observed for the two disorders.

Second, the most direct prediction from C&B's hypothesis is that imprinted genes are over-represented among genes affecting schizophrenia and autism risk. Evidence is available but was only selectively discussed by C&B. Imprinted genes tend to occur in clusters; over half of them (47) are found in just three regions: 7q32, 11p15, and 14q32. However, the same meta-analysis discussed by C&B (Lewis et al. Reference Lewis, Levinson, Wise, DeLisi, Straub, Hovatta, Williams, Schwab, Pulver, Faraone, Brzustowicz, Kaufmann, Garver, Gurling, Lindholm, Coon, Moises, Byerley, Shaw, Mesen, Sherrington, O'Neill, Walsh, Kendler, Ekelund, Paunio, Lönnqvist, Peltonen, O'Donovan, Owen, Wildenauer, Maier, Nestadt, Blouin, Antonarakis, Mowry, Silverman, Crowe, Cloninger, Tsuang, Malaspina, Harkavy-Friedman, Svrakic, Bassett, Holcomb, Kalsi, McQuillin, Brynjolfson, Sigmundsson, Petursson, Jazin, Zoëga and Helgason2003) did not find any of these regions to be among the top 19 most likely schizophrenia gene regions (see Lewis et al. Reference Lewis, Levinson, Wise, DeLisi, Straub, Hovatta, Williams, Schwab, Pulver, Faraone, Brzustowicz, Kaufmann, Garver, Gurling, Lindholm, Coon, Moises, Byerley, Shaw, Mesen, Sherrington, O'Neill, Walsh, Kendler, Ekelund, Paunio, Lönnqvist, Peltonen, O'Donovan, Owen, Wildenauer, Maier, Nestadt, Blouin, Antonarakis, Mowry, Silverman, Crowe, Cloninger, Tsuang, Malaspina, Harkavy-Friedman, Svrakic, Bassett, Holcomb, Kalsi, McQuillin, Brynjolfson, Sigmundsson, Petursson, Jazin, Zoëga and Helgason2003, Table 2). On the other hand, data indicating likely linkage peaks near 7q32 for autism (Badner & Gershon Reference Badner and Gershon2002) seems to support C&B's hypothesis with respect to autism.

Third, C&B overstate the evidence for increased fertility among relatives of schizophrenics. C&B note that “one or more category” of first-degree relatives have higher fertility (sect. 6.4), but such “supportive” evidence includes, for example, the fertility of sisters but not brothers – even when overall sister-plus-brother fertility is lower than the general population (Haukka et al. Reference Haukka, Suvisaari and Lönnqvist2003). Such selective presentation of the data is frustrating. If one studied any group (e.g., men with arthritis) and conditioned on enough relative classes (e.g., number of maternal siblings, cf. Fañanás & Bertranpetit Reference Fañanás and Bertranpetit1995), significant fertility differences could probably be found somewhere. A more balanced presentation of the data would show that siblings and offspring of schizophrenics have equal or lower fertility compared to the general population (Bassett et al. Reference Bassett, Bury, Hodgkinson and Honer1996; Haukka et al. Reference Haukka, Suvisaari and Lönnqvist2003; McGlashan et al. Reference McGlashan, Pedersen, Hoffman and Mortensen2006; McGrath et al. Reference McGrath, Hearle, Jenner, Plant, Drummond and Barkla1999; Rimmer & Jacobsen Reference Rimmer and Jacobsen1976; Srinivasan & Padmavati Reference Srinivasan and Padmavati1997; Svensson et al. Reference Svensson, Lichtenstein, Sandin and Hultman2007), whereas parents of schizophrenics have equal or higher fertility (Avila et al. Reference Avila, Thaker and Adami2001; McGlashan et al. Reference McGlashan, Pedersen, Hoffman and Mortensen2006; Srinivasan & Padmavati Reference Srinivasan and Padmavati1997; Svensson et al. Reference Svensson, Lichtenstein, Sandin and Hultman2007; Waddington & Youssef Reference Waddington and Youssef1996). As noted by C&B, data on higher parental fertility is ambiguous because it could simply indicate that lower parental investment (pre- or postnatally), which must occur in larger sibships, is a schizophrenia risk factor.

How does evidence for a mutational role in schizophrenia and autism risk fit in with C&B's hypothesis?

C&B ignore the compelling and growing evidence that both new and circulating deleterious mutations play an important role in autism and schizophrenia risk (reviewed in Keller & Miller Reference Keller and Miller2006; McClellan et al. Reference McClellan, Susser and King2007). Indirect evidence includes: (a) the apparent fitness costs of autism and schizophrenia; (b) the likely effect sizes of susceptibility alleles; and (c) the increased risks of schizophrenia and/or autism with brain trauma, inbreeding, paternal age, and ionizing radiation. More direct evidence based on DNA sequencing or copy number detection is increasingly available and also supports a mutational role (Cheng & Chen Reference Cheng and Chen2007; Durand et al. Reference Durand, Betancur, Boeckers, Bockmann, Chaste, Fauchereau, Nygren, Rastam, Gillberg, Anckarsater, Sponheim, Goubran-Botros, Delorme, Chabane, Mouren-Simeoni, de Mas, Bieth, Roge, Heron, Burglen, Gillberg, Leboyer and Bourgeron2007; Lencz et al. Reference Lencz, Lambert, DeRosse, Burdick, Morgan, Kane, Kucherlapati and Malhotra2007; Sebat et al. Reference Sebat, Lakshmi, Malhotra, Troge, Lese-Martin, Walsh, Yamrom, Yoon, Krasnitz, Kendall, Leotta, Pai, Zhang, Lee, Hicks, Spence, Lee, Puura, Lehtimäki, Ledbetter, Gregersen, Bregman, Sutcliffe, Jobanputra, Chung, Warburton, King, Skuse, Geschwind, Gilliam, Ye and Wigler2007; Sutcliffe et al. Reference Sutcliffe, Delahanty, Prasad, McCauley, Han, Jiang, Li, Folstein and Blakely2005). It would be useful for C&B to clarify whether they consider mutations as a contributing cause of the imbalance itself, or an unrelated and additional mechanism (one of the “myriad causes,” sect. 1), or inconsistent with their hypothesis.