Overgeneralizations and Obfuscations: A Response to Turkheimer(‘s blog)

Recently, Eric Turkheimer, a foremost behavioral geneticist, wrote a blog about the debate in Criminology around heritability studies, primarily focusing on arguments in our original article (see here and our rejoinder). While I am under no illusion that people will actually read this response/blog, I wanted to respond to Turkheimer’s critique because, like other (pro-BG) responses to our critique, it is grounded in a misunderstanding or mischaracterization of our argument.[a] Like Barnes et al., Wright et al., and Moffitt and Beckley before him, Turkheimer seems unwilling to engage with the substance (explicit focus) of our critique, but rather chooses to interpret it (wrongly) as anti-genetic and “anti-BG.”[b] I feel compelled to address these misguided responses to (critiques of) our critique.

To foreshadow the discussion below, drawing on Turkheimer’s quotes, including his critique, I maintain that Turkheimer agrees with the substance of our critique, with discord only appearing if our arguments are mischaracterized (overgeneralized to topics that they were avowedly not addressing, e.g., the broader field of behavior genetics, the use of twin and adoption studies for non-heritability purposes), and that we cited him correctly.

I respond to each of Turkheimer’s points below. The responses are numbered to correspond to the numbered points he raised in his blog.

(1) Like Wright et al. and Barnes et al., Turkheimer balks at our call for an end to heritability studies as (“sounding a lot like”) scientific censorship. This is both unnecessarily polemical and a rather weak response to a scientific argument, given that this is the way science proceeds when advances uproot the flaws (or lack of utility) of certain scientific practices. This is a gradual process whereby (some) scientists begin to realize the flaws of a paradigm or practice (e.g., the geocentric theory of the universe or phrenology) and bring it to attention and the topic of debate. Ours was an attempt to inform criminologists, who have neither the background in biology nor the time or interest to keep pace with the incredible advances in the life sciences in recent years, about the flaws of the heritability study and to make a case for superannuating (or hastening the decline of) the practice of estimating h2 as well as recognizing the flawed nature of its existing estimates. This is in the spirit of advancing science. As Popper (1962) articulated (and we cited this in our paper) “our knowledge grows only through the correction of our mistakes.”

(2) Turkheimer states: “If noting else, the generic citations of Rutter and Turkheimer as a kind of appeal to authority is a lousy way to argue. What did I say, exactly, and why is what I said relevant to the current argument.” We cited Turkheimer and Rutter not as an appeal to authority but as foremost experts in the field. Barnes et al. responded to our initial critique with the intimation that we were merely extreme environmentalists who were uninformed about biology (presumably in contrast to them), and that, as such, our criticisms should be ignored because we did not have expertise in the areas (aka, we didn’t know what we were talking about.)[c] Rather than as an “appeal to authority” as a means of arguing, we cited Rutter and Turkheimer as experts in the area who also thought that heritability studies had little (to nothing) to contribute to science at the present state of knowledge.

But, what did Turkheimer say?[d] In the paper, we pointed the reader to two pages in two articles that contained the discussion supporting our interpretation of his statements indicating that heritability studies, at present, lack utility and should come to the end. Here are the quotes:

Turkheimer (2011: 598)

“Whether this exercise [heritability studies, which he often refers to as variance partitioning because that is what it is] has any scientific content has been the subject of debate for more than a century, and here is a summary of why I think it does not: it is not about cause. Practitioners of the art wanted it to be about cause, in the sense that the relative magnitudes of the various components were supposed to tell us something about the importance of genetic and environmental causes underlying a trait, but they do not…In the real world of humans, in a given context everything is heritable to some extent and environmental to some other extent, but the magnitudes of the proportions are variable from situation to situation, and have nothing whatsoever to do with the causal properties of genes and environment for the trait in question, unless one is interested in the pointless null hypothesis that one of the components is zero….Plomin and Daniels [1987!!] marked the beginning of the end of variance partitioning as a useful endeavor in human developmental science.”

Turkheimer (2014)

“The laws of behavior genetics are not actual laws, and calling them that may have led to some misunderstandings of what was originally intended. In particular, although the First Law of Behavior Genetics has sometimes been considered an endorsement of a hereditarian view of behavior (e.g., Pinker 2003), the universality of heritability is best interpreted as a reductio ad absurdum of these very distinctions—as a way of observing that the endeavor of figuring out how genetic or environmental a trait is, let alone declaring that it is exclusively one or the other, is pointless. (p.532; emphases added).

And,

“The prospect of this outcome has haunted the nature-nurture debate from its inception, as both sides of the old debate were led to a dead end of thinking that the point of the debate was to evaluate the separate effects of genes and environment. It became clear long ago that neither genes nor environment could be discounted for anything important… Genetically informed research designs can partially, imperfectly, control for the genetic and shared environmental confounds that otherwise cloud causal interpretation of associations like these, and they have been extraordinarily successful at doing so. The quantification of heritability itself is unimportant in such analyses…” (p. 535, emphases added).

We could also point to Johnson, Turkheimer, Gottesman, & Bouchard (2009). In the abstract (p. 217): “We see little need for further studies of the heritability of individual traits in behavioral science…”

Furthermore, noting clearly that “heritability studies are no longer important,” Johnson, Turkheimer, Gottesman, & Bouchard (2009) stated: “To understand why heritability estimates are no longer important, it is necessary to understand that they are completely dependent on the specifics of the samples and environmental conditions from which they are taken…This means that little can be gleaned from any particular heritability estimate and there is little need for further twin studies investigating the presence and magnitude of genetic influences on behavior.” (pp. 217-218, emphasis added).

These are some of the things that Turkheimer said that led us to cite him as calling for an end to heritability studies. And, there are more. Although we wish we had the space to include these quotations, we believe that Turkheimer’s words are consistent with our citing him as an insider in the area who also thinks the studies lack utility at the present state of knowledge.

(3) Turkheimer states: “One ongoing problem I have with a lot of recent anti-BG writing is the whole idea of a “heritability study”.  Evan Charney is another person who trades in this idea.  A heritability study isn’t a thing.  A [sic] heritability is a descriptive statistic, an effect size, not a kind of study.  It is kind of like writing a critique of social psychology and railing against “F-ratio studies.”

First, we point out that Turkheimer (like Barnes et al., Wright et al., and Moffitt and Beckley before him) wrongly lumps our study into a body of “anti-BG” writing. We are clear[e] that our critique is narrowly focused on “heritability studies,” and this necessitated and justified our using the term “heritability study” to describe the studies at the focus of our critique.

Second, I wish to clarify what we mean by heritability studies, although I remain rather astonished that this seems to be a point of confusion. What we refer to as heritability studies are those studies whose aims are to come up with a numerical estimate of heritability (h2), which we defined: “Heritability was defined by Wahlsten (1990: 244) as “the proportion of variance in a measure of behaviour or other phenotype in a breeding population that is attributable to genetic variation” and by Plomin et al. (2012: 87) as “the proportion of phenotypic variance that is accounted for by genetic differences among individuals.” (Burt & Simons 2014: 227).

We selected the name “heritability studies” for these studies, following Charney (2012), because we are narrowly focused on studies whose focus is on partitioning variance between “genetics” and “environments”. Heritability studies are not tantamount to twin and adoption studies because studies of twin and adoption samples can be used for other purposes other than to estimate h2. Our critique, as we note in the paper, is not on these other uses of twin (or adoption) studies. (Thus, for example, Moffitt and Beckley’s point that MZ studies of twin discordance can be used to enhance our understanding of criminal behavior is entirely beside the point of our critique. In fact, I think that in some situations MZ studies of twin discordance offer great promise for shedding light on environmental influences on phenotype variance.)

Turkheimer argues that the “heritability study isn’t a thing. A heritability is a descriptive statistic, an effect size, not a kind of study.” Sure it is a kind of study. A heritability study is exactly how we (and Charney) have described above. What’s not to understand? The heritability estimate is a (time, population) specific effect size, and the study that estimates heritability is a heritability study. This is pretty straightforward. Critiquing studies that merely attempt to estimate h2 is not at all like “writing a critique of social psychology and railing against ‘F-ratio studies’. Social psychological studies that use F-statistics do not have the goal of estimating the F-statistic and they do not present the resulting F-statistic in the abstract as the key finding from the study. No, such studies have the goal of examining whether relationships exist, and they certainly don’t report F-statistics as the key finding of the study. If they did, then they could be appropriately called F-statistic studies. Similar to our use of heritability study, scientists refer to studies that assess the dimensionality of various measures as “factor analytic studies.. Heritability studies are absolutely a thing (as Turkheimer notes in his point 4) like factor analytic studies.

(4) But, Turkheimer does in fact understand what we call heritability studies: “What Burt and Simons think they mean by a “heritability study” is a study that has no point other than estimating the heritability of something.  My friend Ron Yeo used to refer to these as h-squared-equals studies. On this point they can cite me, and the 2011 paper they cite is a reasonable source.  The heritability of criminality doesn’t mean that it has somehow turned out to be “biological” and whether the estimated value in some twin study is .3 or .65 makes very little difference and doesn’t tend to replicate anyway”

Yes, what we “think we mean” is, in fact, actually what we mean (thanks).[f] And, with this, we would point out that Turkheimer does acknowledge that heritability studies (h2-equals studies) are, in fact, a thing.

This is exactly how we cited Turkheimer and this is exactly our argument. As we noted, there had been a profusion of heritability studies (Turkheimer’s seemingly preferred appellation, “h-squared equals studies”) in criminology over the past several years, and these were the studies we were calling for an end to.

(5) Turkheimer: “Barnes et al are right: I do twin studies for a living, and it would be mighty hypocritical of me to declare that they are useless in general.  As a BG person who has spent a lot of time criticizing the heritability concept, I have made it a discipline to never suggest that my doubts about numerical heritabilities should lead to a general dismissal of genetic effects on behavior or the behavior genetic enterprise generally.”

With his statements on point 4, it appears that we did cite Turkheimer appropriately, and he does agree that we should stop estimating heritability studies at the present state of knowledge. Yet, like other BG proponents, he wants to overgeneralize our argument rather than conceding his agreement, twisting it into something it is not (or he is just making irrelevant points that genetic effects aren’t zero, something our paper did not question). This critique is not anti-BG, and nowhere do we dismiss either genetic effects or behavior genetics generally. (We repeat this point several times throughout the papers). We are not broadly criticizing quantitative genetics, nor are we endorsing it, we wanted to call an end to these “h-squared-equals studies,” which were being estimated and published on nearly every outcome without an end in sight.

To be clear, studies that employ the twin design for reasons that are not for the purposes of estimating heritability are not the focus of our critique, which, again, is why we employed the term heritability study (not twin studies).

(6) Turkheimer though takes us to task for not doing something that was irrelevant to our paper: “Where in the remainder of their articles is a concession that genetic pathways rule out certain interpretations of the data, an acknowledgement that criminology has to take genetics into account when it is interpreting its findings?  Like every anti-BG paper I have ever read, they spend the rest of their paper finding something to attack in every single genetically oriented paper they can find.  Twin study?  EEA.  Adoption study?  Prenatal effects.  If all of non-experimental social science were held to this standard it would just go away.” Again, this is beside the point. Our goal was squarely on heritability studies, and “finding something to attack in nearly every single genetically-oriented paper we could find”? Weak sauce, as we were focused on criticizing heritability studies and, yes, our focus was identifying their problems. It was simply out of the scope of our paper to concede that genetic pathways rule out certain interpretations of data. We also did not talk about the effects of labeling by the criminal justice system on interpretations of the effects incarceration on the likelihood of recidivism. Both are out of scope.

(7) Turkheimer states that he thinks that studies of twins can be used effectively for purposes other than heritability. Off topic, but good to know. (Such was clear from his papers we cited.)

(8) Turkheimer asks: “Now the question is whether the studies that Burt and Simons list are h2-equals-studies or more interesting genetically informed social science.  I’m not going to take the time to go through them one at a time, but the several I looked at were bivariate “quasi-causal” (as we call them in our lab) studies.”

I can answer this. The studies we identified (per reviewer requests to point to studies) all involved an “h2-equal” as the sole goal if not a primary goal of the paper.

(9) Turkheimer stated: “Moffitt and Beckley wrote a nice reply about epigenetics and I won’t go into depth about it.  I think that people who think that epigenetic explanations of behavioral differences are important should go ahead and do the research and show it.  I doubt very much that a demonstration of meaningful epigenetic effects would meet with any opposition in the behavior genetic community.  Epigenetics of behavior is behavior genetics.  Epigenetics and classical genetics aren’t at odds on a biological level, and there is no reason they should be at odds in social science.”

I have so many thoughts on this point that I’m actually not even sure where to begin, so I shall go in chronological order.

As I noted above, I found Moffitt and Beckley’s response to be both misguided and off topic. For example, they summarized our argument as follows: “In their original article, Burt and Simons (2014) argued that heritability studies should be abandoned because twin and adoption research is a fatally flawed paradigm. They pointed optimistically to epigenetics research as the way forward.” And, later, they summarized our argument as: “recommending…that criminologists should embrace epigenetics and abandon twin studies.” In short, not exactly. For the former, we focused on ending heritability studies because these studies are flawed conceptually and biologically, and we pointed to epigenetics research as demonstrating the biological fallacy of separate G vs. E influences, and how this new paradigm (postgenomic paradigm) might influence future biologically-informed social scientific work on crime. “Twin and adoption research” are not a “fatally flawed paradigm” because they are not a paradigm. Finally, to repeat (perhaps ad nauseam), we at no point called for an end to the use of studies with samples of twins.

Unfortunately, in our view, rather than engage with our critique, Beckley and Moffitt’s response piece was largely off topic to the debate about the utility of heritability studies. (E.g., Burt and Simons (2014) also suggested that the twin design is irreparably flawed. However, one of the most exciting future uses of twin data for criminology is the study of discordant twins…”). We did not discuss the use of twin sample and methods in general, which is why the value of MZ twin discordance studies is irrelevant. They then proceed to discuss the debates and difficulties in engaging in epigenetics research, which I don’t necessarily disagree with, but again irrelevant to our point. Moreover, at points it seems as though they are arguing that the difficulties (and challenges) of doing social scientific research that accords with current biological knowledge justifies or neutralizes studies that are mired in outdated (biological) knowledge. (Would they have us continue develop knowledge based on the model of the earth as flat following scientific evidence overturning this model because of the difficulties and complexities involved in changing paradigms?)

Neither Turkheimer nor Beckley and Moffitt speak to our primary argument about the flawed conceptual (biological) model of heritability studies nor its lack of utility at the present state of knowledge. And, Turkheimer’s point that epigenetics is behavioral genetics is immaterial. Our critique was avowedly not anti-BG. However, Turkheimer’s statement that “Epigenetics and classical genetics aren’t at odds on a biological level,” is simply not correct, see Meloni (2014a; 2014b) and Charney (2012a) clear and detailed refutations of this statement.

I must also register my surprise that a behavioral geneticist is taking social scientists who see epigenetic findings as relevant and promising to task “for not doing the research and showing it.” Missing heritability? And, this from the scholar who states: “Complex human behaviour emerges out of a hyper-complex developmental network into which individual genes and individual environmental events are inputs. The systematic causal effects of any of those inputs are lost in the developmental complexity of the network. Causal explanations of complex differences among humans are therefore not going to be found in individual genes or environments any more than explanations of plate tectonics can be found in the chemical composition of individual rocks” (Turkheimer 2011: 600). But, the nascent field of epigenetics needs to step up and show their results if they want to be in the discussion (?). Even so, there is, in fact, significant evidence demonstrating the effects of epigenetic differences on behavior (see my discussion here, drawing on a number of various sources, e.g., Landecker & Panofsky 2013; Weaver et al. 2006). To be sure, at this point much of this evidence is confined to non-human animals, yet, this is changing. Scholars are doing this work on humans, and evidence is beginning to accrue that “shows it” (the effects of epigenetic variation on behavioral variation).

(10) (Again, the tired tendency to paint us as anti-BG rather than focusing on the substance of our more focused critique). Turkheimer says we are wrong about GCTA studies. We present evidence; where is his? I have yet to come across one GCTA study that even remotely resembles the high heritabilities found in twin studies of complex social phenotypes. That is our argument. I am more than willing to revise my thinking in light of new evidence, but Turkheimer does not provide any (neither did Wright et al. who did not respond to our response to their use of GCTA studies of height as support for the validity of twin studies of social phenotypes).

Turkheimer concludes with the following: “So in conclusion:  if you want to cite me as a critic of some general version of BG, the citation should be limited to the idea that numerical heritabilities aren’t very important per se, and that studies that do nothing other than estimate them are no longer very important.  The next sentence should be something about how I do maintain that nonzero heritability is important methodologically, and that there are many scientifically useful things to do with twins other than just estimating heritabilities.  Better yet, be very specific about what it is I am supposed to have said when you cite me.”

Again, it seems that we are in total agreement. We did not and have not discussed Turkheimer as a general critic of BG, and cited him as indicating that numerical heritabilities are not very important. We were not able due to space limitations to include all the quotes from Turkheimer, but we did point the reader to the sources and pages where his criticisms of heritability studies were found.

After reading Turkheimer’s blog, I cannot seem but feel he wants to have it both ways. He recognizes and publicly states that heritability studies are not useful, and, yet, wants us to refrain from pointing that out because doing so is somehow anti-BG, anti-genetic, and anti-science and is tantamount to indicting all of genetically-informed social science as balderdash (and thus, makes the critique out of touch with scientific reality). Perhaps, before one can criticize heritability studies, one must first establish a record of supporting genetic effects and the BG enterprise (only insiders can criticize the method?)

In sum, Turkheimer’s critique of our critique is not without substance; yet, as with Barnes et al., Wright et. al., and Beckley and Moffitt, Turkheimer creates a general anti-BG argument out of our narrowly constructed heritability study critique. Rather than conceding that we are in agreement, which it does appear we are, Turkheimer makes our argument anti-BG and us anti-genetic. I find this both highly perplexing and frustrating as it seems to muddy the waters unnecessarily and confuse issues and models that are separate and distinct.

[a] I would like to note that I engaged in a series of pleasant emails with Turkheimer after he sent me a link to his blog, and it seemed that we shared some common ground and were both willing to (re)consider our ideas in light of new information.

[b] For example, Wright et al. (2015) stated: “Consider, too, the linguistic gymnastics used by Burt and Simons (2014, 2015) to claim that nothing positive ‘at the present time’ has come out of behavioral genetic research.” Speaking of impressive gymnastics, this entirely twists our argument to try to feebly come up with the value of heritability studies based on “behavioral genetic findings” and “biosocial data”. We stated that at the present state of knowledge heritability studies lack utility. We have yet to see a response (on point) to this critique.

[c] We should note that none of the authors to who responded to our critique in Criminology (JC Barnes, JP Wright, B. Boutwell, J.A. Schwartz, EJ Connolly, JL Nedelec, K.M. Beaver) have any degrees in the biological sciences.

[d] I will note that we had wished to include quotes from both Turkheimer and Rutter, but we were under strict page limitations. We asked if we could have extra space to include quotes from these to scholars, but this was not allowed.

[e] We noted clearly in our rejoinder, for example: “When it suits their purpose, Barnes et al. and Wright et al. wrongly broaden our critique to behavioral genetics and/or biosocial research generally. We have clearly noted throughout that the heritability study method, including its fallacious technical and conceptual assumptions, is the focus of our critique, not behavioral genetics or biosocial work in general.”

[f] Aside, we find our heritability study label much more interpretable and useful than Turkheimer’s seemingly preferred label “h-squared-equals studies” coined by his friend. This slight difference in conceptualization (or non-recognition of our conceptualization) can greatly impair communication and progress. Point of illustration, this occurs on occasion when my son is either trying be difficult so as not to engage with me or refusing to recognize my clear reference to something because I am not employing his (constantly changing) conceptual framework. In response to my request for him to (please) pick up your airplanes and put them away, he continues resting in his chair and responds, “there are no airplanes on the floor.” I point to the ones right in front of him. “Nope, Mommy, those are rocket ships.”

CITED REFERENCES:

Burt, Callie H., & Simons, Ronald L. (2014). Pulling back the curtain on heritability studies: Biosocial criminology in the postgenomic era. Criminology, 52, 223-262.

Burt , Callie H., & Simons, Ronald L. (2015). Heritability studies in the postgenomic era: The ‘fatal flaw’ is conceptual. Criminology 53: 101-112.

Charney, Evan. (2012). Behavior genetics and postgenomics. Behavioral and Brain Sciences 35:331–410.

Johnson, Wendy, Eric Turkheimer, Irving I. Gottesman, & Thomas J. Bouchard. (2009). Beyond heritability: Twin studies in behavioral research. Current Directions in Psychological Science 18: 217-220.

Landecker, Hannah, & Aaron Panofksy. (2013). From social structure to gene regulation and back: A critical introduction to environmental epigenetics and society. Annual Review of Sociology 39:333–57.

Moffitt, Terrie E., & Amber Beckley. (2014). Abandon twin research? Embrace epigenetic research? Premature advice for criminologists. Criminology 53: 121-126.

Turkheimer, Eric. (2011). Commentary: Variation and causation in the environment and genome. International Journal of Epidemiology 40:598–601.

Turkheimer, Eric. (2014). A phenotype null hypothesis for the genetics of personality. Annual Review of Psychology 65: 515-540.

Wright, John Paul, J. C. Barnes, Brian B. Boutwell, Joseph A. Schwartz, Eric J. Connolly, Joseph L. Nedelec, & Kevin M. Beaver. (2015). Mathematical proof is not minutiae and irreducible complexity is not a theory: A final response to Burt and Simons and a call to criminologists. Criminology 53:113–20.

Advertisements

2 thoughts on “Overgeneralizations and Obfuscations: A Response to Turkheimer(‘s blog)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s