This target article assumes that a mapping between individual differences in IQ and various brain characteristics constitutes a testable theory of intelligence. However, we would argue that even within the advent of the modern neuroimaging revolution, this is contestable on both conceptual and methodological grounds.
We remind readers that the modern notion of g is based on intercorrelation among test performances, but “g tells us little if anything about contents” (Jensen Reference Jensen1998, p. 92). The intercorrelation may be partially or entirely based on noncognitive factors, so there are still serious doubts whether the intelligence envisaged by Jung & Haier (J&H) exists (van der Maas et al. Reference van der Maas, Dolan, Grasman, Wicherts, Huizenga and Raijmakers2006). Critics say that indices of g, such as IQ test performance, measure degree of enculturation into specific linguistic and cognitive styles, together with motivational, anxiety-related, and other test-preparedness factors: in effect, it is a good measure of social class (not well measured by socioeconomic status [SES]; Cole Reference Cole and Sternberg1999).
Just as interpretation from correlations among IQ scores is debatable, so is that from imaging data, as J&H recognize: “all neuroimaging research is correlational by nature” (sect. 5.7, para. 3). Yet their whole theoretical effort is based on correlations between those correlations. It is scarcely surprising that the results are mixed and inconsistent. As J&H say, in “more than 40%” of voxel-based morphometry studies, “tissue density and white matter integrity … correlate substantially” with IQ (sect. 5.3, para. 1) – meaning that in nearly 60% of studies they don't. Similar interpretations apply to the fMRI, PET, and other data. What, for example, does “consistently related … across more than 30% of studies” mean (sect. 6, para. 2)? Sometimes correlations are positive, sometimes negative (e.g., sect. 5.4, para. 2). Finding areas that are activated during cognitive tasks is intrinsically interesting. But that more than 20 different tasks were used, together with unreliability of methodology, makes interpretation difficult. (Giuliani et al. Reference Giuliani, Calhoun, Pearlson, Francis and Buchanan2005, cited for reassurance in sect. 5.1, para. 2, actually say that “Although VBM is rapid and fully automated, it is not a replacement for manual ROI-based analyses. Both methods provide different types of information” [p. 135].) And all this must be appraised within the context of the “noise” created from a multitude of different complex cognitive tasks in different studies. Even fMRI studies on well-defined, replicable tasks, such as hand flexions/extensions, have stressed the interdependence between brain regions (Marrelec et al. Reference Marrelec, Krainik, Duffau, Pélégrini-Isaac, Lehéricy, Doyon and Benali2006).
Even if the correlational patterns, and interpretations of them reported by J&H, were robust, we suspect that better explanations could be found within the “ecology” of IQ test preparedness and brain development. Both IQ and brain region volumes are experience dependent. For example, the high memory demands on London taxi drivers are reflected in bigger hippocampi (Maguire et al. Reference Maguire, Gadian, Johnsrude, Good, Ashburner, Frackowiak and Frith2000). In Western class-structured societies, the majority of developmental advantages and benefits are socially inherited, with huge psychological, as well as material, consequences affecting both IQ and brain development. For example:
1 Although the consequences of malnutrition for brain development are well known, risk factors are clustered in the lower class groups, primarily because their diet “provides cheap energy (sources) … lower in essential nutrients such as calcium, iron, magnesium, folate, and vitamin C than that of the higher socioeconomic groups” (James et al. Reference James, Nelson, Ralph and Leather1997, p. 1545).
2 Adverse substance experience has higher incidence among lower SES groups. Alcohol abuse during pregnancy interferes with trophic factors that regulate neurogenesis and cell survival (Goodlett et al. Reference Goodlett, Horn and Zhou2005). Nicotine dependence affects N-acetylaspartate (NAA) levels in the frontal cortex (Gür et al. Reference Gür, Schütz, Träber, Block, Maier and Schild2006). Similar findings have emerged from studies on drugs and exposure to environmental toxins, particularly lead (Grandjean & Landrigan Reference Grandjean and Landrigan2006).
3 Stress arising from the poor sense of control over circumstances, including financial and workplace insecurity, affects children and leaves “an indelible impression on brain structure and function” (Teicher Reference Teicher2002, p. 68; cf. Austin et al. Reference Austin, Leader and Reilly2005). In childhood and later life, increased stress reactivity can impair aspects of test performance, including self-confidence, attention, and memory (Richardson Reference Richardson2002).
4 Experiences in working class situations create poor cognitive self-efficacy beliefs. These are passed from parents to children (Bandura et al. Reference Bandura, Barbaranelli, Caprara and Pastorelli1996), influencing levels of aspiration and self-confidence, as well as anxiety and distractiveness in test situations (for review, see Richardson Reference Richardson2002). They can deter children from specific kinds of cognitive experience, with consequences for brain development. For example, activity in anterior cingulate regions is related to confidence in cognitive tasks (Fleck et al. Reference Fleck, Daselaar, Dobbins and Cabeza2006).
Note that many experience-dependent effects have been discovered to be (non-genetically) transgenerational through what Harper (Reference Harper2005) calls “this all but ignored” pathway of influence from parents to children and successive generations. That is, parental experiences are reflected in gene regulatory (epigenetic) aspects of children's and even grandchildren's development.
Such considerations suggest that “bigger” brain areas are not a cause of higher IQ. Rather, both are consequences of social experience. Accordingly, J&H's theory merely redescribes the class structure and social history of society and its unfortunate consequences. It seems to us that a host of variables, terms, and ecological factors need to be clarified before their model can be entertained further.
We are also concerned about the empirical latitude permitted in this area, overall, where, for example, a simple vocabulary test, or even a reading test (e.g., sect. 5.6, para. 7), can be taken as a measure of general intelligence. Likewise, we would question J&H's appeal to twin studies which have been consistently criticised for poor empirical standards (Richardson & Norgate Reference Richardson and Norgate2005).
Finally, we note that their whole model is based on a feedforward model of higher cerebral functions, which many would argue is outmoded. Looking for simple deterministic bases of intelligence variation does not reflect how the dynamics of the brain work (Freeman Reference Freeman2001). Having evolved to deal with changeable environments, cognitive systems must wring predictability from deep structures in the dynamic flow of information using massive reciprocal connections and cooperative processing between centres. This suggests quite different foundations for a theory of intelligent systems (Richardson Reference Richardson2006).
