The position of a gene on a chromosome is a locus. Quantitative trait locus (QTL) refers to a region of DNA related to a trait like intelligence. There are often repeat copies of a gene at a locus and the number of copies sometimes can be related to normal or abnormal protein functions. DNA analysis generally breaks strands into fragments using any of several techniques. One technical breakthrough allowed the identification of small variations in the DNA sequence at any point on a strand where a base pair is changed or mutated. These errors are called single-nucleotide polymorphisms (SNPs). For example, if a sequence at a locus typically is GTCGAATTGGAATTGG, sometimes the first T can be a C in some individuals. This variation in the general sequence is a SNP. Most SNPs are non-functional, but some are related to diseases and possibly to traits like intelligence. One estimate is that there are about 30 million SNPs in a person’s DNA. SNPs can be compared between two groups, say defined by high or low IQ scores, in an effort to find segments of DNA, perhaps individual genes, that differentiate the groups. Early studies sampled thousands of SNPs. Now an entire person’s genome can be sequenced and SNPs assessed in genome-wide association studies (GWAS). Such studies generate enormous data sets and the field of genomic informatics has develo ped statistical methods for sorting through all the possible combinations with the ultimate goal of identifying specific genes for diseases, medical conditions, and the myriad of inherited traits. There are now bioinformatic efforts to use cloud computing to accumulate, organize, and analyze Big Data sets of genetic information acquired from DNA analyses.
Proteomics is the study of proteins and how they work. It is now possible to test gene expression for thousands of proteins and their varieties simultaneously, often using a small DNA sample on microarrays with different reactive agents. All together, the ever-evolving techniques and methods of molecular genetics provide detailed assessments deep into the neurobiology and neurochemistry of neurons, synapses, and brain development of function and structure. Despite the complexities and overwhelming amounts of DNA data, in my view, the challenges for understanding intelligence at this level are made finite by the DNA technologies. In fact, given the psychometric problems noted in Chapter 1, progress understanding intelligence on the behavioral level might prove more difficult than on the molecular level.
2.6 Seven Recent Noteworthy Studies of Molecular Genetic Progress
The benefits of a consortium approach are nicely illustrated in a study co-authored by 59 investigators pooling data sets from around the world (Rietveld et al., 2014). This study actually was a combined effort from two consortia: SSGAC (Social Science Genetic Association Consortium) and CHIC (Childhood Intelligence Consortium). These researchers used a conceptually simple and clever two-stage process that began with a sample of 106,736 individuals and ended with an indirect replication in an independent sample of 24,189 individuals. In the first sample, millions of SNPs were assessed (see Textbox 2.2) for each person’s DNA and 69 were related to education attainment level (years of education). Education level is highly correlated to intelligence. In the second sample, these 69 SNPs were tested for any associations with a g-score derived from cognitive test scores. Although not every person completed the same tests, g-scores from different batteries are highly correlated (even over .95) if the test batteries and the sample are sufficiently diverse (Johnson et al., 2008b). Several advanced statistical analyses revealed four promising genes of interest related to very small amounts of variance in cognitive performance. Interestingly, these genes (KNCMA1, NRXN1, POU2F3, SCRT; I am not responsible for how genes are named) are known to influence a glutamate neurotransmitter pathway related to brain plasticity and learning and memory. The pathway involves the NMDA receptor, glutamate binding, and synaptic changes. Despite the small amount of intelligence variance associated with these genes, this study demonstrates the statistical reality that large samples are necessary to find small effects. Such findings also provide hints about molecular mechanisms that may be associated with intelligence that could be the basis for hypotheses about the salient neurobiology.
About the same time, the second study (Hill et al., 2014) used genome-wide analyses in 3,511 individuals to investigate small effects of 1,461 individual genes on intelligence by finding associations with cognitive ability in aggregated networks of functionally related genes. They started with a specific hypothesis that focused on genes related to post-synaptic functioning. After replication in independent samples, proteins related to the NMDA receptor were associated specifically with fluid intelligence. Other aspects of post-synaptic functioning were not related to variation in any other cognitive abilities. NMDA was also implicated indirectly in the study of Rietveld et al., but these post-synaptic findings tied genetic variation of specific proteins to individual differences in fluid intelligence. The key protein was guanylate kinase (MAGUK), fundamentally important for converting neuronal action potentials into biological signals that underlie information processing throughout the brain. This study provides additional hints about the neurobiology of intelligence.
Whereas the Rietveld et al. study used a very large atheoretical shot-gun approach to find needles in a haystack and the Hill et al. study focused on a network of functionally related genes, another group of researchers used a different strategy that focused on a specific gene and its effects on intelligence after traumatic brain injury (TBI) (Barbey et al., 2014). There is a neurochemical called brain-derived neurotrophic factor (BDNF) that promotes and regulates well-functioning synapses. BDNF is related to cognitive functioning in healthy people, especially to aspects of memory and to impaired cognition in Alzheimer’s disease and other brain disorders. Val66Met, a gene associated with BDNF, also is implicated in neural repair mechanisms that stimulate neuro-regeneration in the prefrontal cortex after recovery from TBI. Is BDNF related to intelligence? After TBI to the frontal lobes, some patients show persistent deficits in g-loaded tasks, whereas in other patients there is a preservation of g-loaded task performance. The genetic basis for BDNF is the Val66Met polymorphism (see Textbox 2.2) that has two main variations, Val/Met and Val/Val. The question for this study was whether either of these variants was related to the preservation of intelligence after TBI.
Unfortunately, there are a large number of TBI cases. Many are treated in Veterans Administration (VA) hospitals. Participants in this study came from a group of 171 male veterans who suffered penetrating head injuries during the Vietnam War. For 151 of these individuals, the sites of brain lesions were located by CT scans and confirmed in the frontal lobes. Each participant completed 14 subscales of the WAIS III and each person also had completed the battery of tests in the Armed Forces Qualification Test (AFQT) when they entered the military, prior to the TBI. Both these tests allowed the computation of a g-factor score along with other subfactors. Two groups were defined based on genotyping: Val/Met (n = 59) and Val/Val (n = 97) and the intelligence factor scores were compared between these two groups with sophisticated psychometric analyses.
The results were rather striking. Scores derived from the AFQT did not differ between the groups. In other words, prior to the TBI the veterans’ genotype (Val/Met or Val/Val) had no impact on general cognitive ability. However, there was a substantial difference following TBI. The Val/Val group showed average diminished factor scores for g and other primary factors including verbal comprehension, perceptual organization, working memory, and processing speed. The average score differences were relatively large at about half a standard deviation. The authors concluded that having the Val/Val genotype was associated with cognitive susceptibility to TBI, whereas the Val/Met genotype may help preserve cognitive functioning following TBI. These results may have implications for cognitive rehabilitation strategies that might be more effective than others, although there is a paucity of such research at present. The results also tie variations in the BDNF gene to intelligence and demonstrate some progress toward identifying specific gene variants related to intelligence. Such data can help generate hypotheses abo
ut the step-by-step cascade of neurochemical events at the molecular level that lead from the genetics of BDNF expression to explaining a small amount of variance among individuals in intelligence. It is likely that there are many steps in the cascade and multiple complex interactions with other genetic or biological factors, some of which might occur in an epigenetic context. And BDNF is only one of many factors probably involved.
The fourth study demonstrating progress takes another approach (Davis et al., 2015). These researchers also focused on one molecular factor, a protein called DUF1220 that is associated with brain size and brain evolution. DUF1220 has two main subtypes, CON1 and CON2. Many gene sequences have multiple copies in a person’s DNA and the number of copies can be related to diseases and other traits. In this study, not only was the number of copies of CON2 associated with IQ scores, the association was linear. That is, the more copies of CON2, the higher the IQ score. Brain size, assessed by MRI, also was correlated to IQ score, especially for the surface area of the temporal cortex bilaterally, and right frontal surface area was related to increased dosage of CON1 and CON2. These findings came from a sample of 600 North American young people and were replicated in a smaller sample of 75 individuals living in New Zealand. Although both samples are quite small compared to the previous studies just summarized, the linear nature of the CON2 copy/IQ finding is intriguing, especially because it was strongest in males 6–11 years old. There are a number of reasons to be cautious about this finding, as acknowledged by the authors, and it is too early to accept it at face value. Nonetheless, it represents another example of how the search for intelligence genes is pushed forward with a priori hypotheses about specific genetic factors.
The fifth study is from the CHIC. They reported a GWAS of intelligence in children aged 6–18 years old with a combined discovery cohort total of 12,441 and a replication cohort total of 5,548 (Benyamin et al., 2014). No single SNP was associated with intelligence, but the aggregate of common SNPs accounted for 22%–46% of variation in intelligence in the three largest cohorts. The FNBP1L gene was associated with intelligence, accounting for small amounts of variance in three separate replication cohorts (1.2%, 3.5%, .5%, respectively). Despite the large sample, the authors concluded that even larger samples might be necessary to detect individual SNPs with genome-wide significance.
The sixth study comes from another multisite consortium (CHARGE: Cohorts for Heart and Aging Research in Genomic Epidemiology) of 31 cohorts (N = 53,949). They reported a meta-analysis based on a GWAS of middle-aged and older adults who had completed a battery of four cognitive tests (Davies et al., 2015). This is the largest such study of general cognitive ability to date. Across all the samples, 13 SNPs were associated with general cognitive ability, together accounting for 29% and 28% of the variance in two of the largest samples, respectively. Three genomic regions were associated with these SNPs with special focus on the HMGN1 region. Four genes previously associated with Alzheimer’s disease also were associated with general cognitive ability (TOMM40, APOE, ABCG1, MEF2C). Consistent with the polygenetic model of inheritance, these genes individually accounted for small proportions of variance. These researchers also conclude that even larger samples will be required to identify more genome-wide associations. No one yet knows how many genes may contribute to variations in intelligence, but the very existence of these multicenter collaborations is a giant step toward finding out. In fact, just after this book went into production, an elaborate collaborative study identified two networks of genes (1,148 genes in one network and 150 genes in the other) that were related to general cognition (Johnson et al., 2016). Many of these genes were related to specific synaptic functions that potentially could be manipulated to influence intelligence. I do not have space to elaborate more details of this landmark study, but the hunt for intelligence genes has taken another major step forward.
The seventh encouraging example of progress is from China. In my view, it also is a landmark exercise. It reports a broad systems biology approach (Zhao et al., 2014) that is designed to elucidate complex regulation and interactions and generate hypotheses about the mechanisms that drive them. Whereas Chabris and colleagues had started with 12 candidate genes from previous studies and failed to replicate any of them, these researchers selected 158 genes that had been associated with IQ scores. They mapped the locations of these genes on chromosomes and found some clustering in seven regions of chromosome 7 and the X-chromosome. Many of these genes were known to be involved in various neural mechanisms and pathways. Using a type of network analysis, “IQ-related pathways” were constructed. These pathways primarily involved dopamine and norepinephrine, neurotransmitters involved in many brain functions. The details of this analysis are far beyond our intentions in this chapter, but this report illustrates how molecular genetics can generate testable hypotheses about specific neural mechanisms related to intelligence, and how those mechanisms might be tweaked by drugs or other means. Work like this fuels my optimism that the genetic basis for intelligence is not a retreat to determinism and immutability. Rather the opposite: the genetic basis, once understood, can lead to the remarkable ability to treat or prevent brain disorders that result in low IQ and to the Holy Grail of increasing intelligence across the whole range, as we will discuss in Chapter 5.
Studies like these seven (plus the Johnson et al. study) are the reason this chapter has not dwelt on older (and a few current) criticisms about whether genes are important for intelligence. Although the full role of genes is not yet known, the evidence for major genetic involvement in intelligence is overwhelming. No one ever believed that understanding intelligence on the molecular level would be simple, but the studies and their complex analyses summarized here show that the challenge is not impossible.
On a final note, genetic studies are logistically complex and expensive, especially when large samples are involved. DNA sequencing machines alone, for example, cost about $1–2 million each. Reportedly, in 2012 a single research institute in China, the Behavioral Genetics Institute, had 128 of them, along with super computers. Finding intelligence genes is a high priority. This one institute has over 4,000 scientists and technicians working there and a poster on the wall reportedly says: “Genes build the future.” Consider the race to find intelligence genes and how they work. At the end of the twentieth century, Plomin (1999) stated, “The most far-reaching implications for science, and perhaps for society, will come from identifying genes responsible for the heritability of g …” On one hand, China has substantial investment in this hunt, and on the other hand, a majority of members currently in the US Congress apparently do not believe in evolution. Seriously.
All the studies outlined in this chapter illustrate how quantitative genetic research strategies and sophisticated DNA analyses are being used to establish the genetic basis of intelligence and search for specific genes and how they work. There are now several worldwide consortia working on this effort that add a third methodological element, quantitative neuroimaging to measure brain structure and function. This combination of three research elements targets the identification of genes that influence brain characteristics related to intelligence. In my view, these studies represent a new phase in the search for intelligence genes and what they do. These are exciting findings and we will review them in Chapter 4. First, however, to understand the full impact of the newest DNA studies using twin research designs that target the brain, we now introduce the third element, neuroimaging, in the next chapter.
Chapter 2 Summary
Sir Cyril Burt and Professor Arthur Jensen were early advocates of the importance of a genetic role in intelligence, but their views were attacked and widely rejected.
Modern quantitative genetic studies overwhelmingly support a major role for genes in explaining the variance of intelligence test scores among individuals.
The same studies indicate that environmental factors play a role in early childhood, especially non-shared factors, but this role diminishes almost entirely by
early teen years.
The weight of evidence from modern studies of intensive compensatory education, now rebranded as early childhood education, still fail to find lasting effects on IQ scores.
Progress in the search for specific genes involved in intelligence has been slow and disappointing, leading to the conclusion that many genes, each having a small effect, must be involved.
Advanced DNA technologies applied to molecular genetic studies are beginning to identify intelligence-related genes and how they might work on a neurobiological level.
Review Questions
1. Why are genetic explanations of behavior so controversial?
2. What were the immediate and long-term impacts of Jensen’s 1969 article?
3. What is the most compelling evidence from quantitative genetic studies that genes are involved in intelligence differences among people?
4. How does the influence (or relative contribution) of genetic and environmental factors on variation in intelligence change across development?
5. Why may there be an advantage to intelligence involving many genes of small influence?
6. What is an example of a recent finding regarding intelligence from a specific molecular genetics study?
Further Reading
Human Intelligence (Hunt, 2011). This is a thorough textbook that covers all aspects of intelligence written by a pioneer of intelligence research. Chapter 8 has an excellent discussion of heritability estimation and other genetic issues.
The Neuroscience of Intelligence Page 10
