PKU (Phenylketonuria) is a condition that is often invoked to demonstrate that genetic does not mean unchangeable. It is also often cited as a trait with 100% heritability in normal environments that can, nevertheless, be changed and in the right environment have zero heritability. This post explains why these two sentences are not synonymous and how the second is flawed and reinforces an incorrect idea of heritability. I also examine PKU in terms of measured genetic and environmental factors, arriving at the counter-intuitive idea that no gene x environment interaction is involved. Continue reading
Idea: As conventionally interpreted, heritability indicates the fraction of variation in a trait associated with “genetic differences.” A high value indicates a strong genetic contribution to the trait and “makes the trait a potentially worthwhile candidate for molecular research” that might identify the specific genetic factors involved. I contest the conventional interpretation and contend that there is nothing reliable that anyone can do on the basis of estimates of heritability for human traits. While some have moved their focus to cases in which measurable genetic and environmental factors are involved, others see the need to bring genetics into the explanation of differences among the averages for groups, especially racial groups.
a. Heritability & critique
Heritability is a quantity derived from analysis of variation in traits of humans, other animals, or plants in ways that take account of the genealogical relatedness of the individuals whose traits are observed. Such “quantitative genetic” analysis does not require any knowledge of the genes or “measurable genetic factors” involved.
Turkheimer is “on the left” of behavioral genetics, being much less gung ho about the implications of its findings. Here he gives a clear overview of what the field has shown.
Plomin articulates the confident consensus of behavior genetics, namely, that they’ve debunked the supposed environmentalist orthodoxy in social science that says that everything is social and have established a basis for connecting with molecular genetics to identify the actual genetic factors.
Rutter, a senior psychological researcher (who once worked with Brown on social determinants of mental illness), tries to moderate the “polarizing claims” and “unwarranted extrapolations.”
Taylor 2010 casts doubt on the findings that underlie both Turkheimer and Plomin’s articles by exposing problems with the concepts and methods used to arrive at those findings. Taylor ends with a nudge towards methods that use measured genetic factors as well as measured environmental factors (the latter being the staple of social epidemiology).
b. Interaction of measured genes and measured environments
Moffitt 2005 provides a review of what’s involved in trying to identify interactions between measured genetic and environmental factors. (Use Taylor 2010 to get clear about the difference between this kind of interaction and the classical genotype x environment interaction in quantitative genetics.) Caspi 2002 is one of two 2002 papers that caused a lot of splash. Davey-Smith picks up on the current consensus that the 2002 studies have been hard to replicate and invokes Mendelian randomization as a way to strengthen causal inference about interactions between measured genetic and environmental factors.
c. Data & models about heritability & change (or lack of it)
Dickens 2001 provides a resolution of the paradox that heritability of IQ test scores is reported to be high, but there has been a large increase in average IQ test scores from one generation to the next. We know that genes haven’t changed from one generation to the next, so Dickens’ account is also exposing a flaw in the logic that because heritability of IQ test scores is high within racially defined groups and because there is a large difference in average IQ test scores between whites and blacks, genetic factors are probably involved in that difference.
Rushton 2005 however thinks that 30 years of research has validated that idea.
Taylor 2010 refers to Dickens 2001, but gives a somewhat different spin on its implications.
(This post continues a series laying out a sequence of basic ideas in thinking like epidemiologists, especially epidemiologists who pay attention to possible social influences on the development and unequal distribution of diseases and behaviors in populations [see first post in series and contribute to open-source curriculum http://bit.ly/EpiContribute].)
Caspi, A., J. McClay, et al. (2002). “Role of Genotype in the Cycle of Violence in Maltreated Children.” Science 297(5582): 851-854.
Davey-Smith, G. (2009). “Mendelian randomization for strengthening causal inference in observational studies: Application to gene by environment interaction.” Perspectives on Psychological Science, in press.
Dickens, W. T. and J. R. Flynn (2001). “Heritability estimates versus large environmental effects: The IQ paradox resolved.” Psychological Review 108(2): 346-369.
Moffitt, T. E., A. Caspi, et al. (2005). “Strategy for investigating interactions between measured genes and measured environments.” Archives of General Psychiatry 62(5): 473-481.
Plomin, R. and K. Asbury (2006). “Nature and Nurture: Genetic and Environmental Influences on Behavior.” The Annals of the American Academy of Political and Social Science 600(1): 86-98.
Rushton, J. P. and A. R. Jensen (2005). “Thirty years of research on race differences in cognitive ability.” Psychology, Public Policy, and Law 11: 235-294.
Rutter, M. (2002). “Nature, nurture, and development: From evangelism through science toward policy and practice.” Child Development 73(1): 1-21.
Taylor, P. J. (2010). “Three puzzles and eight gaps: What heritability studies and critical commentaries have not paid enough attention to.” Biology & Philosophy, 25:1-31. (DOI 10.1007/s10539-009-9174-x).
Turkheimer, E. (2000). “Three laws of behavior genetics and what they mean.” Current Directions in Psychological Science 9(5): 160-164.
In 2002 Avi Caspi, Terrie Moffitt and colleagues published two articles in Science that examined psychological traits in relation to measured genetic and environmental factors. In one of them they reported on anti-social behavior in adults in relation to the activity of monoamine oxidase typeA (MAOA) and childhood maltreatment (Caspi et al. 2002). They summarized their results in figure 1. For people who experienced severe childhood maltreatment and have low MAOA activity, the average antisocial behavior was higher than if they had only one of the conditions and much higher than if they experienced low or no childhood maltreatment together with high MAOA activity. In other words, MAOA deficiency was a strong predictor of aggressive behavior only when the child had also been maltreated.
The authors conclude that their results ‘could inform the development of future pharmacological treatments’ (Caspi et al. 2002, 853). In the context of research on childhood experiences in relation to adult behavior, the implication of their conclusion is that, if low MAOA children could be identified, prophylatic drug treatment could reduce their propensity to anti-social behavior as adults. Or, to be more precise, such treatment could reduce their vulnerability to childhood maltreatment precipitating undesired adult outcomes. An easy rejoinder would be that, if childhood maltreatment could be identified and stopped early, this action could reduce their vulnerability to low MAOA levels leading to undesired adult outcomes. Indeed, eliminating childhood maltreatment would seem to be unconditionally positive, while prophylatic drug treatment may have sideeffects, and some of these may not emerge till later in life. The rejoinder is too easy, however. The social infrastructure needed to detect and prevent childhood maltreatment would intrude into many households, require surveillance, monitoring, and intervention by state agencies, divert government budgets from other needs, and so on. The specific outcome may be positive, but the means are not unconditionally positive to all. How would decisions about investment in the social infrastructure be decided? How would individuals decide where to engage with that social infrastructure once it is established?
Once the issue of social infrastructure is brought into the picture, more significant problems emerge with the idea of early detection and intervention on the basis of MAOA status. Notice that the points plotted in Figure 1 are the averages for the respective categories of people. Within each category people show a range of anti-social behaviors. It turns out that, among children who experienced probable or severe maltreatment, the ranges overlap, that is, some of the high MAOA individuals ended up with higher anti-social behavior scores than some of the low MAOA individuals. The potential for misclassifying people as ones who may end up antisocial is not eliminated by adjusting what counts as antisocial. If we count as antisocial only those individuals whose score exceeds some value that is higher than the upper limit of the range for high MAOA individuals, this increases the numbers of low MAOA individuals who do not end up counting as antisocial. If we lower the cutoff score, many high MAOA individuals end up with behavior classified as antisocial (Figure 2). Indeed, playing around with the cutoff score, the best that can be achieved with Caspi et al’s data is a little more than one-third of children correctly classified on the basis of their MAOA status (i.e., low MAOA ending up counting as antisocial and high MAOA ending up not).
The issue of misclassification is especially troubling because, once the resources are invested to screen children for MAOA levels, attention would be focused on all low MAOA children. Indeed, how could this stereotyping be avoided if we do not know from a childhood MAOA assessment whether any particular individual is one who would go on, after maltreatment, to be an antisocial adult? Additional research would be needed to identify other characteristics that differentiate among the low MAOA children (and perhaps help predict who among the high MAOA children are also vulnerable). If that research were successful, additional resources would have to be invested to customize the way that parents, teachers, doctors, social workers treated the different low and high MAOA children and to educate everyone not to treat children according to their MAOA group membership. Just as in the case of PKU, the meaning of new genetic knowledge (in this case in combination with environmental knowledge) is contingent on the presence or absence of social infrastructure; the positive benefits depend on extensive social reconstruction.
Another excerpt from P. Taylor, “Infrastructure and Scaffolding: Interpretation and Change of Research Involving Human Genetic Information,” Science as Culture, 18(4):435-459, 2009