Genetics as a model for cultural understanding

[John McCreery]

Early in our lives we learn a bit of something. Unless we keep going back to it, that becomes our image of what people in that particular field know about what they study. But even if accurate when we learned it, over time that image turns to the straw of which straw-man arguments are made. If we are lucky, however, we have friends who help us catch up with what we've been missing. When it comes to biology and genetics, for me those friends are Andrew Petto (Anj to his friends) and Richard Wilsnack, who post on Anthro-L. One of their recent exchanges seemed to me like it ought to be shared. I asked their permission' and both said yes, with one caveat. I must warn you that neither is a geneticist. All they claim to be is interested observers. We should all be such observers.

First Wilsnack, then Petto.

---------------

Back in June I had a good exchange with Anj Petto about some recent developments in medical genetic research. Then I got diverted by a big trip and family reunion to celebrate my semi-retirement, then a big grant application that kept going awry, and finally some extensive tinkering with my coronary blood vessels. Finally, I can get back to writing (I hope humbly enough) about some important work in an area that I understand only the edges of. I hope this may be relevant to future discussions here, at least on issues of medical anthropology. I do recommend reading this while sipping some kind of long, slow drink, by the fireplace.

A big issue in medical genetic research is that the gene variants known to be associated with some relatively common complex phenotypes (such as autism spectrum disorder (ASD) and schizophrenia) explain very little variation in those diagnoses. Even if we scan thousands or millions of genetic variants, such as in Genome Wide Association Studies (GWAS), the explanations are incomplete. For example, GWAS studies have found no alleles of any one gene that are a major contributing factor to ASD (D Wahlsten, "The hunt for gene effects pertinent to behavioral traits and psychiatric disorders: From mouse to human," Developmental Psychobiology 54 (2012), 475-492). A recent effort to predict a relatively stable non-disease characteristic (adult height) from 294,831 single nucleotide polymorphisms (SNPs) in 3925 individuals could explain only 45% of their variation in height, and did not (could not?) differentiate which SNPs mattered most (J Yang et al., "Common SNPs explain a large proportion of the heritability of human height," Nature Genetics 42: 7 (July 2010), 565-569.)

One response to this problem has been to declare that all genetic explanation of psychiatric health problems is a sham (e.g., J Joseph, "The 'missing heritability' of psychiatric disorders: Elusive genes or non-existent genes?" Applied Developmental Science 16: 2 (2012), 65-83). This kind of polemical all-or-nothing view is understandable, and it points out that people and institutions that have invested careers and huge amounts of research funds in genetic research will strive to defend the importance and value of their work if doubts are raised about it. However, such jeremiads are also simplistic and painfully biased in their use of information.

A more interesting recent argument is that behavioral geneticists have oversimplified their analyses by assuming that all genetic influences (including factors affecting gene expression or suppression) are additive. If there are nonlinear interactions among different parts of DNA (so that the effect of the whole is greater or less than the summed effects of its parts), this might help explain the heritability of some phenotypes (such as psychiatric disorders) that the separate genetic parts cannot explain (O Zuk et al., "The mystery of missing heritability: Genetic interactions create phantom heritability," Proceedings of the National Academy of Sciences 109: 4 (24 January 2012), 1193-1198). A big problem with testing this idea is that if many gene variants have very small effects on phenotypes (such as psychiatric disorders), it is difficult and perhaps impossible to distinguish the small additive effects of genes from small interactive effects, unless we have huge, perhaps impossibly large samples of SNPs or whole exomes.

The best article I have read (accessible to educated lay people) is the one by Douglas Wahlsten cited above (in Developmental Psychobiology). He discusses the difficulties of detecting a multitude of very small additive or interactive effects of genetic components. He also raises the possibility that influences on complex behavior patterns (such as in psychiatric disorders) may be not only polygenic but also heterogeneous: a variety of different combinations of aberrant genetic alleles may influence the same diagnostic behavior pattern (see also J Sebat et al., "Rare structural variants in schizophrenia: One disorder, multiple mutations; One mutation, multiple disorders," Trends in Genetics 25 (2009), 528-535). This genetic heterogeneity underlying similar diagnoses has also been reported for cancer and lipid diseases. Wahlsten also hypothesizes that if multiple gene variants are necessary for illness or increased vulnerability to illness, and the combinations of gene variants for any given illness may vary, perhaps there is a threshold when combined small effects (additive or interactive) of many different parts of the genome are sufficient to produce diagnosable symptoms, even if no one component is always necessary. Finally, he notes that twin comparisons (monozygotic vs. dizygotic) are a dubious way to estimate heritability of faulty genes, to the extent that the fault may result from spontaneous mutations that the parents may not have passed along and that the identical twins may or may not pass on to their offspring (if any). Such spontaneous mutations, believed to have important roles in both autism and schizophrenia, would inflate the correlations of illness in monozygotic twins relative to dizygotic twins.

The idea that there are large numbers of gene variants with individually small effects in human populations is reinforced by two recent studies. MR Nelson et al. ("An abundance of rare functional variants in 202 drug target genes sequenced in 14,002 people," Science 337:6090 (6 July 2012), 100-104) report that function-altering genetic variants may be very common (averaging 1 of every 17 bases), individually rare (few people have the same variant), and geographically variable (so that the set of variants you might find in Europe would not be the same as the set of variants you would find in Africa). JA Tennessen et al. ("Evolution and functional impact of rare coding variation from deep sequencing of human exomes," Science 337: 6090 (6 July 2012), 64-69) identified over 5000 single-nucleotide variants in 15,585 protein coding genes from 2440 individuals. Almost every variant (~96%) predicted to be functionally important was rare (minor allele frequency of < 0.5%), and the diversity of genetic variants in Europeans was different from the diversity in Africans. Both authors concluded that abundant uncommon gene variants have resulted from very rapid relatively recent human population growth combined with low selection to filter out such variants (since nearly every variant has had only a small effect).

None of the articles cited here discussed how DNA material affecting expression or suppression of genes (such as by methylation and phosphorylation) would further complicate analyses of genetic effects on health and behavior (although Wahlsten does mention this problem). Also, there were no comments about environmental influences on genetic effects (as contingencies or multipliers for genetic effects), such as the identified perinatal influences on children's development of ADHD (from smoking, atopic exzema, and gestational diabetes of mothers). There was no discussion of possible compensatory effects from other genes and physiological processes in response to proteins altered by malfunctioning genes. And there was no discussion of the timing of gene variant expression or suppression, which would seem essential to understand the onset and remission of genetically-influenced disorders.

The bottom line (or some possible reasons why you should read this far): The ability of physicians to link specific genetic flaws to specific illnesses is reaching its limit. For many of the most common and complex psychiatric disorders (as well as cancers), there will be no future discoveries of magic genetic bullets that will simplify diagnoses and treatments. The best we can hope for medically is that animal studies of the physiological effects of specific genetic variants will reveal patterns at the physiological level that will provide a better understanding of disease processes. I don't know quite what to say about the implications for anthropological research, but it seems as if genetically differentiating groups rather than individuals has now become more complicated and more difficult than was once anticipated, ever so long ago.

Richard W. Wilsnack
[e-mail]


And now for something completely different ... a really serious answer!

In our discussions earlier in the year, I pointed out that Richard had uncovered the "jet lag" between the cutting edge of genetics research and the trailing edge of its practical applications ... and the even greater gap between current concepts in genetics/genomics and the common understanding even in educated people.

I think that the Zuk article (which is the only one that I have read thoroughly) gets at this well.

We go through school learning Mendelian genetics: dominant/recessive/co-dominant alleles that control the expression of a phenotypic feature. We acknowledge that there are polygenic traits, but curricula and textbooks have very little to say about them.

In the last 50 years, but especially in the last 20-30, the simplicity of those notions has been exploded, yet most disciplines outside of genetics (like medicine and anthropology and psychology, to name a few) are still stuck on the "one-gene:one-protein" model --- which is only slightly more sophisticated than the Mendelian "one-gene:one-trait" model. The upshot is that almost anything that is vaguely interesting in human development and variation is likely to be significantly more complex than any of these models.

THAT is the main justification for objections to Rushton and his colleagues; and to evolutionary psychology as a general field; and to some of the "designer" therapies based on genomics: the archaic view of how genes are inherited and expressed.

We have to admit (1) Mendel got lucky with the traits he picked; (2) may of the gene-based disorders and diseases DO seem to have one major bit of DNA (one locus) in common; and (3) the number of new discoveries in the genetics of disease that find a single locus have been declining significantly compared to the loci that seem to have an effect that is contingent on the contents of various other loci (just pick your favorite condition or disorder at OMIM and see how many loci might be involved: http://www.ncbi.nlm.nih.gov/omim).

I like to use the Bombay Phenotype to illustrate this; thanks to a student who told me that she was blood type AB, her mom was A and her dad was O. It turns out that the old ABO system needs a precursor compound added to the cell membrane in order to attach the A and/or B antigens. A defect in THAT locus causes the expressed B antigen to have nowhere to attach to the RBCs, and so the blood type will appear to be O (aka the Bombay phenotype: http://anthro.palomar.edu/blood/Bombay_pheno.htm).

The other problem is of the type: "We have met the enemy, and he is us!" The short-hand tendency to graft the Mendelian understanding (the gene FOR seed color, for example) onto the loci in our chromosomes comes out in a way so as to distort the role of these DNA sequences. I have had students come to me and say, "I have that gene for XXXX, and I am worried." After my stock response, "I am not THAT kind of doctor (and probably neither is your MD)", I usually say something like this: "Having this gene is a good reason for you to get further screening and consultation with a specialist. However, when we talk about a 'gene FOR' something, what we usually means is that here is a bit of DNA that produces some product in the body and when that DNA is altered in some way or if its products are damaged in some way, THEN a disease MIGHT result with some calculable probability."

The usual response is "Huh?" To which I usually add, "There is no gene whose main role in the body is to give you cancer. Your risk of cancer is increased, however, if that gene is altered in certain ways." Thus, BrCa is NOT a GENE FOR breast cancer (that is, its role is not to produce cancer in a healthy person); it is a locus that increases the risk of breast cancer in people whose DNA sequence appears in a certain arrangement and the degree of risk depends on the appearance and behavior of other genes in the body."

It would be better to get the public to understand that the polypeptide sequences that the DNA encodes are like Lego blocks. A DNA sequence makes one kind of block; the body takes a variety of blocks and builds the final product ... sometimes building only a scaffold, or an enabling enzyme, or something other than a final product. Additional copies of the same blocks are often used in very different compounds. If we could even get THAT across, it would jump the public understanding of genetics decades ahead of where it is now.

And then, of course, there is stuff that we do not (yet) understand. As ANY good research does, genomics research has given us both some good answers and some difficult questions. One is that with data from a whole genome, not ALL the differences in the DNA from one taxon to another is exactly the same (one locus might differ by 3%, and another by 10-15%); sometimes the differences are in those loci where we might expect to find sequences that define a separate lineage. In other words, the simple molecular clock that we have presumed needs more calibration (if you are interested and love Bayesian analysis, see almost anything recently by Cecile Ane').

It is ALL so VERY exciting!

Anj


--
Andrew J Petto, PhD
Editor, Reports of the National Center for Science Education
Department of Biological Sciences
University of Wisconsin -- Milwaukee
PO Box 413
Milwaukee WI 53201-0413
CapTel Line: 1-877-243-2823; Ext 4142296784
FAX: [phone]
https://pantherfile.uwm.edu/ajpetto/www/index.htm

[John McCreery]

Fifteen views, no comments. I wonder what that says about the composition of OAC. I suspect that our numbers include few biological anthropologists and there was too big a gap between the subject "Genetics as a model for cultural understanding" and the content of the conversation. Let me try to close that gap. I begin with the observation that any text can be read on multiple levels, for the immediate content of what is said (here observations about recent developments in genetics) or, more abstractly, as an instance of a way of thinking about the world that transcends particular disciplines. Given my own interests, I was reading on both these levels, with, however, a focus on the second.

I was particularly struck by Petto's description of what he called Wilsnack's discovery of the "jet lag" between the advancing frontier of genetics and most, even educated, people's understanding of genetics. Thus, we continue to see fierce debates about the role of genetics in shaping human behavior in which genetics is supposed to be the simple one gene-one trait (or behavior) mapping exemplified by claims that X has discovered "the gene" for Y. Meanwhile, geneticists have already discovered that such claims are, in most cases, far too simplistic. But how different is this, I ask myself, from other arguments in, for example, politics or economics, where this or that decision or factor is seen as "the explanation" for the events in question?

I found myself thinking of A.N.Whitehead's observation in _Science and the Modern World_ that the scientific revolution of the seventeenth century had bifurcated reality into two incommensurable parts, the invisible reality of atoms and forces, which was taken to be the really real reality, and the sensuous stuff of everyday life, which was taken to be epiphenomenal—the central proposition in the whole business being that if we had a complete understanding of the invisible really real we would also understand everything else.

In Gerald Weinstein's _Introduction to General Systems Theory_ I found a refinement of what I had learned from Whitehead. Here reality was divided into three zones. The really real of the seventeenth century physicists had divided into (A) things explainable with simple mechanical models and (B)things explainable by probabilistic models, which still left (C)everything else, for which no mathematics were available. In General Systems Theory, however, A and B are not taken to be more real than C. The difference is epistemological, in how we know, rather than ontological.

Now, serendipitously, I find myself reading Nate Silver's new book _The Signal and the Noise_, in which Silver points to two typical flaws in data-driven research. The first is omitting relevant data, the second overfitting, which is to say precisely describing noise and distracting attention from the core of what is going on. In the Petto-Wilsnack discussion, omitting relevant data, the influence of the other genes, proteins, processes that influence a gene's effects is the issue. What, I wonder, would be overfitting in this context? How does one trace the mechanisms by which culture affects the operation of genes and vice-versa, without getting lost in irrelevant detail? How do we separate the signal from the noise? Not, I'm afraid, by trotting out old Nature-Nurture arguments that are, in their very nature, categorically flawed.

[John McCreery]

Anj Petto has replied to my comment on Anthro-L

-----------

I think that overfitting is exactly what concerned some of the authors of the pieces that Richard mentioned in his post. A sort of overfitting is what Rushton's work does; and the problem is using an outdated view of complex interactions among explanatory variables. 

There are two aspects of Rushton's work that exemplify this problem. (1) The reliance on the outdated notion that a single variable --- so-called 'g' --- is an underlying component of the scores we get with IQ tests. At best it seems to account for 50% of the variation in the data and seems to have some correlational stability across disciplines. This is useful, of course, in the process of examining the data (and remember, the original tests were to identify differences in ability between "normal" and "retarded" children --- not to quantify intelligence in its own right). 

The nature of principal components analysis is such that one or a few of the axes has strong(ish) associations with an array of variables. The problem is that the SIZE of the association (the r^2, as it were) is not the same thing as the significance of the impact. An axis can have a small effect of the r^2, but have a powerful effect on the outcome of the model; and we often do not look more deeply into those 5th and higher axes that may contribute little to the overall r^2, but which can be important in understanding the differences between two subsamples that otherwise see the same on most of the other variables. 

(When we used a factor analysis to render teachers' responses to a questionnaire after their participation in a series of professional development workshops, we found 5 axes for 16 variables; 30.9% of variance in the first axis, and 38.8% in the first 3 axes all together. Several of the variables did not load on any of these axes; a couple of them aligned with more than one --- often in opposite ways. So, our remaining questions were "What about those variables that did not make it into the top 5 axes? Are they important; and how do they affect the other relationships?" [Petto AJ, Patrick M, Kessel R. Emphasizing Inquiry, Collaboration, and Leadership in K–12 Professional Development, in Yager RE, ed. Exemplary Science: Best Practices in Science Teaching Today. Reston (VA): National Science Teachers' Association, 2005. pp. 147–60.] So, if we were to look for the variance unexplained by this analysis, would we presume that the genetic differences among the participants were the cause? After all, there WERE significant differences between elementary and high school teachers; or teachers and administrators or professional staff.)

The problem is not that there isn't some set of associations that one can relate to a genetic component; the problem is that the way that 'g' expects that genetic component to be expressed is quite decidedly outdated. If, instead of looking at IQ across Rushton's races, we were to look at wealth accumulation, we would see that persons of color accumulate less wealth, even when we control for income, education, age, occupation, and so on (or, as Chris Rock explains" "Shaquille O'Neal is rich; the white man who signs his check is wealthy"). Could anyone seriously argue that our controlling for these socieconomic variables leaves us to conclude that failure to accumulate wealth is genetic?

Yet THAT is the model, and it is one typical of the research in the first half of the 20th century when genetics was simple and all we had to do was identify and eliminate all the environmental influences so that all that was left was the effects of the genes: V(total) = V(genetic) + V(environment). Problem was that there was always a pesky 'error' term ... the variance unexplained by the model. 

(2) The second issue is like unto the first. There are things that we CAN measure and analyze and those that we SHOULD. It is good when the two sets overlap completely; it is rare that they do. So, if we control for a whole raft of socioeconomic, demographic, politicolegal, and other culturally informed variables and are left with a certain amount of unexplained variation ... what is the rationale for our concluding that this amount corresponds to a genetic input? It only really tells us that these other variables leave some component unexplained.

We need an explicit model of genetic influences on complex behaviors and their consequences (such as social status, wealth, and so on) that can be tested. However, just to take the negative evidence --- the variance left over after all the main factor axes have been correlated with outcomes --- can significantly misinform the analysis. As Jon Marks is fond of saying: "Show me the genes!" If we want to argue that the data support a genetic explanation, then it is absolutely necessary to have a contemporary genetic model for this influence AND to analyze the actual genes of the subjects. This was difficult to do a decade ago, but it is quite routine now --- the only problem is that we don't really know where to begin looking among the 3 billion base pairs of DNA, but at least we could start by asking a simple question: Are there significant genetic differences between members of one population and members of another whom we presume to have genetic differences in some feature (and, are those differences consistently and persistent associated with one population more than the other)? 

Such research CAN be fruitful and illuminating. For example, there is some interesting research in susceptibility to hypertensive cardiovascular disease in people of African descent that finds such an association (though research is still ongoing, it appears that it was not slavery that made the difference, but that European populations underwent a genetic bottleneck and lost some alleles related to sodium conservation that are still found in Africa).

So, overfitting with genetic models is the use of whatever left over variance we cannot associate specifically with sociocultural and demographic variables and call it a 'genetic' influence.

[Huon Wardle]

I looked at this and was puzzled by the idea of genetics being a model for culture. We have tried this with 'memes' and in my opinion the results are fairly unhelpful. For better or worse the word 'gene', like the word 'culture', has escaped the hands of the specialists and is out there being used, often in politically charged ways, typically in ways far wide of its in initially intended or current uses, though with a trace of what was originally meant. Looking back, the Human Genome project evoked immense triumphalism and many fears in equal measure; so it is not surprising that it should have spawned all sorts of ideas about the power of genes - 'it is in the genes...' Social scientists are quick to see any explanation from the side of heredity with extreme suspicion - after all there is a history there Galton, Huxley, eugenics, scientific racism etc.

As regards schizophrenia and IQ, over the years claims about the heredity versus environment as causes have been batted back and forth. When we shift from a clearly demarcated biological phenomenon like breast cancer to what are inherently behavioural, social, phenomena like schizophrenia or IQ we are entering an interpretive quagmire. It is certainly useful to note that if schizophrenia has a genetic cause at all then it derives from a heterogenous array of interactions, but that is really only scratching the surface since we then need to do the same kind of multirelational exploration in the field of social behaviour; thereby recognising interpersonal behaviour as causative in its own right. Unless we want to go on 'overfitting' we have to give weighting to behaviour and genetics and not see one as causing the other unidirectionally. After all there is a lot of work now on gene expression, epigenesis and so on.

[Keith Hart]

Entering this thread at all is a lot of work and, as Fran said, the OAC is straddling the divide between social media and academic discussion. Then most of us have a position on nature/nurture that is unlikely to be moved by the way this question is framed. I have a particular line: the American ideology, as presently constituted, draws heavily on genetic explanation for evolutionary and cognitive science. This is another way of escaping from history into a vision of nature that has no place for cultural decline. But at the rate of posting here right now, someone may engage in the next few weeks. Maybe those of us who post often should retreat for a while and see what happens.

[John McCreery]

Huon, Keith, my bad. I knew I was going a step too far with that subject line. But gentlemen, I have to wonder if either of you read the exchange between Wilsnack and Petto with an open mind, looking for fresh information, instead of just scanning for confirmation of existing prejudices. The fact is that both are good guys, totally on our side when it comes to refuting crass racism "supported" by bad science. Where they differ from us, and I suspect, most here on OAC is that they actually know something about good science. Had you read a bit more carefully you might have realized that the topic under discussion is why racist science is bad science, a combination of simplistic ideas (one gene=one trait) and sloppy methods (assigning all the unexplained variance to a single cause labeled "g" for "genetic"). Then, you might also have observed that as they describe the current state of the art in genetics, it is all about processes, interactions, and function in different contexts, which is—take note—what sound hermeneutics/cultural understanding of words and symbols and artifacts is also all about.

That was what caught my attention, which is primed, I like to think, by interest in the history and philosophy of science. Thus, where you saw John bringing up that tired old nature-nurture debate again, I saw myself observing a convergence in the way both scientists and scholars in other disciplines are coming to see a world which is messy and probabilistic at every level we examine it and the possibility of conversations in which people from different disciplines share common concerns discussed in a common language, instead of huddling behind the barriers that disciplinary prejudice erects. I suspect that I am foolish to be so optimistic.

[Francine Barone]

I get where John was coming from when he posted this exchange. The title sufficiently grabbed me and it was a good intellectual challenge to read the Petto/Wilsnack posts with it in mind. I agree with Huon's post, the first part of which covers one of the main points I was going to make about the popular usage of "culture", only expressed more concisely than I probably would have managed.

I imagine, though, that the reason for so few replies out of 30 views is the density of the original exchange and a lack of guidance from the OP. I wonder how many non-medical/biological anthros ("actual" non-geneticists) made it past the middle of Wilsnack's post. Not that non-specialists couldn't find their way, but their first impression might be that specialist knowledge is necessary to arrive at more abstract connections between genetics and culture given the source material. You have to persevere to the end before figuring out that it's not necessary.

I would never suggest that we delve any less deeply into topics here, but I wonder sometimes if we forget the mixed audience we purport to appeal to. In this regard I feel we can work on framing new threads in ways that will attract the widest pool of participants. One of the most difficult balances for our academic/social network comes down to the right amounts of catchy, public appeal and rigorous, in-depth analysis. A tough one for sure; maybe impossible. In this case, John's follow-up post explaining his rationale for posting - or at least some key sign posts - could have come earlier in the intro, IMO. Perhaps that would have helped more "quick glancers" to engage with the text and not give up before reaching the end.

[John McCreery]

Francine, good points, taken on board. My original thought was to allow Wilsnack and Petto to speak for themselves, reserving my own musings for the follow up. I can see where that wouldn't work so well for readers who found Wilsnack tough sledding. Also, I would never have come up with what I wrote in response to Huon and Keith, if I hadn't felt challenged by their remarks. 

[M Izabel]

The title grabbed me, but I saw no culture in the post but all biology.  I expected it to be about a genetic structure as a model for culture.  

I do wonder why the trend in social sciences is going biological.  Do social scientists need to be taken seriously like how biological scientists are popularly read and followed?

I just want to share that back home we have a concept of "mana" (inheritance) that is used biologically in our language and culture.  When a criminal is a son of a criminal father, we say, "mana ka sa tatay mo" (you are like your father or you follow your father's footstep or you are what your father is). We also have this proverbial statement, "Hindi bumubunga ng bayabas ang mangga" (Mango doesn't bear guava).  I think this cultural logic is used in kinship formations such as in marriage where the (psychological and behavioral) backgrounds of a groom and a bride are, generally, important to their families before their wedding is set.  They are afraid of genetic contamination locally called "masamang lahi" (bad gene or undesirable trait)..  

[John McCreery]

For what it's worth, I am no expert in these areas, I would suggest the following scenario.

Ideas about inherited characteristics may be a human universal. I personally know of no people who don't have them. The specific content varies, from full reincarnation (Tibetan lamas)to mixtures of lines ("He's got his Uncle Harry's ears"). As in the Philippines, the specifics of these ideas tip ideally have a lot to do with the way in which kinship and other social relations are conceptualized.

Mendel's pea experiments demonstrated the possibility of accounting for variation in terms of "genes." It wasn't, however, until Watson and Crick described DNA that scientists began to get inside genes and explore the chemical processes that account for their effects and how they interact. Even the interactions have turned out to depend a lot on the environment in which they occur.

Meanwhile, however, the idea of "genes" was picked up and popularized as a way to give "scientific" grounding to racist and other ideas related to trait inheritance. While science was moving in the direction of more detail and digging into mechanisms (like physicists moving on from atoms to subatomic particles to quarks and strings) the popularizers ignored all that.

So, the current state of affairs is one in which ideas derived from the earlier, crude "atomic" ideas about genes, bad science based on those initial crude ideas, earlier folk ideas about inherited characteristics, and the new post-DNA biochemistry of genes are all jumbled together. The unfortunate thing about a lot of our typical nature-nurture debates is that the early crude science and the later bad science are conflated in a bogeyman while the new post-DNA biochemistry is ignored by people who are simply or willfully ignorant of its existence.

Turning, however, to the biological trend in current social science. One crass but plausible hypothesis concerns funding. Biology taps into the relatively lavish funding still available for projects that can claim some relevance to medical, public health, or psychological issues, while funding for social science and humanities projects that cannot claim this relevance declines.