There is a certain mechanistic appeal in theoretical science's types and dimensions of personality. If we can believe in them, such paradigms hold out to us the hope of rendering a welter of human differences comprehensible in terms of cause and effect, and quantifiable in terms of position along some fundamental axis. To the extent that such models are valid, they garner a great deal in terms of predictive, explanatory, or at least descriptive power. To the extent that they are invalid, they lose a great deal in the projection of multifaceted reality onto flat theory. So theoretical abstractions are useful only insofar as they remain grounded in the phenomena that they purport to describe. Crespi & Badcock (C&B) are keenly aware of this tension between theory and phenomenon, symbol and referent, and this is why they conclude with an apology for “overly simplistic terms” and “broad, crude strokes.” Despite this acknowledged coarseness of their model, C&B have little for which to apologise – for it is the very breadth and force of such strokes that carries science across ideological limits and into new territories.

Like Baron-Cohen's (Reference Baron-Cohen2002) empathising–systemising duality, which it extends and supplants, this mentalism–mechanism duality has something of an Aristotelian feel. Indeed, it recalls medicine's first explanations of disease in terms of imbalance of the humours. In terms comprehensible to the ancient Greeks we might define the psychotic spectrum as an excess of the mentalistic humour and the autism spectrum as an excess of the mechanistic humour, and observe that these imbalances can be associated with unusual cognitive skills – just as Aristotle characterised what we now know as mood disorders as an excess of black bile, and observed among his fellow philosophers a certain tendency towards melancholic constitutions (Pies Reference Pies2007). The difference, of course, is that we don't imagine our humoural abstractions as unitary, physical substances. Rather, we are aware that at genetic and biochemical levels of analysis these mentalistic and mechanistic humours are implemented by dauntingly complex genetic networks which interact with the environment and with each other as development unfolds. This underlying complexity, though, is precisely the condition that makes these insubstantial humours so useful: In parallel with efforts to describe autism and schizophrenia in terms of network interactions among specific genes, we can now describe these conditions in terms of classes of genes united by their selection for maternal and paternal reproductive fitness. Thus, the work of C&B enables a trait-level analysis of autism and psychosis that complements gene-level analysis.

As C&B point out, this trait-level view in terms of mentalism and mechanism can in some regards muster greater descriptive and explanatory power than the more immediate view in terms of paternally and maternally imprinted genes themselves, because abnormal biases constructed by such genes can interact with normal programmes and gradients of brain development, spurring knock-on effects as the process of interactive specialisation of neural and cognitive systems unfolds (Johnson et al. Reference Johnson, Halit, Grice and Karmiloff-Smith2002; Karmiloff-Smith Reference Karmiloff-Smith2007). Support for the notion of such multi-level cascades of perturbed development comes from studies of schizophrenia demonstrating deficits in early sensory processing (Butler & Javitt Reference Butler and Javitt2005; Butler et al. Reference Butler, Martinez, Foxe, Kim, Zemon, Silipo, Mahoney, Shpaner, Jalbrzikowski and Javitt2007; Uhlhaas & Silverstein Reference Uhlhaas and Silverstein2005) and relating auditory frequency discrimination to deficits in affect recognition (Leitman et al. Reference Leitman, Hoptman, Foxe, Saccente, Wylie, Nierenberg, Jalbrzikowski, Lim and Javitt2007) and visual size discrimination to deficits in theory of mind (Uhlhaas et al. Reference Uhlhaas, Phillips, Schenkel and Silverstein2006); from studies of autism demonstrating enhanced perception of first-order stimuli (Mottron et al. Reference Mottron, Dawson, Soulières, Hubert and Burack2006) and associating theory-of-mind performance with joint attention (Charman Reference Charman1997) and perceptual disembedding (Jarrold et al. Reference Jarrold, Butler, Cottington and Jimenez2000); and from models relating C&B's cited variations in brain size to systems-level variation in the balance between short-distance and long-distance connectivities (Lewis & Elman Reference Lewis and Elman2008). In each of these instances, what seems on face an alteration of higher-order cognition is seen to be related to, and in some cases a consequence of, perturbations at lower levels of processing – and in this regard the cited conditions are not disorders of the “social brain” exclusively. Though such relationships may be most obvious in cases of clinical disorders, they seem unlikely to be limited to disordered cognition and instead may extend to normal variation – that is, to the entire continuum of cognitive variation between the mentalistic and mechanistic extremes.

Further refinement will be necessary to explore aspects in which the autism spectrum and the less straightforwardly defined “psychotic spectrum” may have been miscast as opposites. As C&B note, some empirical aspects are not comparable between these two populations with existing data. Schizophrenia is diagnosed much later than autism, and therefore there remains a dearth of early developmental observations. For example, although the autistic amygdala is indeed abnormally large during childhood and therefore seems a counterpoint to the abnormally small amygdala in schizophrenia, by adulthood the autistic amygdala likewise has become abnormally small (Aylward et al. Reference Aylward, Minshew, Goldstein, Honeycutt, Augustine, Yates, Barta and Pearlson1999; Rojas et al. Reference Rojas, Smith, Benkers, Camou, Reite and Rogers2004). Likewise, observations of frontal hypoactivation in autism may depend strongly on the behavioural task and on the latency at which activation is measured – in some contexts autistic frontal systems are hyperactivated (Takarae et al. Reference Takarae, Minshew, Luna and Sweeney2007). Regarding sensory filtering, autism shares with schizophrenia a decrease in prepulse inhibition (McAlonan et al. Reference McAlonan, Daly, Kumari, Critchley, van Amelsvoort, Suckling, Simmons, Sigmundsson, Greenwood, Russell, Schmitz, Happé, Howlin and Murphy2002; Perry et al. Reference Perry, Minassian, Lopez, Maron and Lincoln2007), and although people with autism do often focus attention very strongly, in some cases maintaining attention is an issue, and many people with autism fulfill criteria for attention-deficit/hyperactivity disorder (Corbett & Constantine Reference Corbett and Constantine2006). Familial association of autism and mood disorders (DeLong Reference DeLong2004) calls into question the inclusion of mood disorders within the “psychotic spectrum” in this context. The connectivity issues too remain to be sorted out; both autism and schizophrenia have been described as deficits in brain connectivity (Frith Reference Frith2005a), but these may be secondary to developmental interactions whose antecedents differ between the two conditions. Despite such areas for refinement, the mentalism–mechanism duality has strong potential to change the way we think about the relation of clinical disorders to normal human cognitive variation, as well as the evolutionary forces that have produced both.