Thinking Inside and Outside the (Black) Box ...

Commentary Human Development 2006;49:343–346 DOI: 10.1159/000096533

Thinking Inside and Outside the (Black) Box: Behavioral Genetics and Human Development Richard Rende Brown Medical School, Providence, R.I., USA

Key Words Adoption Gene-environment interaction IQ development

Behavioral genetic research has generated much interest and commentary from developmentalists over the last few decades. Certainly there has been considerable acceptance of the behavioral genetic model in developmental research and recognition of the yield of the various research designs, including the adoption paradigm along with the twin method as well as other variants [Rende & Plomin, 1995; Rende & Waldman, 2006]. What is still at issue, however, is what to make of the findings to date, and what these findings suggest in terms of intervention and public policy concerning the development of behavioral traits and disorders, including the complex phenotype IQ. The paper by Richardson and Norgate illuminates the areas that still generate debate. Behavioral genetic researchers have used the core methods (primarily twin and adoption studies) to estimate proportions of variance on traits (or individual differences in the population) that can be ascribed to either genetic or non-genetic influences. In the case of IQ, the studies to date provide a robust estimate of heritability which seems to increase with age, and which also appears to dwarf the effect of the environment. Much of the commentary offered by Richardson and Norgate aims to draw attention to a number of limitations of the adoption paradigm which could effect the estimation of heritability. These points are well taken and are consistent with the caveats that behavioral geneticists sometimes (but not always) make when interpreting their findings [see Rende, 2004]. The position adopted in this commentary is that the critique raises both old and new questions about behavioral genetic research. A key old question, which continues to stir debate in some circles, is the meaning of the statistics generated by the quantitative genetic model. Simply put, what is heritability and what does it mean to determine that something is highly heritable? These questions require a close look

© 2006 S. Karger AG, Basel 0018–716X/06/0496–0343$23.50/0 Fax +41 61 306 12 34 E-Mail [email protected] Accessible online at: www.karger.com www.karger.com/hde

Richard Rende Transdisciplinary Research Group, Butler Hospital 345 Blackstone Blvd, Providence, RI 02906 (USA) Tel. +1 401 455 6617, Fax +1 401 455 6618 E-Mail [email protected]

inside the black box of behavioral genetics, namely the anonymous components of variance that are inferred and ascribed to latent (unobserved) factors. Descriptions of the heritability statistic abound in the behavioral genetic literature [e.g., Rende & Plomin, 1995; Rende & Waldman, 2006], but it is worth reinforcing that heritability is a descriptive statistic, with an estimated effect size that is based on a number of assumptions (e.g., an additive genetic model, no gene-environment interaction or gene-environment correlation) applied to a particular population of individuals studied at a particular moment in time using specific constructs of measurement (e.g., definitions of the phenotype) as well as measurement techniques. The number of disclaimers offered in the preceding sentence serves as a testimony to the limitations of the method and these have been acknowledged frequently by behavior genetic researchers. The difficulty, I believe, resides in the overemphasis on heritability and especially the magnitude of effect by both proponents and critics of behavioral genetics. Although the advances in biometrical modeling that have occurred over the past two decades have yielded much information on the etiology of traits and disorders, modern behavioral genetic papers could give off a false sense of precision, and could be read as providing concrete evidence of biological determinism, if one does not attend to the realities of the methodology and the specific way in which the parameters should be interpreted (e.g., Rende, 2004; Rende & Plomin, 1995; Rende & Waldman, 2006). Simply put, a significant heritability statistic may be interpreted as providing evidence consistent with a genetic contribution to the phenotype being studied, given all the caveats associated with the design. A larger effect size (higher heritability), as compared to a lower effect size (lower heritability), simply means that one can be more confident that there is a genetic effect (as defined in this model) that is greater than zero. The heritability statistic does not have biological meaning and does not refer in any way to the actual effect sizes of genes or genetic systems within the broader biological and social context. The behavioral genetic paradigm has been useful in bringing attention to a role for genetic influences on nearly all traits and disorders. For example, prior models of autism that focused on environmental risk factors such as ‘refrigerator mothers’ were rejected on the evidence of heritability from twin studies. On the other hand, twin studies of psychiatric disorders provided compelling evidence against deterministic single gene models simply by suggesting that the heritability was always less than one [see Rende, 2004], and actually offered an empirical basis for more probabilistic models that incorporated the effects of multiple genes as well as non-genetic influences. One take on all this is that most behavioral genetic studies have essentially given credence to the suggestion that both genes and environment are critical factors that contribute to the etiology of individual differences on nearly all behavioral traits and disorders, hardly a controversial conclusion but clearly a rational one with empirical support. It is important to recognize that quantitative genetics is skewed toward detecting genetic effects. The descriptive statistic ‘heritability’ actually will encompass any effect that is even partially attributable to genetics, such as gene-environment interaction and gene-environment correlation. The only term that is manipulated in the quantitative genetic model, when applied to any behavioral genetic design, is the degree of genetic relatedness – environmental influence is not directly examined and is treated as a residual after accounting for the genetic component. This limitation

344

Human Development 2006;49:343–346

Rende

of the model may certainly explain why environmental effects are typically not detected in quantitative genetics, but it does not refute the utility of the model for attempting to detect genetic influence. For example, going back to the roots of the twin paradigm, the idea was to compare identical and fraternal twins in order to see if any difference in similarity existed which would be consistent with some form genetic influence [Rende, Plomin, & Vandenberg, 1990]. The behavioral genetic model may be viewed as a litmus test for genetic influence that provides a reasonable screen for the likelihood of involvement of genetic factors, but a more limited test of environmental influence. The preceding commentary touches on some of the old issues in interpreting behavioral genetic studies. Moving outside the black box of behavior genetics, the fundamental difficulty developmentalists have had with the construct of heritability, or more specifically demonstration of highly heritable effects on traits, is the erroneous assumption that these findings imply biological determinism and a lack of effect that the environment could have on the phenotypes of interest. It is noteworthy and indeed critical to recognize that many studies have altered the classic behavioral genetic paradigms to examine environmental effects and have shown that heritability estimates for a given phenotype may change dramatically under different environmental conditions [for a review, see Rende & Waldman, 2006]. The reason for this is straightforward: heritability estimates are descriptive statistics, and like any class of descriptive statistic, apply only to a given population under given conditions – when differing contexts are put into the model, the variance attributed to genetic influence may fluctuate accordingly. Thus many of the concerns voiced by Richardson and Norgate could, in principle, effect heritability estimates of IQ if they were incorporated into the biometrical models. More to the point, behavioral genetic findings have few implications for the potential of environmental efforts such as intervention strategies aimed at specific individuals and groups of individuals. Behavioral genetic findings do not necessarily have relevance to these efforts because they apply strictly to individual differences, not individuals. Thus, while adoption studies of IQ show little evidence of environmental influence, studies of adoptee populations show profoundly positive effects of the adoptive family rearing environment. For example, children who are adopted score higher on IQ than their nonadopted siblings or peers and do not differ significantly from their nonadopted siblings or peers on IQ testing [van Ijzendoorn, Juffer, & Poelhuis, 2005]. Children reared in profoundly depriving institutions in Romania show extensive gains in cognitive functioning after rearing in adoptive families [Rutter et al., 2004]. These studies attend to issues that are vastly different from those raised by adoption studies using the behavioral genetic paradigm, and provide clear and tangible evidence that family environment substantially influences cognitive development. The Richardson and Norgate critique illuminates many of the limitations within the black box of behavioral genetics. These caveats to the method are not controversial but they are not especially new either. Rather, the same debate continues, with the overall premise seeming to be either acceptance or rejection of behavioral genetic findings (or evaluation inside the black box), rather than expansion outside of the black box. Although dynamic-systems-based approaches (such of those mentioned by Richardson and Norgate) are appealing, they will require translation into either new research designs or adaptations of existing methodologies that will yield a body

Behavioral Genetics and Human Development


345

of empirical findings that may be critiqued to the same degree as the behavioral genetic data base that has accrued over the past few decades. Thus it would be helpful if critics of behavioral genetics move the argument along by suggesting, developing and implementing alternate research strategies (and not just theories) that can go beyond the limitations of the adoption method (and for that matter, the twin method). The overall evaluation of the application of the adoption paradigm to studying the genetic and environmental contributions to IQ requires a balanced view from both within and outside the black box of behavior genetics. Within the box, the inherent limitations of naturalistic designs, the issues concerning the definition and measurement of the phenotype and the particular meaning of heritability within the quantitative genetic model are all important considerations that need to be recognized. That said, the evidence supporting some type of genetic influence on IQ may be used to support research on the genetic (and biological) contributions to the development of a wide range of cognitive functions, with the hope being that such work would be used to gain insight into neural processing deficits that may interfere with optimal learning. This work would have little to do with rank-ordering individual differences on IQ scores in a population, the fundamental framework for determining heritability. The work supporting the advantageous effects of adoption on cognitive performance serves as a concrete reminder that rearing environment matters and that optimal cognitive development most likely requires the most supportive and nurturing environment in childhood and beyond. Ultimately, our basic research findings become important for how they contribute to a fuller understanding of human development and the factors that best promote adaptive outcomes. With respect to behavior genetics, such translation of findings requires an appreciation of the complexities within the black box, the yield given those complexities, and the appropriate scope of application outside the black box along with the development of new research methods that may provide alternate approaches to the complex dynamic interplay of diverse biological, psychological and social systems that affect cognitive development.

References Rende, R. (2004). Beyond heritability: Biological process in social context. In C. Garcia Coll, E. Bearer, & R. Lerner (Eds.), Nature and nurture: The complex interplay of genetic and environmental influences on human behavior and development (pp. 107–126). Mahwah, N.J.: Lawrence Erlbaum Associates. Rende, R., & Plomin, R. (1995). Nature, nurture, and development of psychopathology. In D. Cicchetti & D. Cohen (Eds.), Developmental Psychopathology. Vol. 1: Theory and Method (pp. 291–314). New York: Wiley and Sons. Rende, R., Plomin, R., & Vandenberg, S.G. (1990). Who discovered the twin method? Behavior Genetics, 20, 277–285. Rende, R., & Waldman, I. (2006). Behavioral and molecular genetics and developmental psychopathology. In D. Cicchetti & D. Cohen (Eds.), Developmental Psychopathology, Vol. 2 (2nd edition) (pp. 427–464). New York: Wiley and Sons. Rutter, M., O’Connor, T.G., & the English and Romanian Adoptees (ERA) Study Team. (2004). Are there biological programming effects for psychological development? Findings from a study of Romanian adoptees. Developmental Psychology, 40, 81–94. Van Ijzendoorn, M.H., Juffer, F., & Poelhuis, C.W. (2005). Adoption and cognitive development: A meta-analytic comparison of adopted and nonadopted children’s IQ and school performance. Psychological Bulletin, 131, 301–316.

346


Rende