More on Murray: What is Biological Determinism?

More on Murray: What is Biological Determinism?

Reactions to my post yesterday about Human Diversity mostly involved my invocation of the idea of genetic determinism, henceforth GD.

I know that GD is sometimes used as an all-purpose tool to attack any assertion of a role for genetics in human behavior, but that isn’t my point. In fact I think the hereditarian argument requires GD, isn’t interesting without GD, and that there are good examples of actual GD. The point is that asserting GD requires additional evidence over and above what is actually available for the vast majority of the human differences we are talking about here.

As usual, let’s start with marital status. Marital status is heritable, and there are real-world consequences of that heritability. If you are studying (as I have) the effects of divorce on children in the family, you can’t just compare kids in divorced and non-divorced families, because the environmental pathways are confounded with genetic ones. But suppose a friend of yours is getting divorced. You probably wouldn’t say, well of course, that’s because you have genes that caused you to get divorced. Why not, exactly? Marital status, we have agreed, is heritable in twins, and if no one has gotten around to the GWAS of marital status yet it is only because it hasn’t seemed worth it. The genetics of divorce is not scientifically different from the genetics of personality or intelligence or mental illness.

Divorce is heritable, but it is not GD. What I mean by this is that we have utterly no idea about the how or why or when of the particular genes that happen to be involved in the correlation between marital status and genotype. Would our friend’s genes have led to divorce in a world with better access to mental health care, or if he had had a kid, or if he didn’t drink, or…. on and on and on. The list of things that might– almost certainly do– modify the “causal” (the word barely applies) pathway between a bit of DNA and divorce is so vast, so uncontrolled, that we know that we can’t say anything more than, “genotype and phenotype are modestly correlated.”

This universal, causally indeterminate heritability leaves the hereditarian in a dilemma. You could say, no, the heritability is all I need. Divorce is heritable, and that means one way or another you have to think about genes when you think about divorce. That’s fine, but it is true for everything, and it leads to an empty GE interactionism– both genes and environment contribute to the genesis of behavior– that is never going to get Murray where he wants to go. If you want to write a scary book about how the problem of divorce in America is the result of our genetic makeup, it won’t do to just point to the first law of behavior genetics. You need to say there is something specific about our genes that is making us get divorced in a specific way, and you can’t do that.

Groups, like I say, are just collections of individuals, and the exact same argument applies to the explanation of group differences. Divorce rates are higher in the US than they are in Korea. Marital status is heritable in both places, and there are identifiable genetic differences between Koreans and Americans. Is the difference in divorce rates the result of genetics? Presumably not, and once again the situation isn’t really any different for divorce than it might be for personality or IQ. And the hereditarians have the same choice: if they want they can say, everything is heritable, all groups differ to some extent genetically, and therefore all behavioral differences must have something to do with (or “be partially determined by” or “have a component of” or whatever weaselly phrase they want to trot out) genetics. But once again that won’t get them anywhere, because it applies to everything. It’s just a generic assertion that genes are involved in all human development.

Contrast that to Huntington’s Disease, which is GD. If our friend comes down with HD, it is perfectly sensible to say that they got the disease because they have the gene. If there is a country somewhere with a higher rate of HD than the US it is because of genetic differences between the two countries.

I would sum up Murray’s argument as the assertion that group differences in personality and IQ, among other things, are more like HD and less like marital status than has previously been acknowledged. What’s the evidentiary basis of that assertion? What kind of additional evidence do you need? This is what is never spelled out in hereditarian arguments. I think there are two main possibilities.

The first, which I have discussed before, is knowledge of mechanism. People sometimes respond to this idea as though I am just making up an impossibly high criterion for GD, but that isn’t the point. The problem with asserting GD for marital status is that we have no knowledge of the scope and potential moderators of the relationship between genotype and phenotype. Would person X be prone to divorce in all conditions? We don’t know and have no particular hope of finding out. For HD, we more or less understand how the gene causes the behavioral disability, and that allows us to rule out all sorts of uncontrolled possibilities that might moderate the relationship. In the same way, we don’t have to worry about moderators of HD across groups. The gene causes HD in Japan in more or less the same way it does here.

I know it seems pretty unlikely that we will achieve that kind of understanding for marital status, or in my opinion for intelligence or personality, so I want to sketch out another possibility that may apply more to psychology. Suppose for a moment that it turns out that the Plomins and the Murrays are correct, and the 2-3% of the IQ variance we can currently account for (within families) with a PGS eventually increases to 40%, about where it is for height. Families come to routinely use it to select embryos, producing an amazing crop of hyper-intelligent children. Moreover, it turns out that the PGS works in pretty much the same way everywhere, rich and poor, male and female, Asia, Africa and North America. I think we would pretty soon starting treating the PGS as GD, even if we didn’t understand the mechanism. It would be like the effect of poverty on cognitive ability, or smoking and lung cancer. A big effect size, plus a meaningful theoretical, if not exactly mechanistic, basis. It would have genomic construct validity. Interesting idea to pursue.

I emphasize that this is not where we are, and I don’t think we are ever going to get there. But it explains why the Plomins and Murrays are always leaning on the inexorable progress of science to base their arguments on some imagined genomic future that is sure to arrive before 2030. The structure of the argument is like this: Soon we are going to have personal flying cars, and society had better start building appropriate traffic control structures, or everything is going to be a mess when it happens! Except instead of some harmless science fiction fantasy, they are anticipating a genomic dystopia in which poverty has turned out to be a biological given and racism has turned out to be empirically verified.

A final note is that this conversation has been going on for a long time. The vast majority of Human Diversity could have been written by Arthur Jensen in 1990. And as a result, I have been saying these things for a long time myself. The following is from 1998, using “weak” and “strong” biological explanation in place of garden variety heritability and GD:

Proponents of biological explanation do not generally defend the claim that one can or ought to reduce complex behavior to biological terms. Instead, they report the results of experiments that purport to demonstrate that behavior has turned out to be biologically based. However, to the extent these experimental results are necessary consequences of the physical instantiation of psychological processes, they provide only the weakest support, analogous to the support a behaviorist could claim on the basis of an experiment showing that behavior is “influenced” by schedules of reinforcement. This confounding of the philosophical and empirical underpinnings of biological explanation has provided biologism’ s adherents with a nearly impregnable defense: They claim biological .explanation as a strong empirical result on the basis of experiments that depend solely on the necessary instantiation of behavior in the body and brain.

Eric Turkheimer

Eric Turkheimer is the Project Leader for the Genetics and Human Agency Project. Eric is a clinical psychologist and behavioral geneticist. For thirty years he has been involved in empirical and theoretical investigations of the implications of genetics for the genesis of complex human behavior. Current projects include understanding the interaction between socioeconomic status and the heritability of intelligence, and philosophical analysis of the ethical status of work that purports to demonstrate biologically based differences in behavior among racial groups.

  • Rick Frey
    Posted at 00:37h, 30 January Reply

    This argument explaining genetic determinism makes much better sense than the quick reference in your blog on the Murray book, so sorry for jumping on that. But I still think what you’re saying misses the most basic understanding of hereditarian arguments.

    Imagine the thought experiment of some grad student doing data analysis on the broad collection of Minnesota twin studies. They are the dumbest grad student in the world when it comes to genetics, but they are experts at recording data, analyzing data, running any stats assessments of data.

    It isn’t their belief in the hereditarian argument that would create the correlations, it isn’t their friendship with Murray or membership in a white supremacist group that would show a range of hereditarian results across traits. It’s there in the data. It can’t be avoided, it just exists.

    It seems like your argument (to sum it up like you summed up Murray’s) is since we don’t know how genetics affects divorce, unless we’re talking about pure GD in cases like Huntington’s, it’s worthless to notice these correlations.

    If you want to argue that reliable trends in data are useless, you can. But the future is data science and these trends are not meaningless noise. There is powerful information in this data that has significant predictive utility. Some of this utility is helpful for the people who know it (an increased likelihood of getting certain cancers so the person screens earlier and more regularly). There is data that can be used dangerously (increased likelihood of getting a disease so rejected by insurance company).

    The hereditarian argument doesn’t require GD, just because you think it makes it more interesting or easier to argue against. The hereditarian argument is math applied to data. The data and the stats just exist. If you want to say the data is wrong, you could try that argument. If you want to say the math and stats are wrong, you could argue that. But if the data is accurate and the stats are correct, there are significant correlations in the data related to heritability. Obviously people can interpret this data incorrectly, but you seem to be trying to say that there’s no rug under Murray’s feet, there’s no there there. The data are abundantly clear.

    Neither Murray nor any respectable population geneticist would ever write a book about how genes determine divorce. No idea where you got that idea from. Folks all over the world are studying genes and height. At some point, there will be genetic testing for it, embryos with indicators of greater height will be selected for. It won’t be a 100% accurate predictor–it’s not pure GD, but it’s not random either. Many many traits are in between HD and divorce, and the closer they are to HD, the more valuable it is to look at the data and patterns. Whether you like it or not, there is reliable, powerful predictive utility to that data and if we don’t get better at thinking about it and talking about it, we’re in big trouble.

    • Sean Creighton
      Posted at 22:31h, 03 February Reply

      I appreciate these blog posts and the “gloomy prospect” blog.

      @ Rick Frey, I’ve a few questions:

      (1) You say that the data just exists about IQ. What do you mean? Are you saying there are particular genetic markers associated with the various demands of an IQ test that explain differences among people? If so, what are they? Is there a database I am unaware of that finely details a nice set of genes for each aspect of intelligence? Where is this existent data, even if it is incomplete?

      (2) Is there a handy guide for applying math to data? I’m not sure how to manipulate symbols in the way you mention the grad student above doing so. After I gain these math skills, should I select the data at random, in a sort of haphazard fashion? I’m not sure where to start.

      (3) At what point do the correlates become deterministic? I took determinism to mean that the genes are causing the outcome (i.e,, no environmental difference will matter for the expression) and the heritability hypothesis to be saying, “look, we controlled enough factors to believe that genes are doing x percentage of the work.” You seem to be saying the Hereditarians aren’t looking for genetic causes but instead are correlation junkies who accidentally find correlations that are made useful because they correlate. I may have misread in thinking that the use of heritability studies was tied to the genetic influence.

      I think Eric did a nice job elucidating the first “law” that “everything is heritable.” For instance, if we wanted to find genetic markers for people who like to pop balloons and birthday parties, it seems plausible that such markers would exist. Finding such associations doesn’t strike me as very valuable. In the case of HD, it’s certainly valuable because we have isolated genes doing a whole lot of explanatory work, no matter what culture or what balloon popping propensity one has. But with divorce, how do genes help us understand why Susan cheated on Dan in one scenario, Steve went to prison for 27 years and got divorced in another scenario, and two people got married at age 19 and then changed a lot by age 33 in another? What genes unite all those bizarre situations? Why would we think genes make a difference here? That’s a weird project to get off the ground, haphazard correlation finding or not.

  • John Clarke
    Posted at 15:15h, 30 January Reply

    I’m no expert on these things, but unless the identical twin and adoption studies have been wildly misrepresented it seems to me that intelligence is closer to HD at least in the United States. If you have ok parents and go to an ok school and don’t have major accidents or illnesses your intelligence will probably rise to its natural level.

    For example if it takes an IQ of 115 to be an accountant and 130 to be a research scientist, I can’t imagine what better parenting or schooling could do to have turned the average accountant into a PhD in chemistry. Now maybe that’s a little extreme jump in IQ but how much of an IQ gain do Murray’s critics think is possible with a typical middle class child in the USA?

  • Joe James
    Posted at 12:20h, 31 January Reply

    The comparison with divorce isn’t apt. Divorce is regulated by statute – under what terms can you get one (until recently you needed cause/fault) and what happens financially after the divorce. Not like intelligence at all.

    We have adoption studies, twin studies, brain scan studies that correlate intelligence to glucose metabolism and also studies that show brain weight corresponds to intelligence. With respect to the race/IQ debate we also have adoption studies, admixture studies, etc. Conversely, we know that early intervention programs such as Head Start, while showing some benefit early on, have by age 18 no positive effect on IQ (which is what you’d expect if intelligence is highly heritable).

    I don’t see how anyone benefits by setting up an impossibly high standard (fine the genes). If these things didn’t have implications for race and social policy I don’t think it would be controversial, eg, that there is a non-trivial genetic component to race differences in intelligence.

  • Darin Johnson
    Posted at 17:58h, 31 January Reply

    I believe the comparison with divorce is okay. The differences in statutes are analogous to differences in environment: they affect how genes are expressed. If divorce laws were changed, I may be more or less likely to divorce, and it may change the calculated heritability (e.g., if divorce is illegal, heritability will be zero), but it wouldn’t change the fact that there’s a genetic effect. So I don’t see how it this supports Turkheimer’s point.

    And I WOULD say to a divorcing friend that his genes may have raised the odds of it happening. Why is that even controversial?

    Well, I may not say it, but I’d think it.

  • Joe James
    Posted at 13:41h, 02 February Reply

    “I believe the comparison with divorce is okay. The differences in statutes are analogous to differences in environment: they affect how genes are expressed.”

    I guess my point is that putting intelligence on a Huntington’s Disease —> Divorce spectrum isn’t necessarily the best way to look at it. I’d try to create a spectrum of traits such as divorce, religiosity, intelligence, vegetarianism and see where intelligence falls.

    For example, the US has never had a national church and attendance has been voluntary (maybe a few exceptions but none in the last couple hundred years). So religiosity is influenced by family and societal pressure. Religiosity ebbs and flows. It seems to be ebbing now. Or take vegetarianism/omnivorism (I think that means a mixed plant/meat diet). Most people can afford either and the ratio of each groups varies.

    But compare these things to intelligence. Based on everything I’ve read, intelligence is much more heritable. People can go back and forth on their commitment to vegetarianism and religion, but you are pretty much struck with your intelligence and the b/w IQ gap stays roughly the same. Looked at it that way, I guess intelligence is more like HD than divorce.

    Sure, you could shoot someone in the head and his IQ would go to zero. You could ban or restrict divorce and the heritability would drop. But that’s obvious. The point is: in today’s USA why believe that intelligence isn’t largely heritable for most people (ok parents, ok schools, no abuse). Just because you could abuse a child and decrease his IQ 10 points doesn’t mean the average person can learn differential equations with the best possible upbringing.

    In his 2007 debate with Charles Murray, James Flynn admitted that it is difficult to come up with an explanation for the persistent B/W IQ gap in the US.

    • Darin Johnson
      Posted at 16:18h, 02 February Reply

      Unless I’m mistaken, Turkheimer himself has published such a spectrum. As I recall, heritability of religiosity is comparable to heritability of IQ — something around 0.5.

      Talking about Huntington’s Disease is just squid-ink. Nobody expects a direct, single-gene causation relationship for complex traits that are themselves not binary. You don’t either have IQ or not have it like you do with Huntington’s Disease. Charles Murray himself understands (I promise!) that heritability of traits like IQ would change in a different environment — so if GD requires believing something else, as Turkheimer suggests, then Murray is not GD. And neither is anybody else who is passingly familiar with gene-environment interactions.

Post A Comment