A Quantitative Genetic Analysis of Cognitive Abilities during the Second Half of the Life Span

Little is known about the importance of genetic effects on individual differences in cognitive abilities late in life. We present the first report from the Swedish Adoption/Twin Study of Aging (SATS A) for cognitive data, including general cognitive ability and 13 tests of specific cognitive abilities. The adoption/twin design consists of identical twins separated at an early age and reared apart (46 pairs), identical twins reared together (67 pairs), fraternal twins reared apart (100 pairs), and fraternal twins reared together (89 pairs); average age was 65 years. Heritability of general cognitive ability in these twins was much higher (about 80%) than estimates typically found earlier in life (about 50%). Consistent with the literature, heritabilities of specific cognitive abilities were lower than the heritability of general cognitive ability but nonetheless substantial. Average herit abilities for verbal, spatial, perceptual speed, and memory tests were, respectively, 58%, 46%, 58%, and 38%. Evidence of a genetic influence on cognitive function has accumulated rapidly in the last few decades. There is still some disagreement about the magnitude of the genetic effect, but it appears to be very substantial indeed: Results from dozens of twin and adoption studies converge on the conclusion that about half of the variance of general cognitive ability (IQ) can be accounted for by genetic factors (Bouchard & McGue, 1981 ; Chipuer, Rovine, & Plomin, 1990; Loehlin, 1989). Although less is known about the genetics of specific cognitive abilities such as verbal ability, spatial ability, perceptual speed, and memory two trends have emerged: Specific cognitive abilities appear to yield estimates of genetic influence that are significant but lower than estimates for IQ, and memory tests appear to be subject to less genetic influence than other tests of specific cognitive abilities (Plomin, 1988). The conclusion that significant and substantial genetic influence contributes to the variability of scores on cognitive tests is much nearer the beginning than the end of the story of genetics and cognitive abilities (Plomin & Neiderhiser, in press). One gap in our understanding about genetic influence on cognitive abilities is the virtual absence of information for the entire second half of the life course. In this article, we present the first report of a quantitative genetic investigation of cognitive abilities in the second half of the life span. The study, the Swedish Adoption/Twin Study of Aging (SATS A), uses the powerful combination of adoption and twin designs in that it includes identical and fraternal twins reared apart in addition to the traditional twin design involving twins reared together. One of the most basic developmental questions is whether the relative importance of genetic variance (heritability) differs with age. The proportional importance of genetic and environmental factors could change throughout the life span for a number of reasons. Although we are born with a full complement of genes, not all are operating at any one time. Genes may be 4 'turned on" only at specific stages during the life span, while others may be inactivated. Temporal genes may be involved in the timing of specific age-related events (Fairer, 1987; Paigen, 1980). Thus, changes in the activity of genes could result in differences in genetic variance at various points during the life span. It is perhaps more obvious that environmental influences vary throughout the life span. Some life span developmental theorists predict that total variance increases as one accumulates experiences. Thus, it would be reasonable to assume that heritability decreases during early development as experiences accumulate or that environmental influence becomes increasingly important later in life (Horn & Cattell, 1966), with the accumulation of wounds from life's ' 'slings and arrows of outrageous fortune." The net effect on heritability estimates (i.e., the relative importance of genetic variance) will, of course, depend on whether total phenotypic variance is changing, and whether the effects of genetic or environmental factors are accumulated over time or are occasion-specific. If a single set of genes operates throughout development and information from the environment is " stored over time," heritability will decrease over time (Eaves, Long, & Heath, 1986). However, if environmental effects are occasion-specific, heritability will increase with age. Although early twin studies focused on age differences in twin resemblance for cognitive ability (Merriman, 1924; Thorndike, 1905), interest in developmental issues disappeared until the past decade (Plomin, 1986). Contrary to expectations of proportionally increasing environmental variance during development, research in early childhood suggests a steady increase in the heritability of IQ scores, at least up until the early school years (Fulker, DeFries, & Plomin, 1988; McCartney, Harris, & Bernieri, 1990). Whether heritability of general and specific cognitive abilities changes during adolescence is less clear, and very little is known about this issue during adulthood or old age. The only previous behavioral genetic study of cognitive abilities in late life is the New York State Psychiatric Institute Study of Aging Twins, which was concerned primarily with biological aging (Kallmann & Sander, 1948). Six cognitive tests were administered to a subsample of 75 identical and 45 fraterAddress correspondence to Nancy L. Pedersen, Department of Epidemiology, Institute of Environmental Medicine, The Karolinska Institute, Box 60208, S-10401 Stockholm, Sweden. 346 Copyright © 1992 American Psychological Society VOL. 3, NO. 6, NOVEMBER 1992 This content downloaded from 152.14.136.96 on Sat, 8 Jun 2013 07:46:13 AM All use subject to JSTOR Terms and Conditions PSYCHOLOGICAL SCIENCE N.L. Pedersen et al. nal twin pairs between the ages of 60 and 89 (Kallmann, Feingold, & Bondy, 1951). Unfortunately, twin correlations were not reported, and thus it has not been possible to use the results of this study to estimate the magnitude of genetic influence on cognitive abilities late in life. Mean twin intrapair differences were reported, and these suggest genetic influence in that identical twins yielded smaller intrapair differences than did fraternal twins, except in the case of a test of digit memory (Jarvik & Bank, 1983). The only other data relevant to late life come from a report of a pilot study for SATS A based on 12 cognitive tests administered to 34 pairs of fraternal twins who had been reared apart and were 59 years old on average (Pedersen, McClearn, Plomin, & Friberg, 1985). A measure of general cognitive ability, an unrotated first principal component score based on the 12 cognitive tests, yielded a twin correlation of .52 for this small sample, after the effects of age, age at separation, and differences in the twins' degree of separation were partialed out. This correlation is consistent with a hypothesis of substantial genetic influence. However, the 95% confidence interval for the correlation is from .19 to .75, which suggests caution should be exercised in estimating heritability from these data. Two twin studies of cognitive abilities in the middle adult years (average age of 40) also found heritabilities for cognitive abilities that greatly exceed the estimates of 50% typically found earlier in the life course. For example, heritability of IQ was estimated as 80% in a Norwegian study of 40 pairs each of identical and fraternal twins from 30 to 57 years of age (Tambs, Sundet, & Magnus, 1984). In a report of about 45 pairs of identical twins who had been reared apart and ranged in age from 19 to 68 years, heritability of IQ was estimated at 75% (Bouchard, Lykken, McGue, Segal, & Tellegen, 1990). Based on the evidence from midlife and later life and the substantial corpus of studies in adolescence and young adulthood, we hypothesized not only that cognitive abilities later in life would show genetic influence but also that heritabilities might be higher than the heritabilities of 50% typically found in adolescence and young adulthood. We also predicted that specific cognitive abilities would show results similar to those found earlier in life: The magnitude of genetic influence for tests of specific cognitive abilities was expected to be lower than for general cognitive ability, and genetic influence was expected to be lower for memory tests than for other specific cognitive abilities.