Jayman on the Three Laws of Behavior Genetics

Jayman on the Three Laws of Behavior Genetics

Charles Murray links to an article about the three laws of behavior genetics, recently updated by others to five.  

As is often the case when people of a hereditarian persuasion write about the three laws, Jayman gets it almost entirely wrong, reacting only to the headline, “Everything is Heritable” and ignoring the great majority of the paper that completely undercuts his conclusions. The paper is actually titled, “Three Laws of Behavior Genetics And What They Mean”, and it is in fact much more about the what-they-mean part than about the laws themselves. And what they mean is: Behavioral differences are in a weak sense heritable, but that doesn’t mean that behavior in general has a genetic explanation. Quite the opposite.

At the beginning of the post, Jayman has the chutzpah to suggest that I have “since spent [my] time trying to undermine [my] own discovery.” Actually, it’s worse than Jayman suggests: I spent most of the paper itself undermining the interpretations he wants to assign to it. That’s pretty bad: I misunderstood the conclusions of my own paper!  You can read the paper yourself, it is linked here.  But since Jayman doesn’t show any indication that he actually took the time to read it, let me summarize for you.

  • First Law. All human behavioral traits are heritable.
  • Second Law. The effect of being raised in the same family is smaller than the effect of genes.
  • Third Law. A substantial portion of the variation in complex human behavioral traits is not accounted for by the effects of genes or families.

The first law states that all behavioral differences are heritable, in what I had earlier called the “weak” sense: all this means is that on average, people who start out more similar genetically wind up more similar phenotypically, for anything and everything. There is that paper going around about how dog ownership is heritable. Of course it is. This doesn’t mean that there are genes for dog ownership, it doesn’t mean there is anything resembling a genetic mechanism determining dog ownership, it doesn’t mean that we have reduced free will regarding dog ownership. If you happen to be conducting studies of parent-child similarity for dog ownership, you need to take it into account; otherwise, it is as close as you can come to being consequence-free.

The second law refers mostly to the results of twin studies, in which the effect of the shared family environment often appear to be small. There are many reasons why this is so, but none of them add up to, “families don’t matter.” In fact the paper concluded that most of the difference between genes and family environment in twin studies is methodological. DNA is conveniently discrete and relatively static; environments are continuous and changeable. Behavior genetics was greatly forwarded by the existence of twins, with their simple accounting of rMZ=1.0, rDZ=.5. There is no equivalent in the environment. I proposed a little thought experiment:

How similar was my rearing environment to that of my siblings? And how similar was it to the environment of my adopted sibling, if I have one, or to the environment of my biological sibling who was raised by someone else? The apparent victory of nature over nurture suggested by the first two laws is thus seen to be more methodological than substantive. In a world in which there were occasional occurrences of “identical environmental twins,” whose experiences were exactly the same, moment by moment, and another variety who shared exactly (but randomly) 50% of their experiences, environmentalists could reproduce the precision of their rivals,and like the behavior geneticists could measure with great precision the total contribution of the environment while knowing almost nothing about the developmental processes that underlie it.

The third law says that most human behavior winds up in what is called the nonshared environment: unpredictable, even in twins. As it has turned out, the third law has continued to be true, even in the era of measured DNA and GWAS. The best polygenic scores for behavior account for no more than 10-15% of the variance. Elsewhere, I argue that the nonshared environment also turns out to be the location of free will, which Jayman contends (ridiculously) is undermined by the existence of heritability.

As for the fourth and fifth laws. The fourth law, which says that genetic effects are broken up into the minute effects of many genes, is certainly true. I think I probably disagree with the authors of that paper about the implications of that idea, but they basically get it right.

I had never heard about the fifth law until I read this post. It says that associations between phenotypes are confounded by genetic associations.  The author of the blogpost making this point, by the way, manages not to credit me for the first three laws. And although I am out of the law-numbering racket,  the presumably new one is a dumbing-down of what I have called “the phenotypic null hypothesis”: the multivariate structure of genetic variance is usually the same as the structure of the phenotypic variance. This is another one that hereditarians interpret as an indication of genetic primacy, but as usual they get it exactly backwards. If people who own dogs are more likely to spend time in parks, and it turns out there is a genetic correlation between dog-ownership and park-time, it is not because “the same genes” cause dog-ownership and park-time: it is because phenotypic dog-ownership causes park-time, and whatever genes are associated with either go along for the ride. That’s a complicated story which I’ll have to explain another time.

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.

  • Gustav Holst
    Posted at 23:17h, 29 May Reply

    Dear mr Turkheimer.

    I got this post linked to me on Twitter.
    A group of people has been discussing twin studies on Twitter the last week.
    My own effort I’ve put into trying to get some grip of the methodology itself.

    I’m a physician living in Sweden. It has been a while since I studied genetics, and I know only a limited amout about statistics, so to be honest I find some of it a bit difficult.

    Still, I wrote a comment on a scienceblog here, which I’ll translate into english, for the sake of interest:

    I’d be happy to get your comments on it, if possible.

    ”I think I’ve understood the basics of twin studies.
    I have an objection to some of them, below:

    (It’s possible I’ve missunderstood some of it. In that case, I’m sorry.)


    Basically you look at the similarity (correlation, concordance) for the group as a whole, between siblings within pairs of MZ twins and DZ twins, respectively.
    You assume that an identical genotype gives an identical phenotyp, (in absence of other simultaneous differences).
    Identical environment also causes identical phenotyp, (or at least doesn’t cause any difference).
    Different environment (in relevant areas) on the other hand causes difference in phenotype.

    Similarity in pairs of twins is ascribed to similar genotypes, and simlar environment, (called ”shared environment”).
    Differences are ascribed to differences in genotype (that is, in DZ twins), differences in environment (“individual environment”), and also errors in measurement and “coincidence”.

    For exampel: If 95 % of MZ twins are equally tall, or body height correlates in another way in the group, that is said to depend on equal genes for a part of the similarty, and to equal environment for another part of the similarity.
    Similarity here perhaps depending on genes in x percentage of them, and on environment in y percentage of them, (or some other way of describing similarity).
    (Similar descriptions can be made when comes to similarities among DZ twins.)


    Here I see a problem:
    From my point of view genes are a basis when comes to most thing regarding humans.
    All else being equal, identical genotyp as in MZ twins should therefore make a pair of MZ sibilings exactly similar, as a start.

    And – when it comes to similarity, you can not be more similar than “exactly the same”.
    Shared environment, therefore, can not convey any “extra similarity”, in MZ twins, (nor in DZ-twins).

    Eg: Two MZ twins have a birthmark, that is genetically caused. They don’t get “more equal” from being exposed to equal environment, such as dying their hair in the same colour, or removing their appendices. They remain equal.

    (They have changes qualitatively.
    That though, is something different.)

    In the sam way genotype can not convey more similarity if the enivironment is identical.
    (The MZ twins, growing up in exactly the same enviroment and being identical, dont get more similar just because both get gray hair from genetical reasons – they just change qualitatively.)

    (On the other hand one can imagine different environment causing similarity in persons that are different for genetical reasons. Eg parents studying extra with one sibling, ie one of two DZ twin sibilings, that has trouble with, say, mathematics in comparison to the other sibling.
    And one can imagine differences in genotype smothing out differences caused by environment – ie genes protecting from coronary disease in spite of bad diet in one of the twins and giving overt disease in of them both at the same age, or such.)

    To me, (as an “idealized concenpt”) all similarity in MZ twins comes from identical genes and identical environment, both at the same time/or “in parallel”.
    Both of them have an impact qualitatively, but they can not convey ”more similarity”, (nor difference).
    (All similarity can therefore, if idealized, be ascribed to genes and identical environment, but they can not be separated.)

    – Differencens within pairs of MZ twins, though, are to be ascribed to differences in environment, that has “driven” them apart.
    (And to errors of measurement, and “coincidence”.)

    (There are some side-tracks or special cases regarding this, but I’ll omit them in this comment.)


    A similar discussion can be made when comes to DZ twins.
    (I can add that one later if needed.
    Possibly there is also a bit more to say on “individual environment” when it comes to DZ twins, but I’ll leave that out for now as well.)


    An alternative model to this description is that some traits don’t have a genetic base at all, but are founded in some sort of “blank slate”, that can be dominated entirely by environmental factors.
    For instance in a subgroup of people, or when it comes to a part of a trait in all of us.
    For instance a number of MZ twins that are equally tall related to genes, and another number of MZ twins that are equallt tall because of equal environment.
    In those with “blank slate”, the high could vary more or less randomly if it were not for environmental factors that determined their height.

    To me, the current model for twin studies seem to be based on this later idea that there is a number of persons that are, or a part of traits in most that is, a “blank slate”, (- that can be dominated entirely by environment).
    I wonder whether that, in reality, is really the case.

    (My own suggestion/model, on the other hand is a bit ”idealized”..
    But still.)


    I’m not sure how they’ve reached the current model for calculating genetic inheritance contra ”shared environment”.

    It’s quite possible that I’ve misunderstood someting in the concept of twin studies.

    In my view, though, I’m rather sceptic to the paradigm that -either- similar environment -or- similar genotype causes similar phenotyp.
    I therefore wonder whether the model that tries to separate “shared environment” from genetic inheritance is reasonable.
    I would rather say that you can’t separate the effect on of “shared environment” and “shared genetics” on -similarity- between twins, through comparing correlations among MZ twins contra DZ twins.
    (In fact, it may be probable that they both convey similarity at the same time, “on top of one another”, in all MZ and DZ twins – having a qualitative effect on traits, but no additive effect to one another on similarity between siblings.)

    On the other hand, the effect of ”individual environment”, “coincidence” and errors of measurement can probably be sorted out, more or less in the same manner as today.”

    Like I’ve written, I may have misunderstood something fundamental.
    If not, it would be very interesting to get your comment on this.
    I’m happy to try to explain what I mean better, if something is unclear.

    Thaks you, in advance!

    Best whishes!

    /Gustav Holst, Karlstad, Sweden

Post A Comment