If intelligence is partly determined by our genes, how does brain development relate to IQ? An attempt to answer this question measures the size of the outer layer of the brain, the cortex, with surprising results.
Shaw and colleagues (page 676 of this issue)1 have investigated whether there is a relationship between intelligence and physical dimensions of the brain. Specifically, they measure the thickness of the cortex; the complex computations carried out by the brain depend on the firing of the cortical cells. The authors' results indicate that intelligence can be related to how the cortex changes during development.
Rather than making structural measurements in post-mortem brains, Shaw and colleagues used magnetic resonance imaging (MRI) in living subjects. This allowed the authors to obtain images from people whose IQ could also be tested so as to look for correlations between the two measures. Moreover, detect-ing anatomical features associated with an individual's intelligence requires a large pool of subjects, because any effects may be small and could be missed if the sample size is inadequate. The use of imaging, rather than post-mortem measurements, allows data to be gathered from a sufficient number of individuals.
The authors scanned 307 children from the age of six years and followed them through adolescence with further scans. For each child, the authors estimated intelligence using subtests of the Wechsler Intelligence Scales — the most commonly used IQ tests. An alternative approach would have been to look at a cross-sectional sample of children and adolescents of different ages, scanned only once each. But, as the authors note, such methods are open to many objections: for example, teaching practices may change over time, which would affect the IQ scores.
Shaw and colleagues find no significant correlation between cortical thickness and intelligence in their data from young children. Yet they cite a study of adults by McDaniel2 that reports a modest correlation of 0.3 between intelligence and the total volume of the brain. The reason for the different results could be that the relevant factor is the total area of the cortex rather than its thickness, but it turns out that this is probably not the case. As the children were followed up, the nature of the relationship changed. In young children, the correlation tended to be negative, but in late childhood, around the age of ten, it was positive.
The authors illustrate this point by plotting continuous curves of cortical thickness for subjects from the ages of seven to nineteen, dividing the sample into three groups on the basis of their scores in the IQ tests: those of ‘superior’, ‘high’ and ‘average’ intelligence. IQ measures are normalized to the age group, and should in theory remain the same as the children age. Figure 2 on page 677 shows the curves for cortical thickness in brain areas that show different developmental patterns according to intelligence. Children in the group with superior intelligence have a thinner cortex in these areas in early childhood, but cortical thickness increases sharply until age eleven compared with the other groups, before decreasing through adolescence. The authors note that those of superior intelligence show a prolonged period of prefrontal cortical gain and the most rapid rate of change.
These differential changes do not occur in all cortical areas. The most notable positive correlations with IQ in late childhood occur in the prefrontal cortex. This region lies at the top of the information-processing hierarchy, receiving highly processed information from all five senses3. The brain areas showing the biggest difference in the shape of the growth curve between those with superior intelligence and the other groups lie in the lateral and medial frontal gyri. But are these the areas that are most active when subjects perform IQ tests? This aspect can be assessed by functional MRI, which provides an indirect measure of the increase in arterial blood flow to areas in which cellular activity is increased. Previously, subjects have been scanned while taking non-verbal tests that measure IQ, and increased activity has been found in the lateral and medial prefrontal cortex — regions that are among those highlighted by Shaw and colleagues' developmental measures4,5. Furthermore, individual differences in IQ are correlated with the amplitude of the functional MRI signal in the lateral prefrontal cortex6.
We know that variations in general intelligence, or g, among people depend to a great extent on genetic differences7. So, if g is highly heritable and the increase in the thickness of the prefrontal cortex is related to g, it is tempting to assume that this developmental change in brain structure is determined by a person's genes. But one should be very wary of such a conclusion. The body's development is intimately linked to interactions with its environment. For example, in a classic experiment, Rosenzweig and Bennett8 showed that the thickness of the cortex in adult rats is affected by the degree to which the animals' early environment is enriched in terms of activities. Even in human adults, structural changes can be seen in the cortical grey matter as a result of practice9. Thus, it could be that people with superior intelligence also live in a richer social and linguistic environment, and that it is this that accounts for the sharp increase in the thickness of their prefrontal cortex in late childhood. However, Thompson and colleagues10 previously looked for genetic influences on brain structure by comparing the cortical thickness of pairs of identical and non-identical twins. They found that some regions, including the frontal cortex, are, to use their words, under “tight genetic control”.
Shaw and colleagues speculate that differences in the shape of the growth curves of cortical thickness could be influenced by various factors. These include the number of neurons that collect in the subplate under the cortex during late fetal development, the development of the myelin sheath that insulates the fibres of the neurons, and the selective elimination at puberty of neuronal connections that are not useful. Testing these hypotheses will require animal experiments that measure cellular development. Studies in animals have the advantage that the relative influence of genetics and experience can be disentangled, and so should provide a clearer picture of how intellectual ability is affected by the factors that underpin cortical development.