The interface between genetics and psychology: lessons from developmental dyslexia

Abstract

Developmental dyslexia runs in families, and twin studies have confirmed that there is a substantial genetic contribution to poor reading. The way in which discoveries in molecular genetics are reported can be misleading, encouraging us to think that there are specific genes that might be used to screen for disorder. However, dyslexia is not a classic Mendelian disorder that is caused by a mutation in a single gene. Rather, like many other common disorders, it appears to involve combined effects of many genes and environmental factors, each of which has a small influence, possibly supplemented by rare variants that have larger effects but apply to only a minority of cases. Furthermore, to see clearer relationships between genotype and phenotype, we may need to move beyond the clinical category of dyslexia to look at underlying cognitive deficits that may be implicated in other neurodevelopmental disorders.

Keywords: dyslexia; genetics; neurodevelopmental disorders; specific language impairment; twins.

Figure 1.

Modified version of liability threshold model of disorder, based on Falconer [13]. Multiple genetic and environmental risk factors combine to create a normally distributed trait of liability to disorder. In Falconer's original work, high liability was associated with disorder, but here it is shown reversed, to make it more compatible with developmental disorders, where quantitative measures of ability can be seen as proxy indicators for liability. Disorder is diagnosed when (negative) liability falls below a cut-off level.

Figure 2.

Illustration of logic of the DeFries–Fulker method. Probands (MZ_p and DZ_p) are selected as those with scores below a cut-off (shaded area of normal distribution). The interest is then in how far mean scores of MZ and DZ co-twins (shown as black bars) regress to the mean. If the only factor leading to low scores is something specific to the individual (non-shared environment, which includes chance events), then the expectation is that the average co-twin score will be at the population mean, for both MZ and DZ twins. If the only causal factor is something shared in the environment of a twin pair, then both proband and co-twin should have the same mean scores. If genes are the only causal factor, then the expectation is that the mean for MZ co-twins will be the same as the probands’, because they share all their genes with the corresponding MZ proband, whereas the DZ co-twin mean will regress halfway to the population mean, because they share on average half their segregating genes. In practice, we assume all types of causal factor add together, and we estimate their relative contribution by considering the whole pattern of results (bottom panel). If the data are rescaled such that the proband mean is 1 and the population mean is zero, then the non-shared environment effect () is the difference between the MZ proband mean and the MZ co-twin mean, and the genetic effect () is twice the difference between MZ and DZ co-twins. The shared environment effect () can then be computed by subtraction as .

Figure 3.

Illustration of (a) additive and (b) interactive gene–environment effects at a single locus. The genotype is aa, aA or AA, corresponding to 0, 1 or 2 copies of the major allele.

Figure 4.

Mean scores on language, IQ and literacy-related measures by genotype. The sample is categorized according to whether they have a risk haplotype in DCDC2 and/or a risk haplotype in KIAA0319. Redrawn from figure 3a from corrected data provided by Powers et al. [47].