Jung & Haier (J&H) have done a nice job of collating the little data about the neural substrate of intelligence. It is useful and we do have to start somewhere; there are of course the usual reservations about drawing conclusions about the mind from images of the brain (Coltheart Reference Coltheart2006; Seron & Fias Reference Seron and Fias2006), but J&H have been suitably cautious, sieving the data down to groups of mutually inclusive spots.

The involvement of Brodmann area (BA) 39 is of interest. There is some very solid lesion data from aphasics to support involvement of this area in intelligence, but at least some of the circuitry therein may only have a peripheral role. This area is known to be significantly abnormal in dyslexic dysfunction (Rae et al. Reference Rae, Lee, Dixon, Blamire, Thompson, Styles, Talcott, Richardson and Stein1998; Rumsey et al. Reference Rumsey, Andeason, Zametkin, Aquino, King, Hamburger, Pikus, Rapoport and Cohen1992), although there is a poor relationship (Gustafson & Samuelsson Reference Gustafson and Samuelsson1999) between the presence of dyslexia and IQ. (Albert Einstein is a topical case in point, having both dyslexia and abnormal BA 39; see Kantha [Reference Kantha1992]; did Einstein's abnormal brain development allow the increased development of other circuits, such as those dealing with mathematical reasoning?)

The very intelligent brain is an interesting thing and may not necessarily be wired the same way as the average one. This suggests that a continuum may not exist in neuroanatomical intelligence substrates (i.e., a very intelligent brain may not just be a supercharged version of an average one); male brains being a generalized case in point, where the IQ distribution has significant wings. The gender distribution among those with extreme mathematical ability (or extremely dull brains) is highly skewed towards males (Benbow & Stanley Reference Benbow and Stanley1983). How does this happen? Why is the male brain more susceptible to disorders such as dyslexia (Rutter et al. Reference Rutter, Caspi, Fergusson, Horwood, Goodman, Maughan, Moffitt, Meltzer and Carroll2004)? A clue may yet come from epigenetics.

It seems like stating the obvious, and J&H have highlighted it in their article, but the inherent fuzziness of the concept makes “intelligence” difficult to define (Sternberg Reference Sternberg1990). J&H have identified intelligence as relating closely to g, itself a difficult to define concept. There is an argument to mount that heterogeneous functions in g (e.g., visuo-spatial vs. verbal) have different neural substrates (e.g., Benga Reference Benga2006), and the data may be skewed by the tendency of researchers to use tasks loaded on visuo-spatial ability in their functional imaging studies simply because they are easier to implement in many cases than language tasks (e.g., the verb-generation task study [Schmithorst & Holland Reference Schmithorst and Holland2006] does not include BA 39). From an evolutionary perspective, spatial abilities were acquired by the brain earlier than verbal abilities, which appear wired into the brain (especially the female brain) wherever there is room! Overlaid on this is the effect of gender, where differences in brain regions involved in intelligence have been repeatedly shown. Furthermore, there is strong evidence of possible epigenetic effects (Davies et al. Reference Davies, Isles and Wilkinson2005; Davies & Wilkinson Reference Davies and Wilkinson2006). In the case of Turner syndrome, where persons have only one sex chromosome (45,X), evidence has emerged that the superior and middle temporal lobes are greatly enlarged (Rae et al. Reference Rae, Joy, Harasty, Kemp, Kuan, Christodoulou, Cowell and Coltheart2004). The bigger the lobe, the worse the performance on temporal lobe-related tasks (Rae et al. Reference Rae, Joy, Harasty, Kemp, Kuan, Christodoulou, Cowell and Coltheart2004) (another example of bigger not necessarily being better). The enlargement is related to the parent of origin of the lone X chromosome, and other lobes, such as the parietal lobe which is relatively smaller, are similarly affected (Cutter et al. Reference Cutter, Daly, Robertson, Chitnis, van Amerlsvoort, Simmons, Ng, Williams, Shaw, Conway, Skuse, Collier, Craig and Murphy2006). These changes in brain volume are not insignificant and have the potential to influence attempts to measure changes in brain volume with intelligence. In males, where the parent of origin of the X chromosome is always maternal, this effect may be easier to deal with than in females, where X-inactivation is mosaicized and could have an infinite number of possible patterns. There are of course 23 other chromosomes, and not only gender effects but age effects, hormone effects, and the list goes on. It has recently been shown that just a single amino acid substitution in an enzyme can sharpen signal-to-noise characteristics in microcircuitry (Winterer et al. Reference Winterer, Musso, Vucurevic, Stoeter, Konrad, Seker, Gallinat, Dahmen and Weinberger2006).

Speaking of signaling efficiency (at the risk of overemphasizing my piece of the elephant), the emergent P-FIT model “elucidates the critical interaction between association cortices within parietal and frontal brain regions which, when effectively linked by white matter structures … underpins individual differences in reasoning competence in humans” (sect. 4, para. 2, my emphasis). Being effectively linked is obviously of importance, and structural and metabolic hot spots in functional imaging studies are not yet able to show how this linkage takes place. The size of the linking cabling shown up on the diffusion tensor image (Deary et al. Reference Deary, Bastin, Pattie, Clayden, Whalley, Starr and Wardlaw2006; Mabbott et al. Reference Mabbott, Noseworthy, Bouffet, Laughlin and Rockel2006) and connectivity analysis approaches may in future illuminate how, when, and where this happens (Rypma et al. Reference Rypma, Berger, Prabhakaran, Bly, Kimberg, Biswal and D'Esposito2006). How one brain communicates more efficiently than another, virtually identical brain, is a question central to any query about intelligence. Brains with more bioenergetic capacity appear to work better (Jung et al. Reference Jung, Brooks, Yeo, Chiulli, Weers and Sibbitt1999; Rae et al. Reference Rae, Scott, Lee, Simpson, Hines, Paul, Anderson, Karmiloff-Smith, Styles and Radda2003b); indeed ability can be improved by as much as one standard deviation at a task heavily loaded on g, (Raven's Advanced Progressive Matrices), simply by adding spare bioenergetic capacity (creatine) (Rae et al. Reference Rae, Digney, McEwan and Bates2003a).

The P-FIT model is a good start given our limited data, but may also be a dangerous thing if taken too seriously. Is there really a one-size-fits-all? Do ultra-bright brains really fit inside the box, or do individual differences become more important at this end of the scale?

The bottom line is that, beyond all the other confounding factors, molecular and systems biochemistry also needs to play a bigger role in research into cognition and inform our theorizing about the basis of intelligent behavior. Systems approaches need to be expanded to include all major variables, and study-group sizes need to be a lot bigger to filter the doubt from the dogma. Otherwise we shall never know what the elephant really looks like.