Genetics, Review
comments 14

Giving Genes Their Due, But Not More

A review of Behaving: What’s Genetic, What’s Not, and Why Should We Care? by Kenneth B. Schaffner. Oxford: Oxford University Press (2016), 304 pages.

No one gets anxious about using genetics to help explain a medical disease like cancer or heart disease. But using genetics to help explain a normal behavior like aggression, or a psychiatric disorder like depression, can be an entirely different story. At first blush, this difference in response to using genetics to explain different features of the same animal seems odd. After all, it’s not as if medical geneticists, on the one hand, and behavioral and psychiatric geneticists, on the other, employ different research methods. The difference, of course, is that the behavioral and psychiatric geneticists investigate features of ourselves that we take to be central to our humanity: our ways of acting and being in the world. To use genetics to try to explain those features elicits the anxious question, is human behavior genetically determined?

Few people have been thinking about that question for as long, or with as much devotion to the scientific facts and philosophical subtleties, as the philosopher of science, Kenneth Schaffner. In his magisterial, wise, and succinct new book, Behaving, he disentangles its two separate but related components. The first, which he devotes the lion’s share of the book to illuminating, concerns reductionism: specifically, can behavior be reduced to genes? No, it can’t. But it can, at least in principle, be reduced to, or explained in terms of, a mind-bogglingly large number of variables — including genes — which interact over time. The second concerns determinism: even if genes alone don’t determine behavior, does the fact that behavior is determined mean that freedom is an illusion? No. But it does mean that we have to jettison the sort of freedom that children sometimes imagine — freedom untethered to our bodies and histories.

In the course of decreasing the anxiety associated with genetic determinism, Schaffner’s book also decreases the anxiety associated with the fantasy of “designer babies” — a fantasy which depends on the notion that just by “editing” genes we can produce any trait we want, from great athleticism to great intelligence. By dispelling this wildly simplistic notion, Schaffner’s book serves not only as an anxiety reducer — or “anxiolytic” — but also as a “mood stabilizer”: it helps stabilize the mania that can afflict those who envision the Human Genome Project as the key to the future of medicine. Schaffner provides a balanced account while never losing sight of what has been and will be achieved by using genetics to explain medical, behavioral, and psychiatric traits — especially if integrated with insights at myriad other levels of analysis, from the genetic and neuronal to the psychological and social.

A Judge and a Behavioral Geneticist Have a Conversation

Schaffner begins with three Socratic dialogues (minus any Socratic snarkyness or dead ends) that elegantly introduce the basic concepts and methods of behavioral genetics. They are worth rehearsing here. The dialogues feature a Behavioral Geneticist and a fictional Judge. Based on the breathless headlines she’s read over the years, the Judge anticipates that she will increasingly confront the results of behavioral genetics research in her courtroom. This provides the Behavioral Geneticist with a pretext for explaining how such results can — and cannot — help explain human behavior, and how such results are — and are not — relevant to everyday understandings of behaviors like aggression, traits like performance on IQ tests, and disorders like ADHD. (Because there is no difference between the concepts and methods of behavioral genetics and psychiatric genetics, from here on out I will use “behavioral genetics” to include the use of genetics to illuminate behaviors and traits, whether or not they are associated with a psychiatric diagnosis.)

Two radically different sorts of investigation are undertaken by behavioral geneticists, and the dialogues introduce a basic but crucial distinction between them. The first uses “classical” methods to demonstrate that genes help explain observed differences in human traits and behaviors, whereas the second uses “molecular” methods to determine which genes or genetic differences are generating those observed differences. The distinction is important — the distance is enormous between being able to say that a trait “is genetic” and being able to say which gene variants are contributing to the emergence of that trait (much less being able to say how they are contributing).

The basic idea for the classical method has been around since the pioneering statistician and father of modern eugenics, Francis Galton, published “The History of Twins” in 1875 — long before anyone knew anything about DNA. In its simplest contemporary form, geneticists compare identical and fraternal twins on a trait of interest, whether heart disease, schizophrenia, or performance on IQ tests. The first premise of such investigations is that identical twins are nearly 100% genetically similar and fraternal twins share on average only 50% of their genetic material. The second premise is that identical twins and fraternal twins are raised in equally similar environments. If one accepts those premises and observes that genetically identical twins are more similar with respect to some trait than fraternal twins, then one has reason to make the simple but profound inference that genetic factors help explain why the identical twins are more similar to each other than are the fraternal twins. Over time, by deploying ever more sophisticated variations on that basic logic, behavioral geneticists have demonstrated that identical twins (whether raised together or apart) are not only more similar with respect to traits like height and weight and heart rate, but are also more similar with respect to traits like depression, schizophrenia, aggression, and intelligence.

As Schaffner’s Behavioral Geneticist patiently explains to the Judge, such classical studies produce what are called “heritability estimates.” These are the numbers that are invoked when it is said that depression “is 40% genetic” or that intelligence “is 60% genetic.” They are estimates of how much of the variation with respect to a given trait in a given population can be attributed to variation in genetic factors and how much can be attributed to variation in environmental factors. However, in a different environment the observed variation can be different, and thus so can the heritability estimates.

To say that heritability estimates can be different in different environments is not to say that heritability estimates tell us nothing! (Indeed, how our genes can affect the environments we choose is an area of behavioral genetic research.) An old but ever-relevant example of how much heritability estimates can tell us comes from the 1960s, when behavioral geneticists used classical studies to discredit the then-popular idea that schizophrenia and autism were due solely to bad environments — in particular, to “refrigerator mothers.” The good news is that these studies helped relieve already-devastated mothers of the burden and social stigma associated with believing that their mothering had caused the disease in their child. The bad news is that the knowledge gleaned from those classical studies does not help diagnose or treat — much less prevent — a disorder like schizophrenia. To go from noticing that genetic differences were making a difference to knowing which genetic differences were making a difference, geneticists had to move from the classical twin methods to the modern “molecular” methods.

The Genome: A “Molecular Crystal Ball”?

This move only became possible in the second half of the 20th century, when researchers began to understand the molecular structure of genes and how to map and sequence human genomes. Indeed, the purpose of the Human Genome Project (HGP), which officially launched in 1990, was to map the genome and to specify the sequence of the base pairs, the As, Gs, Cs, and Ts, that are the building blocks of genes. The fervent hope was that knowledge of those sequences would lead rather quickly and directly to understanding and treating human disease. In reflecting back on that time, the geneticists Linda and Edward McCabe speak ruefully of the dream that an individual’s genome would be like a “molecular crystal ball.”

This idea of identifying “genes for” diseases made intuitive sense. After all, one year before the official launch of the HGP, in 1989, Francis Collins — who would go on to direct the National Human Genome Research Institute and who now directs the entire NIH — did co-discover “the gene for” cystic fibrosis, which constituted a prime supporting case in point for the idea dubbed OGOD: One-Gene-One-Disease. If a rare medical disorder like cystic fibrosis could be caused by one gene, then maybe common medical diseases like heart disease could, too. And if common medical diseases could be caused by single genes, then maybe the same was true for psychiatric disorders and behavioral traits.

Sure enough, in the 1990s, articles in the scientific and lay presses announced discoveries of “genes for” everything from bipolar disorder to aggression. But as Schaffner’s Behavioral Geneticist tells the Judge, those findings (which sparked the Judge’s initial interest) could not be replicated. “Genes for” diseases like cystic fibrosis and Huntington’s and Tay Sachs were exceptions to the rule. “Failures to replicate” reminded geneticists of the yawning gap between discovering that a trait “is genetic” and figuring out which genes help explain it.

Genetic Reductionism: A Panacea or a Boondoggle?

One of the fascinating features of Schaffner’s book is his commitment to telling the story of how he came to reform — not renounce — his own vision of reductionism. When he began his career in the 1970s, he resonated with the hardcore genetic reductionists, who dreamt that understanding the operation of genes would be a panacea: a cure for our ignorance with respect to how disease and behavior come into being. But already at that time people who called themselves developmentalists (such as the much-discussed evolutionary biologist Richard Lewontin) were challenging that dream, suggesting that, especially in the context of behavior, genetic reductionism was a boondoggle.

To understand how Schaffner arrived at a middle path, it helps to understand the developmentalists’ challenge. According to Schaffner, that challenge boils down to five core concepts, two of them helpful and three overstated. The first helpful one concerns “contextualism” — the idea that genes do not have inherent meaning, but only acquire meaning “in context with other genes, and in the environment that is cellular, extracellular, and extraorganismic” (p. 95). The other helpful (or at least wholly unobjectionable) core concept is “nonpreformationism” — the developmentalists’ rejection of the very old idea that genes contain within them little copies of the traits with which they are associated.

As for the overstated ones, they include the core concept of “parity” — the idea that genes have no more explanatory power than many other features of the organism and environment. Schaffner dismisses this as an exaggeration, at least insofar as it ignores the extent of our current understanding of the molecular structure and function of DNA. “Unpredictability,” their fourth core concept, is also exaggerated: genes can contribute to some predictions. As for the developmentalists’ fifth concept, “indivisibility,” Schaffner reminds us of the extent to which reductionism can make incremental progress in “dividing” behavior into analyzable components.

To better illustrate his revised vision for reductionism, he introduces the humble roundworm, a wonderful organism for research purposes precisely because we have such highly detailed knowledge of its genes, neurons, neuronal connections and circuits, and of the typical behaviors it engages in during its short life. In his characteristically even-handed way, Schaffner actually begins his account of worm behavior with one of those exceptional cases that can mesmerize journalists, pop psychologists, bioethicists, and others — a case where mutations in a single gene do indeed appear to be the necessary condition for a behavior: specifically, in this case, for determining whether a roundworm eats alone or in groups. In other words: one gene appears to determine the worm’s dining preference!

But then the remainder of his discussion of the roundworm illuminates what’s wrong with the One-Gene-One-Behavior idea — and more generally, with the One-Gene-One-Disease (OGOD) idea. To show why the “gene for style of eating” example is an exception to the big rule of thumb that behaviors cannot be reduced to genes, much less to single genes, Schaffner introduces eight smaller “rules.” These emphasize the interactions, occurring on multiple levels of analysis (from genes to neurons and nutrients), which change over time, and which shape and are shaped by the cellular, extracellular, and extraorganismic environments.

For example, “social deprivation,” he patiently explains, can adversely affect even the development of worms. Those raised in isolation were slower to respond to taps on the plates that constitute their environments (the “tap withdrawal reflex”), were physically smaller, and had delayed development — and the delay was correlated with the altered expression of a gene coding for a protein involved in the tap response. Schaffner quotes the researcher’s conclusion: “Experience … can alter both gene expression and the structure of the nervous system” (p. 92). Even in the roundworm, there is no “gene for” the tap response; instead, the tap response is the result of a complex network, including, at a minimum, genes, neurons, and environments. If we hope to explain behavior, then, according to Schaffner, we need a “network perspective.”

If this “network” type of genetic explanation holds for most behaviors, including even more complex organisms than worms and fruit flies, such as mice and humans, it raises barriers both to any simplistic type of genetic explanation, and the prospects of easily achievable medical and psychiatric pharmacological interventions into behaviors (ital. added, p. 95).

In other words, to appreciate the leap from genes to worm behaviors should put us on notice that there will be even more “barriers” in going from genes to human behaviors, disorders, and diseases. The once-intuitively plausible idea of the genome as a molecular crystal ball has come to seem quaint.

It is essential to recognize, however, the difference between the notion that behaviors can be reduced to the operation of genes and the idea that behaviors can be reduced. The former notion, according to Schaffner, is wildly inaccurate, but the latter is not. The fact that we can’t achieve what he calls “sweeping reductions” of the sort first fantasized about at the start of the Human Genome Project does not mean that the enterprise of reductionism is a bust. It means, among other things, that we need to accept the fact that, in complex systems, we should expect what he calls “patchy” or “partial” or “creeping” reductions. Genes can help to illuminate one “patch” of the huge field or network that would in theory constitute something like a complete explanation of a behavior.

Finding a Path Forward to Understanding Human Behavior

Schaffner nimbly moves from worms to human beings. What geneticists have not been able to discover regarding human personalities should reassure, even gladden, skeptics.

At the turn of the century, some psychologists and geneticists hypothesized that there were three domains of personality temperament — novelty seeking, harm avoidance, and reward dependence; each linked to a distinct neurotransmitter — dopamine, serotonin, and epinephrine; and thus linked to “genes for” the production and regulation of one of those neurotransmitters. The idea was that specific gene variants associated with the regulation of dopamine, for example, had significant effects on novelty seeking. Again, those initial results failed to replicate. Among the reasons for those failures was the mistaken assumption that single “candidate” genes would, independent of their interaction with other genes and environmental variables, have large effects on traits as complex as personality. Combine that mistaken assumption with the all-too-human appetite of scientists, university PR departments, and journal editors for big, exciting findings, and voila: a variety of subtle statistical errors crept in.

Even the study of the interaction of genetic and environmental variables in the early 2000s was plagued with replication problems, perhaps due to their depending on the idea of “candidate” genes with large effects. Since then, extraordinary advances in technologies designed to compare genome sequences, combined with powerful new statistical methods, make it increasingly possible to detect genetic variants associated with tiny effects. The new, emerging picture boils down to this: common complex traits are the result of hundreds or thousands of gene variants of small effect size, which often interact with other gene variants as well as a gigantic range of environmental variables. It remains to be seen how much of practical value will result from this. Moreover, as Schaffner observes, it may be that huge categories like “novelty seeking” and “harm avoidance” are just too vague or indistinct to establish pathways from genes to behaviors like these. Again, to know that personality “is genetic” is massively different from knowing which genes are at work, much less how they are contributing to a given trait.

While Schaffner’s account of personality genetics may dishearten aficionados of genetic explanations, his account of schizophrenia should gladden them. Schizophrenia, too, is a large and heterogeneous category, but researchers have made headway in characterizing that heterogeneity — in specifying the symptoms and subtypes of schizophrenia. It’s in the context of schizophrenia that Schaffner elaborates on his conception of successfully reductionist scientific explanations. Such explanations, whether of schizophrenia or any other disorder or behavior, will have to be “interlevel”; in other words, they will need to draw on what is known at the level of ions, molecules, cells, cell-cell circuits, and organs — and will have to tell a story about how, over time, the factors at those different levels interact with each other and their environments.

In the case of schizophrenia, this includes genes implicated in the production and regulation of specialized nerve cells, specialized parts of those nerve cells, connections among those nerve cells, and, ultimately, brain wave patterns thought to be associated with the activation of those neuronal circuits and associated with at least some features of schizophrenia. Need one say that the model he describes is not anywhere close to complete? (Nor is the elaboration of this model, which has recently received high-profile attention.) Rather, it offers a “creeping” reduction — incremental progress in using the tools of genetics and neuroscience to understand one patch of the massively complex phenomenon we call schizophrenia.

Clearly, this model shouldn’t inspire euphoric expectations of imminent cures. Again, to his credit, Schaffner is adamant in stating that, “DNA sequence per se increasingly seems impoverished as a biological explainer” (p. 197). And, again, this is not to say that DNA sequence is unimportant — it’s just not important in the simple ways we once imagined, which notably still linger in the imagination.

A Grownup Conception of Freedom

So, is human behavior genetically determined? Different from what a sweeping genetic reductionist would hope, we have seen that the answer is plainly no. But nor is human behavior not determined. On the contrary, Schaffner thinks that human behavior is determined — and that it admits of reductionist explanations. Does this mean freedom is an illusion? No, it doesn’t, even if it does mean that we have to give up conceptions of freedom of the sort that best-selling authors like Sam Harris like to set up in order to knock down. Yes, we have to give up the idea of freedom as an extra-natural capacity or force that is somehow insulated from the impact of the natural and social forces at work in the world. But accepting that our behaviors are determined by natural and social forces that, at least in principle, admit of explanation does not mean that we have to give up the conception of freedom that mature adults should want, or that, as Daniel Dennett puts it, “is worth having.”

To get at what such a conception of freedom is, Schaffner introduces philosopher Harry Frankfurt’s influential distinction between first- and second-order desires. Consider, for example, an alcoholic with insight into her alcoholism. She might have a second-order desire not to drink, while also having a first-order desire to drink. The person who cannot bring her first-and second-order desires into alignment lacks what warrants being called free will. If, on the other hand, she can get those first- and second-order desires into alignment, and if she can, as it were, desire what she wants to desire, we can say that she is free.

The behavioral geneticist and philosopher of psychiatry, Kenneth Kendler explains how human beings can, “through their decision-making capacity, intervene in causal pathways from genes to behavior.” Kendler’s first example is alcohol dependence. We know from classical behavioral genetics studies that alcoholism “is genetic” in the real but limited sense that the genes that children inherit from parents can put them at increased risk of becoming alcoholics. We also know, however, that children of alcoholics are also at increased “risk” of becoming teetotalers — practicing complete abstinence from alcohol; Donald Trump’s response to his father’s and brother’s alcoholism is a case in point. Kendler and Schaffner both want us to notice how a grownup conception of freedom retains a place both for genes and for choice. In other words, human decisions can be an essential factor in the multilevel causal network that gives rise to our behaviors. If we notice that genes, neurons, hormones, neighborhoods, cultures, histories — and human desires and choices — can be among the determinants of human behavior, determinism should be less anxiety-producing.

In offering his view of the sort of freedom of choice that any grownup should want, he reminds us that scientific researchers choose which level of the causal network they will study. There is nothing wrong with having a preference for a given level of analysis, but there is something wrong with forgetting that a preferred level won’t be the only one needed to make headway in the sorts of reductions that can contribute to practically useful explanations.

An Anxiolytic and a Mood Stabilizer

This brings us full circle to the growing anxiety swirling around the idea of “designer babies,” and more specifically to the idea that it will be possible to use “gene editing technologies” like CRISPR-Cas9 to engineer traits like intelligence. As we begin to appreciate that such traits involve hundreds or thousands of genes interacting with each other and with the cellular, extracellular, and extraorganismic environments, then the less seriously we can take the notion that it will be possible to enhance such traits by making changes at the level of the gene.

Moreover, as mentioned earlier, understanding this complexity can help stabilize the mania precipitated by the Human Genome Project. Ever since its launch in 1990, we have heard ecstatic claims about the imminent arrival of medical diagnoses, treatments, and preventive interventions tailored to individual genomes. While it is absolutely crucial to appreciate the real and important strides in diagnosis and treatment linked to advances in understanding the genome, it is equally important to appreciate that, with few exceptions, knowledge at the level of the genome alone will likely not be able to produce as much clinically relevant information as was once promised.

As we taxpayers begin to pour hundreds of millions of dollars into the Human Genome Project’s offspring, The Precision Medicine Initiative, we should hold its leaders to their word when they say that they are getting the mania under control. Given how ardently some of the leaders of that initiative — not least Francis Collins — have been committed to a geneocentric approach, and given how mesmerizing and cheap gene-sequencing has become, it may take significant effort on their part to live up to their new promise of pursuing a more multilevel and, dare one say, balanced approach. Reading Schaffner’s book could strengthen their resolve to live up to that promise.


Erik Parens is a senior research scholar at The Hastings Center, a bioethics research institute in Garrison, New York, and is the author of Shaping Our Selves: On Technology, Flourishing, and a Habit of Thinking.


  1. Jan says

    One should make a distinction between prediction and explanation. If we can predict that an individual with genetic variants a1, b1 and c1 is, on average, going to have a higher phenotypic value (say, higher height or IQ) than an individual with variants a2, b2 and c2, then that information can be used for the purposes of genetic engineering. There is no need for us to know how those genetic variants cause differences between individuals — no need for information about the no doubt complex interactive process between genes, gene products and environments that ultimate gives rise to a particular phenotype. If we can use information about causal allele differences for embryo selection or for tweaking genes in embryos, then genetic engineering is a reality, regardless of the reasonableness of a “network perspective” or similar ideas.

    If behavior genetics has established something, it has established that phenotypic similarity in humans is linearly related to their genetic relatedness. This means that an additive genetic model has great explanatory power and that identifying those additive genetic variants enables genetic engineering. While there are ethical concerns, it’s clear that such engineering is going to happen in any case.

  2. I don’t want to make the perfect the enemy of the good. First, this article makes some very good points. And many laypeople might find it interesting and useful, but I could help come away with the feeling like you’re saying a lot without saying anything (or at least without saying much). Unfortunately, this account is highly confused.

    Here’s an example:

    While Schaffner’s account of personality genetics may dishearten aficionados of genetic explanations, his account of schizophrenia should gladden them.

    I don’t see how. Both are just as heritable.

    What I take away from this is that yes, genes do cause behavior, but the causal pathways involved in how they do so are long and complicated. Great. But at the end of the day that doesn’t change the key fact: genes cause behavior.

    To get at what such a conception of freedom is, Schaffner introduces philosopher Harry Frankfurt’s influential distinction between first- and second-order desires. Consider, for example, an alcoholic with insight into her alcoholism. She might have a second-order desire not to drink, while also having a first-order desire to drink. The person who cannot bring her first-and second-order desires into alignment lacks what warrants being called free will. If, on the other hand, she can get those first- and second-order desires into alignment, and if she can, as it were, desire what she wants to desire, we can say that she is free.

    Here’s the problem, and trouble that many people who talk about free will run into: “second order” desires are just as heritable as the first. Meta-cognitive actions (thinking about thoughts/behaviors) are just as heritable as the “lower” behaviors themselves. Genes make Donald Trump a teetotalers just as it makes others in his family alcoholics.

    I have to be honest, the more I read this article, the less impressed by it I became. Let me give some essential truths:

    1. All human behavioral traits are heritable. Generally substantially so. This includes the meta-cognitive (“second order”) thoughts that are the supposed realm of free will

    2. While it true in theory that environment modulates heritability, there is little evidence of this in practice. Traits highly heritable in one environment are generally found to be highly heritable in all other environments. Furthermore, we’re in our environment, the developed world of the 21st century, and not in some putative other environment where heritability might be different. We have to react the heritability estimates we have.

    3. You certainly can modify human traits, including intelligence and personality, by changing genes. This includes both through genetic editing or through selective breeding. Indeed, the latter is the basis of agriculture, which we know works.

    4. The “environment” does impact the expression of genes, but it’s not at all the environment most people think. Most of the differences between individuals living at the same time that can’t be boiled down to differences in genes stem from pathogens or random variation in development in utero (developmental noise). It isn’t things we have any control over. Even the differences between generations stem from grand environmental differences, things that affect us all, like technology, and hence meaningless for individuals within a generation.

    5. Sooner or later, we are going to be able to alter the human genome through technological means. That will mean will have the ability to affect the prevalence of the type of people that exist. There is no getting around this issue, nor should this be a major reason for concern.

    • expeedee says

      Well said JayMan. You are good at wading through the obfuscation and presenting a succinct and much clearer description of genes and behavior.

    • Jerryskids says

      Genes make Donald Trump a teetotalers just as it makes others in his family alcoholics.

      Do you know what specific genes differ between Donald and his brother that makes this so? Or are you simply asserting the opposite of what the author is asserting? “Here’s a difference between Donald Trump and his brother, therefore it must be genetic” versus “Here’s a difference between two genetically similar people, therefore not all differences can be explained by genetics”? I ask because your post seems to suggest some fundamental disagreement with the idea that we’re not all reducible to our genetic structure but there’s really nothing concrete there to substantiate that.

  3. The history of progress in science has always been about extending models which are reductionist and determinist. These models – to the extent that they are successful – deflate ideas of conscious agency and free will. Explanatory models deflate norms. Norms only flourish in the absence of explanatory models. There’s no simple emotionally satisfying way to have it both ways. Even without material reductionism, the logic of cause and effect deflates free will. What could lead a conscious agent to initiate an action? Particular limited understanding combined with particular limited motivations. Where do these come from? Etc.

  4. @SlyGoat; Read and understand Karl Popper’s masterwork, The Open Society and Its Enemies premise to The Poverty of Historicism.

    In a word, we and YOU are victim of the failed dialectical syllogism foist by Hegel and Marx, based on a gross mistranslation of Plato’s Republic.

    Historicism, the logic-like syllogism that events necessarily cause particular effects, fails for free will.

    • I was thinking entirely of physics, not history. Particular historical events and trends, known by humans, are profoundly general and chaotic (sensitive to measurements and relationships we aren’t measuring very well). Physics doesn’t seem to operate deterministically either down at the quantum level, but it certainly works better than historical analysis.

  5. Pingback: Outside in - Involvements with reality » Blog Archive » Chaos Patch (#115)

  6. Pingback: Whewell’s Gazette: Year 2, Vol. #41 | Whewell's Ghost

  7. Pingback: 把基因摆回属于它的位置(上) | 微博日报

  8. Pingback: » [译文]把基因摆回属于它的位置

  9. Pingback: A lire : Giving genes their due but no more - Anthropotechnie

  10. Pingback: Why You Are Who You Are Today (Psychological Explanation)

Leave a Reply