neuro folly: the quest for biological bases for politics

The sociological fad for seeking straightforward biological bases for complex social phenotypes has spread in recent years to political science. A recent collection of articles and blog posts rehashes the rhetoric that garnered widespread criticism in sociology. Essentially the scholarship establishes an entirely noncontroversial position–that biological influences on political ideology are “nonzero”–and spins this into the much broader, and indefensible, claim that inter-person variation in biology is (a) unidirectionally causal; and (b) a significant source of variation in political ideology.

Erik Voeten writes here and here about Fowler and Dawes’ study (gated, sorry) of the effects of specific genotypes–two genes in particular–on voter turnout that exemplify the common rhetorical leap. The study he refers to can reasonably be held to demonstrate that “these two genes might matter a little in influencing voter turnout”, but far from the causal home run implied by the article’s title (“Two Genes Predict Voter Turnout”). Essentially, these two genes — MAOA and 5HTT — show modest effects on voter turnout, likely because of their role in regulating serotonin, a crucial and multipurpose neurotransmitter. Serotonin is implicated in a huge array of social and emotional phenomena, and these likely cluster in ways that result in modest but discernible effects on voter turnout.

Verhulst et al. argue in their recent AJPS article that personality traits do not in themselves cause political ideology but “the primary connection between personality traits and political ideology rests on common genetic precursors of each” (p. 48). Similarly, Hatemy and McDermott call for bringing physiological psychology (which “attempts to explain behavior by studying the physiological processes that control it,” p. 17) into the study of politics.  Still, they hedge:

There is no single “gene” that “causes” any particular social behavior. Rather, the nuanced interaction of countless genetic markers contributes to the regulation of complex neurobiological pathways designed to receive, and remain exquisitely sensitive to, a vast array of both internal and external inputs. Genetic factors may indirectly predispose  certain individuals toward particular traits or characteristics under certain conditions through genetic expression  (epigenetics), and the regulation of hormones that change states of emotion, cognition, and biochemistry which begin  with some social or environmental stimulus. And these environments can be global, local, or even within the cell.  However, even in the case of a seemingly straightforward characteristic such as height, many different factors, such as in utero influences and childhood nutrition, combined with epigenetic mechanisms, contribute to any given phenotypic  outcome…. attitudes and behaviors represent the expression of series of neurobiological processes informing cognitive and emotive mechanisms which interact with the environment. (p. 21)

Similarly in Smith et al.’s AJPS article:

What we actually observe is a pattern of relatively stable MZ [monozygotic twin] correlations…. at more typical levels of genetic similarity, environmental factors contribute more to political similarity. Our analyses are thus consistent with a susceptibility jmodel, i.e., the heritable underpinnings of adult ideology operate largely by creating varying levels of susceptibility to specific sorts of ideological appeals.

The Smith et al. article’s title promises epistemology, but little is to be found. Mostly it considers the so-called Equal Environments Assumption (EEA). This is the assumption that differences in environmental experiences (pre- and post-natal) between MZ (“identical”) and DZ (“fraternal”) twins do not affect outcomes of interest. Generally defenses of the EEA, including this one, do not deny that there are important and systematic differences in environment but claim that these are insufficient to explain greater similarity between MZ than DZ twins. However what would constitute “sufficient” causes is not specified, and there is reason to think, from economics as well as life-course sociology, that small causes may iterate into large outcomes. Still, the conclusion is hard to quibble with: “nothing in the EEA critique credibly supports [sic, I assume] a claim that genetic influence on political temperament is nonzero or non-trivial. ”

An important defender of the biopolitical paradigm, Larry Arnhart, sums up his position thus:

biopolitical science must be a complex science of emergent evolution that embraces not only genetic evolution but also the cultural history of political regimes and the individual history of political actors.  Human genetics constrains but does not determine human cultures and human judgments.

Once again, this is a nearly entirely unobjectionable, but also not very interesting, claim about genetic influence. The rhetorical pattern is very common: assert that biological basic processes are important precursors of political ideas or behavior, accuse skeptics of being head-in-the-sand anti-science types (as Kieran so well undermined), and then demonstrate a minor finding that violates epistemological principles about causality and holds that the environment remains likely the dispositive cause.

I do not doubt that biological factors influence social and political outcomes. However, I think:

  • These factors are not necessarily unidirectional; for example, environmental cues likely affect the production and uptake of serotonin and, therefore, the behavior of genetically-cued behaviors including but not limited to voting. So the relationship between environment and genes is almost certainly not a simple interaction (as in GxE) but likely a chain of indederminate length of genetic and environmental co-causes.
  • Because of this chain character, the very practice of estimating heritability is not just imprecise but literally, epistemologically, impossible. All complex behaviors are 100% genetic and 100% environmental, and the task of partitioning variance is therefore an exercise in shoveling fog[1]. To extend Arnhart’s statement: it is impossible to contemplate the practice of voting without a whole host of common human traits that are certainly evolved (the capacity for causal reasoning, vision, locomotion, opposable thumbs, and so on). It is also impossible to contemplate the practice of voting without a historically-contingent regime of subjectivity, individuality, etc., that is certainly discursive and environmental. Each has conditioned the other, iteratively, over the course of human history.


[1]I got the metaphor from Courtney Bender’s The New Metaphysicals which has nothing at all to do with this topic.

Author: andrewperrin

University of North Carolina, Chapel Hill

11 thoughts on “neuro folly: the quest for biological bases for politics”

  1. “The sociological fad?” Biological approaches to behavior are still wildly unpopular in sociology. Let’s be empirical: how many articles in AJS/ASR/SF/SP are biological? A few and they are mostly written by your colleagues.


    1. I’d be happy to stipulate that the size of the fad is small. Also that my sense of its size and importance may be the result of pluralistic ignorance due to my departmental location.


  2. Serious comment: If you read twin/sibling studies closely, the variance is broken down into variation due to genes, variation due to family environment, and “other.” The “other” is often huge. That leaves a huge amount of space for culture. These studies do not exclude culture-> biology causation and the other issues you mention.

    BTW, I think it is a cheap shot when people accuse critics of being anti-science. But having taught multiple cohorts of soc grad students, there is truth in the stereotype. The types of people who sign up for sociology really, really resist biological accounts of behavior and insist that there is no link between genes, hormones, etc on behavior. There are people who do actively resist the modest hypothesis that biological factors are a factor worthy of attention.


    1. I’m not sure I’ve seen this breakdown – the more typical I’ve seen is heritability, nonshared environment, and shared environment. Culture may play roles in both shared and nonshared environments, so certainly these studies don’t preclude the (additive) effect of culture. Hoewver I do think they preclude the reverse-causal (or, more appropriately, causally complex) pathways because they assume causes are fixed at the point of measurement. In other words, the degree of heritability varies between populations, but not between individuals within a population because heritability is *defined as* a characteristic of a trait within a population.

      I do not doubt that there is a large and vocal group of head-in-the-sand sociologists who reflexively deny biological effects. But the main point of my post–which I still believe to be true–is that the “worthy of attention” part remains still to be demonstrated: that fairly substantial investments of time and money, and enormous investments of intellectual power, have produced little value in terms of distinct effects of biological variation on social outcomes. Thus the accusation of being anti-scientific is certainly unfair.


  3. “The sociological fad for seeking straightforward biological bases for complex social phenotypes has spread in recent years to political science.”

    I am skeptical. The DNA research in poli sci did not draw on nor was it inspired by sociologists. It draws on no sociological ideas, no sociological methods. I checked the foundational articles and they cite no sociologists (except to acknowledge a dataset). So there is no tangible evidence that anything diffused from sociology (no less a fad) to poli sci as far as this research goes. Having read in this area for about a decade, I think the influence flowed from behavioral genetics to poli sci. The citations seem to bear me out on this.

    More importantly, there is no sociological fad (in the first place) for “straightforward biological explanations of complex social phenotypes.” It’s the opposite — the few people I know who work in this area are out to show the opposite: the goal is to develop convincing evidence that genetic factors and social context combine in complex ways. We emphasize the complexity of social settings and, because of this complexity, the complexity of any biological explanation. This is thematically the opposite of much of the poli sci work, which wants to show how specific gene candidates are associated with political orientations — in other words, explanations that do not reference the complexity of social settings. There are a few exceptions in sociology, but these very few folks (I can actually think of only one!) don’t publish their stuff in sociology.

    There is giant world out there of people who study genes and behaviors. That world would happily proceed without any input from sociology at all. I myself would prefer if sociologists try to influence this research — to keep issues of social context (its conceptualization and measurement, an appreciation for its nuance) in the foreground.

    And one quick other note: this research is extending beyond gene candidate studies — GWAS and expression studies (which are kind of a hybrid GWAS/candidate approach). The study of genes and behavior has made mistakes along the way, learned alot (I think), and is building. That’s science.


    1. Both Jeremy and Mike are far more involved in, and knowledgeable about, this subfield than I am, and my goal was not really to (re)-introduce the debate about biological explanations within sociology. My knowledge of the subfield is nearly entirely as a consumer, as I enjoy reading and paying attention to the relevant research in sociology, behavior genetics, and genetic psychiatry, as well as political science, but I do not claim expert or producer status! And I’m certainly happy to stipulate that the, ahem, genealogy of the political science studies doesn’t lead back through sociology.

      I do think that the ontological assumptions in the literature–even in sociology–continue to rest on the assumption that the biological is more basic than the sociological, and that therefore we should understand biological causes as being *refracted through* social contexts, but not being fundamentally caused or changed by those contexts. Consider, for example, a phenotype of adolescent heroin use. This could be the result of a straight GxE interaction (say, MAOA and the availability of heroin). It could also be the result of a GxExE interaction (a gene “for” disobeying parents, an environment of parental disapproval, and the availability of heroin). Unobservable and entirely plausible causal chains of virtually any length of strung-together G and E elements can be drawn, and just-so stories concocted to explain them. Crucially, not only the magnitude but the direction of the G elements in such a chain is are conditional on the E elements (and vice versa). And both the magnitude and the direction may reasonably change over many iterations.

      I think the GWAS and even gene-expression studies fall into precisely that trap. Imagine if a student were to come into her advisor’s office, excited about a new study: “I have a dataset that includes 20,000 people and about 300,000 independent variables. I don’t know what any of these independent variables really means, but they do seem to covary sometimes. What I’d like to do is grab some dependent variable and see which of the 300,000 independent variables predicts it. Then once I’ve figured that out, I’ll see if I can figure out what causal pathway there might be to explain that prediction. What do you say?” I know what I’d say, and I know what any other sociologist would say — if those independent variables weren’t genes. But since they’re genes, they enjoy an undeserved privileged status as causes precisely because of their association with Science. (BTW, I suspect that this is the reason why genetic psychiatry hasn’t come up with much in its GWAS studies either, and in general successful GWAS outcomes are those with very specific biological *dependent* variables.)


      1. The analogy I had for GWAS ~2 years ago was that it was a little like trench warfare, when generals would send 10,000 soldiers over the top, and then when that wouldn’t work conclude that the next step was to send 100,000 soldiers over the top. But I think evidence has been coming together that GWAS can be informative if one can forge the confederacies needed to get the sample size. GWAS of schizophrenia, for instance, looks stronger now than it did in the 2009 article you linked to above.

        The part from GWAS that I think is especially exciting is the development of techniques to do analysis of unrelated individuals that allow for more sophisticated forms of the variance decomposition that we normally associate with twins. This is sometimes called GCTA. Multivariate GCTA is possible, and I think has even now been done, and there are some pretty clear social science applications to the idea of being able to evaluate the extent to which the heritability of two different behaviors/traits are due to the same loci or different loci. Distinguishing among some of the alternative “just so stories” you pose provide examples.

        Some big questions for how all of this goes forward are essentially sociological: (1) to what extent will medical datasets with GWAS play ball with social science investigations and (2) to what extent will social science datasets play ball with one another. I can readily see a world where the social science datasets that end up with genome-wide data don’t cooperate well with each other and the end product is a lot of underpowered stuff.


Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your 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