At minimum, in my view, reasoning research reports should include “intelligence” as a key word for indexing, and the relationship of reasoning tests to intelligence should be acknowledged in the discussion of results that show brain/reasoning relationships. Intelligence reports should do the same for reasoning. Moreover, there is a growing recognition that the results of cognitive/imaging experiments might change dramatically depending on whether the participants are selected for high or average IQ scores (Graham et al., 2010; Preusse et al., 2011; Perfetti et al., 2009). New imaging technologies like MEG may allow even more detailed analysis of information flow during reasoning/problem-solving, especially if levels of intelligence are included in the research design of studies. More collaboration between reasoning researchers, with cognitive expertise, and intelligence researchers, with psychometric expertise, is the best way to integrate these two rich empirical traditions.
4.5 Common Genes for Brain Structure and Intelligence
In Chapter 2 we discussed quantitative and molecular genetic findings related to intelligence, but we deferred presentation of genetic studies of intelligence that included neuroimaging. Now that neuroimaging has been introduced as it is used in studies of intelligence, let us consider the powerful combination of genetic and neuroimaging methods to study intelligence.
As the hunt for specific genes continues, the newest and most compelling quantitative genetic twin studies of intelligence go beyond the simple question of whether there is a genetic component or not. They focus on what the genetic component may do in the brain, even without knowing any specific genes. Paul Thompson and colleagues reported the first twin study that used MRI to assess and map the heritability of cortical gray matter volume and relate it to intelligence (Thompson et al., 2001). They studied a small sample of 10 MZ pairs and 10 DZ pairs. The estimated genetic contribution for gray matter volume based on similarity between twin pairs varied across brain regions, a somewhat surprising finding at the time. The highest genetic contributions were in cortical areas of the frontal and parietal lobes. Moreover, the correlation between IQ scores and gray matter in the frontal lobes was statistically significant. Based on the unique combination of IQ testing and neuroimaging in twins, this study provided unique evidence of what had been suspected by many researchers: individual differences in intelligence were due, at least in part, to the genetics of brain structure, specifically cortical gray matter volume. Despite the small sample, the importance of this finding was underscored by its publication in the prominent journal Nature Neuroscience. An accompanying commentary noted that the high heritability implied that gray matter development apparently was less sensitive to experience than might be expected (Plomin & Kosslyn, 2001).
Researchers in the Netherlands have published a compelling series of findings that draw on larger samples of twins. We introduced some of their findings in Chapter 2 concerning shared and non-shared environment influences on intelligence. Here we will summarize more of their important MRI findings that indicate there are common genes for intelligence and brain structure. In 2002, they published findings about gray and white matter heritability and intelligence, also in Nature Neuroscience (Posthuma et al., 2002). Heritability estimates were high for both and whole-brain white matter was slightly more heritable than whole-brain gray matter. Moreover, by comparing MZ and DZ twins, the authors found that the correlation between gray matter volume and general intelligence was due entirely to genetic factors. No variance was attributable to shared or non-shared environmental factors. Later, they expanded their sample of twins to increase statistical power and examine correlations between gray, white, and cerebellar volumes and different intelligence factors (Posthuma et al., 2003a). All three volumes were correlated with working memory capacity and related to a common genetic basis. Processing speed was genetically related to white matter volume. Perceptual organization was related both genetically and environmentally to cerebellar volume. Verbal comprehension was not related to any of the three volumes. This group also showed that gray and white matter in specific brain areas had a common genetic basis with IQ (Hulshoff-Pol et al., 2006).
Similar results from the Netherlands were found in 112 nine-year-old twin pairs (van Leeuwen et al., 2008), indicating early genetic influences on intelligence in the maturing brain. A longitudinal MRI study of adult twins examined changes in cortical thickness over a 5-year period and found the degree of cortical change (also called plasticity) had a strong genetic basis (Brans et al., 2010). Change was related to IQ. Higher IQ scores were related to cortical thinning over time in the frontal lobes and to thickening in the parahippocampus, an important brain structure in the temporal lobes involved with memory. The actual cortical changes were a fraction of a millimeter, but even small amounts of brain tissue can be important. In the last chapter, we noted an earlier study reported cortical thinning during early childhood was related to higher IQ scores (Shaw et al., 2006). This adult twin study of IQ and brain plasticity in frontal lobes and the parahippocampus concluded that both variables might have some common genetic basis. One novel finding concerned the subcortical parahippocampus, unusual because most studies have focused on the cortex. Another MRI study drawing on the Netherlands twin data examined whether the volume of several subcortical areas was related to IQ. Only the volume of the thalamus, an important hub of brain circuit connectivity, was related to IQ and a common genetic component was implicated for both (Bohlken et al., 2014). Although cortical thickness has been associated with IQ in several studies, there also is an indication that cortical surface area may show even stronger associations with cognitive abilities and related genes based on a study of 515 middle-aged twins that compared both thickness and surface area measures (Vuoksimaa et al., 2015). This field is quite dynamic as new approaches to data analysis evolve and extend previous findings with increased accuracy. They contribute to the weight of evidence regarding genes and brain structure with mostly consistent findings for intelligence.
White matter integrity is a particular focus of intelligence research given its heritability and the DTI results we have noted here and in the last chapter (see also DTI Animation 4.3 and Figure 4.2). The Thompson team used DTI in a sample of 92 Australian twins (23 MZ and 23 DZ pairs) to quantify a measure of white matter fiber integrity called fractional anisotropy (FA) (see Chapter 3). They mapped the heritability of FA throughout the cortex and found the highest values in frontal and parietal lobes (bilaterally), and the left hemisphere occipital lobe (Chiang et al., 2009). IQ scores (FSIQ, PIQ, VIQ) were correlated to specific fiber tracts (higher integrity was associated with higher IQ) and these correlations were also mapped. Based on cross-trait mapping, they concluded that common genetic factors mediated the correlations between IQ and FA, suggesting a common physiological mechanism for both. When the data are displayed as maps, the results are compelling. Figure 4.11 shows the distribution of FA variance for genetic, shared, and non-shared environment. Figure 4.12 shows the cross-trait mapping for FSIQ. In 2011, the Chiang group expanded these findings. Based on a larger sample of 705 twins and their non-twin siblings, they examined effects of age, sex, social–economic status (SES) and IQ on the hereditability of the FA measure (Chiang et al., 2011b). There were complex interactions for various brain regions, but in general, genetic influence was greater in adolescents compared to adults, greater in males than females, greater in those with high SES, and in those with higher IQ.
Figure 4.11 Maps of genetic and environmental influences on white matter integrity (measured by fractional anisotropy, FA). Each row shows a different axial brain view (horizontal slice). Red/yellow shows strongest results. The left column shows the significance of genetic influences. Other columns show the strength of the FA measure for genetic, shared, and non-shared environment, respectively.
Adapted with permission, Chiang et al. (2009, their figure 4).
Figure 4.12 Overlap of common genetic factors on FA (fractional anisotropy) and FSIQ (left column) based on cross-trait analysis
of areas shown in Figure 4.11. The right column shows statistical significance. Each row shows a different axial (horizontal slice) brain slice.
Adapted with permission, Chiang et al. (2009, their figure 7).
Schmithorst and colleagues, as noted in the previous chapter, had reported age and sex differences in the earliest DTI studies of intelligence. A comprehensive DTI study of 1,070 children aged 6–10 years from the Netherlands supported the Schmithorst findings and further reported FA correlations with non-verbal intelligence and with visuospatial ability (Muetzel et al., 2015). A three-year longitudinal study of adolescent twins and their siblings used DTI and graph analysis to map the heritability of white matter fiber integrity (Koenis et al., 2015). The efficiency of white matter networks, assessed with FA, was highly heritable, with genetic influences accounting for as much as 74% of variance. Moreover, there was a provocative finding related to intelligence. For the subgroup of individuals who showed a change in IQ scores over the three-year period, individual increases in scores were correlated to increases in local network efficiency in the frontal and temporal lobe areas. These findings are shown in Figure 4.13. The authors speculate that finding ways to promote the efficiency of white matter networks may optimize teenage cognitive performance. Most parents of teenagers will try anything. So will most teenagers.
Figure 4.13 Correlations between three-year change in IQ score and change in local brain efficiency measured with FA. The largest purple spheres show the strongest IQ/efficiency change correlations.
Adapted with permission, Koenis et al. (2015).
There are now so many studies in this area that it is easy to be confused. Here’s the short story. Genes influence brain networks and intelligence. Until specific genes and their expression are identified, we cannot distinguish directly whether genes influence brain morphometry, which then influences intelligence, or whether genes influence intelligence, which then influences brain morphometry. It is also possible that many genes influence both brain morphometry and intelligence (pleiotropy) and only some of them are common to both.
As you see so far in this chapter, the quantitative analysis of neuroimaging has become quite sophisticated, with complex multivariate statistical methods. The quantitative analysis of genetic data is also quite complex. Teams of researchers that include mathematicians in addition to imaging experts and genetic experts now carry out this research. Intelligence is increasingly a focus of interest and experts in intelligence research are becoming part of such teams. Summarizing results in this chapter from the combination of the neuroimaging and the quantitative genetic domains without oversimplifying is a challenge, but at minimum, the progress and excitement in the field should be clear to all readers. We are light years past earlier controversies about whether there is a role for genetics for understanding individual differences in intelligence. The challenge in the next section is explaining the complexity of neuroimaging analysis combined with the even greater complexity of molecular genetic analysis. The details may be difficult and the gene nomenclature of letter/number combinations seems irrational. But here’s the main point. The results are exciting and speak to the optimism that the hunt for intelligence genes is gaining ground.
4.6 Brain Imaging and Molecular Genetics
There are now many studies that combine neuroimaging and genetic analyses, so we must choose which ones best illustrate progress. Let’s continue with more in the sequence of papers by Chiang and colleagues. In a DTI analysis of 455 twins and their non-twin siblings, they found an association between white matter integrity and the ValMet polymorphism related to BDNF, the brain growth factor involved in normal neuron function. They suggested that BDNF might be related to some intellectual performance indirectly by modulating white matter development in some fiber tracts (Chiang et al., 2011a). In another paper, this group pursued the idea that the genetics of white matter brain wiring is fundamental to intelligence (Chiang et al., 2012). They focused their imaging efforts in a novel way to help identify specific genes related to brain connectivity and intelligence. As we noted earlier in this chapter, in 2009 they had reported that there were genes in common for the integrity of white matter tracts and intelligence based on cross-trait maps of the respective heritability of each. They expanded this approach using DTI and DNA data from a sample of 472 twins and their non-twin siblings. The basic idea was to use statistical methods of clustering many variables based on their similarity. First, thousands of points within white matter fibers were clustered to find brain systems with common genetic determination. The white matter measure was FA, as described previously. Then they used DNA in a genome-wide scan and network analyses to identify a network of genes that was related to white matter integrity in major tracts. FA in some hubs in the white matter network was related to IQ scores (their figure 9). The results of this analysis are complex and include 14 specific genes listed in their table 5, along with what was known about each one’s function in 2012. We include their table here (Table 4.1) to illustrate how intelligence potentially can be tied to brain function on the molecular level, although none of the entries in the table have been replicated. This kind of study illustrates both the complexity of understanding how genes function and a major neuroscience direction for future intelligence research. There is a long road between observations like these and making any practical use of them. However, if replicated, identifying genes related to intelligence and how they function can point to potential mechanisms for enhancing intellectual performance if the cascade of genetic influences on functional molecular events can be manipulated at the right stage of brain development. This includes manipulating more general genetic effects on brain structures like white matter integrity that may influence intelligence indirectly (Kohannim et al., 2012a, 2012b). We will discuss more about enhancement in Chapter 5.
Table 4.1 Genes possibly related to intelligence, their chromosome number, and their suspected function. Reprinted with permission from Chiang et al. (2012, their table 5).
ACNA1C Calcium channel, voltage-dependent, l type, alpha 1C subunit 12 Voltage-sensitive calcium channels
CTBP2 C-terminal binding protein 2 10 Encoding a major component of synaptic ribbons
DDHD1 DDHD domain containing 1 14 A probable phospholipase that hydrolyzes phosphatidic acid
DMD Dystrophin X Anchoring the extracellular matrix to the cytoskeleton
FAIM2 Fas apoptotic inhibitory molecule 2 12 Protecting cells uniquely from Fas-induced apoptosis
FHAD1a Forkhead-associated (FHA) phosphopeptide binding domain 1 1
GRM8 Glutamate receptor, metabotropic 8 1 Encoding receptors for glutamate
HADH Hydroxyacyl-CoA dehydrogenase 4 Mitochondrial beta-oxidation of short-chain fatty acids
KAZN Kazrin 1 Cell adhesion and cytoskeletal organization
LPIN2 Lipin 2 18 Controlling the metabolism of fatty acids
OPCML Opioid-binding protein/cell adhesion molecule-like 11 Binding opioids in the presence of acidic lipids; probably involved in cell contact
SCN3A Sodium channel, voltage-gated, type III, alpha subunit 2 Mediating the voltage-dependent sodium ion permeability of excitable membranes
SYN3 Synapsin III 22 May be involved in the regulation of neurotransmitter release and synaptogenesis
SYT17a Synaptotagmin XVII 16
Information in this table is derived from the NCBI gene database (www.ncbi.nlm.nih.gov/gene) and the GeneCards database from the Weizmann Institute of Science (www.genecards.org).
a Information about the function of this gene is not available in either of the two databases.
In the rapid evolution of molecular genetic studies, the next major advance is the aggregation of huge samples in worldwide multicenter collaborations formed to investigate brain diseases and normal cognition. We noted some of these in Section 2.6. These groups, and the resulting publications, are a triumph of logistical, political, and scientific achievement. One of the largest is named ENIGMA (Enhancing Neuro Imaging Genetics through
Meta-Analysis). One of their papers included a finding that related individual differences in intelligence to a specific variant in a gene called HMGA2 that is related to brain size (Stein et al., 2012). Their discovery and replication samples included thousands of participants who had completed neuroimaging in addition to cognitive and DNA testing. As in other studies, the finding explained only a tiny fraction of variance in intelligence, but it illustrates a definitive victory for finding a gene needle in a haystack of DNA.
The Neuroscience of Intelligence Page 18