Walker Response 1
In Defense of the Genetic Virtue Program: A Rejoinder
Mark Walker
The target paper has a vaulting ambition: to convince readers that we ought to attempt to reduce evil in our world by constructing an interdisciplinary program, which I call the Genetic Virtue Program (GVP), to enhance the biological aspects of virtue. Most of the contributors to this forum are not sympathetic to the project—to put it mildly. Yet, one of the surprising things, at least to this author, is that comparatively little is said about the paper’s overall ambition. Jamie Bronstein offers the idea that better socialization may be key: “No one would argue that there have not also been great evils; but the historical record doesn’t support the level of improbability for further moral improvement through socialization that Walker would like to assign.” However, nowhere do I claim that there is no prospect for further moral improvement through socialization. Indeed, I have recently published a piece that specifically recommends that we improve socialization efforts to enhance virtue by publically tracking prosocialization efforts.[1]
The part of Bronstein’s conjecture that is true is that social reforms, however helpful, are only going to eliminate evil so far. Think of improvements in children’s health since the nineteenth century. Cultural changes, combined with new health and hygiene practices, helped to substantially lower rates of serious disease and death. But it is folly to think that social measures alone will eliminate the need for biological interventions, for example, antibiotics and immunization programs. The social reforms Bronstein mentions—“successful civil rights movements for racial equality, and reforms in capital punishment and the treatment of prisoners”—are to be welcomed, of course, and we should hope that one day they will be fully realized. But these reforms have not immunized humanity against genocide in Cambodia, Rwanda, or Yugoslavia to name but a few examples in the panoply of recent human genocides. Nor does Bronstein explain how the single improved socialization practices cited—scholastic improvements in Harlem—can be applied to moral improvement.[2]
Of course, we can hope all we want—hope, after all, requires no evidence—but it goes against all reasonable probability to extrapolate from reformed educational practices in Harlem to think that such evil acts as murder, rape, or genocide will disappear any time soon, or that they may be applied to making the general population more virtuous. Two other respondents take another tact. Larry Arnhart speculates that evil may be an ineliminable feature of the human condition, while Robert Sprinkle says, in reference to the hope expressed in the paper for the elimination of evil, “I, for one, never held such a hope.” Yet talk about evil in the abstract should not blind us to the fact we are talking about real victims. I don’t claim to know that we can eliminate evil, but likewise, it unduly privileges our epistemic position to say that improvements in biology (coupled with improvements in socialization) will never eliminate evil.
Overall, I am indebted to the contributors to this forum for raising many important issues and specific challenges to the arguments of my paper. My further replies are grouped below under three headings: science, ethics, and eugenics.
Science
Several respondents found my characterization of the scientific issues associated with the GVP to be unduly optimistic, if not downright naive. To assess these charges it will be helpful to say something first about the general scientific landscape in which the GVP is located.
The science of the GVP is a proper subset of behavioral genetics. In the most general terms, behavioral genetics seeks a scientific account of the role of genetics in the behavior of organisms, including the human organism. The field’s scientific ambitions are varied but include the goal of explaining, predicting, and influencing behavior in terms of genetics.[3] To illustrate this point, consider that behavioral genetics has had an abiding interest in psychopathologies such as schizophrenia. Schizophrenia affects a little less than 1% of the population, but, through familial studies, behavioral geneticists have found that one’s chances of developing schizophrenia increase tenfold (from 1% to 10%) when a sibling has the disease.[4] A longstanding hypothesis is that dopamine regulation is implicated in schizophrenia, and behavioral geneticists have recently found differences between schizophrenics and the general population regarding the existence of genes associated with dopamine regulation.[5] The goal of influencing behavior in this case is ultimately therapeutic—to reduce the incidence of schizophrenia. In terms of actual practice this may mean recommending to those who are genetically predisposed to the disease to avoid its environmental correlates, such as substance abuse. It also means creating pharmacological agents based on knowledge of underlying genetic differences; and, more prospectively, perhaps using genetic technologies to alter or select against genetic correlates of the disease.
Consider now the basic structure of the science section of the paper. The argument assumes that behavioral genetics is basically on track in its ambitions to explain, predict, and influence behavior, and then cites evidence that a behavioral genetics of virtue is a serious possibility. So—and this cannot be emphasized enough—my paper has no independent argument for the general claim that behavioral genetics is, at least for the most part, on track in its aims and methodology; rather, the paper accepts that the issue has been resolved in favor of behavioral genetics. Naturally, anyone who questions behavioral genetics in general will perforce have reason to question my proposal for a behavioral genetics of virtue. The reasons for not defending behavioral genetics in the paper are the usual ones of space and competency. That said, it should be noted that although the discipline is only about a hundred years old, behavioral genetics is well-entrenched in the academy. This is not to say that it does not have its detractors,[6] but what discipline doesn’t have detractors?
The scientific objections raised by my critics apply to behavioral genetics in general, not specifically to the GVP. For example, Athena Andreadis finds the GVP to have a “terminally wobbly foundation” in part because it is based on animal and twin studies. Guilty as charged. The argument that the science of the GVP is a subset of behavioral genetics is indeed based on such evidence. However, animal and twin studies form an integral part of behavioral genetics as currently practiced. Their limitations, including a specific criticism that Andreadis mentions—adopted twins are likely to grow up in similar environments—are not new, and indeed, often cogitated in the literature.[7] If the science of the GVP is built on “terminally wobbly foundations,” then so too is much of behavioral genetics.
Another attack on behavioral genetics is evident in Sprinkle’s ruminations. He almost certainly is correct that any hope of a Mendelian type of reduction of virtues is beyond the pale of serious speculation, and once we realize that, the “polygenetic, intensity of expression, and gene-environment interaction problems pile on.” He suggests that the problem pile is so large that the science of the GVP requires “A philosopher’s ‘jump to hypo space’ — which is to say, to hypothetical space.” But notice that he has not told us why researchers investigating the genetics of schizophrenia are not obligated to make a similar “jump to hypo space.” After all, genes for schizophrenia are polygenetic, and show intensity of expression and gene-environment interactions. For instance, gene-environment interactions are dramatically evident in the slightly less than 50% concordance rate of schizophrenia in identical twins.[8] No doubt researchers investigating the genetics of schizophrenia would be surprised to hear that the problems they face require a jump to hypo space. In any event, the problems Sprinkle mentions plague behavioral genetics in general, not the GVP in particular. So, the fundamental dilemma for this line of criticism is that either behavioral geneticists researching schizophrenia have left the terra firma of Mendelian genetics for hypo space, or their research methodologies show the way forward to investigate the possibility of a behavioral genetics of virtue. One can’t have it both ways.
To assess some of the other scientific criticisms it will be helpful to focus on a specific technology for possible implementation of the GVP, namely, pre-implantation genetic diagnosis (PGD). Imagine this situation: a couple produces 10 embryos and each has its full genome sequenced. (The price on full genome sequencing is falling rapidly; soon cost will not be an issue, at least for many in the wealthier nations). This assumption is more than fair since PGD is already practiced today—embryos are sometimes selected on the basis of genetic correlates for disease, sex, and sibling tissue matching. Also, the technology is much less spectacular than genetic engineering: anything that can be done with PGD can be done by genetic engineering (at least in principle), but not vice versa. For instance, selecting against Tay Sachs disease is possible today in labs where PGD is performed. Although the technology is beyond our present grasp, in principle the same result could be achieved by genetically engineering an embryo with genetic markers for Tay Sachs. On the other hand, genetic experiments such as inserting a foreign gene—as in the experiment that inserted a jellyfish gene into a rhesus monkey nearly a decade ago, are virtually impossible to perform through PGD. To replicate this experiment with PGD one would have to hope that the jellyfish gene spontaneous mutates in a monkey zygote and then select this zygote. The odds of this happening are astronomically small.
With the technology of PGD in mind, consider the following remark by Andreadis: “The most fundamental problem with the GVP is that there are no genes for virtue or even inclination for it, just as there are no genes for intelligence.” One interpretation of this remark is that the difference between human and chimp intelligence has nothing to do with genes. If there are no genes for intelligence, then genetics is not going to explain the difference. A different, and perhaps more charitable understanding of the remark, might be interpreted as implying that genes are only part of the story about why humans are smarter than chimps. As Andreadis notes, “Genes encode regulatory, catalytic, and structural proteins and RNAs,” so some very complex story will have to be told about the fact that genes, which encode for proteins and RNAs, contribute to human intelligence. Part of this story will include environment-gene interactions. But, ultimately—and this is the important point—at least part of the explanation is in terms of differences in DNA. But this can’t be right as an interpretation because the GVP assumes that differences in DNA are only part of the explanation.
Perhaps the remark about “no genes for intelligence” is intended in the sense that complex traits, like intelligence and virtue, likely have a large number of contributory genes, and that altering any single allele is likely to have a small and unpredictable effect. Yet, is not clear that anything in particular follows from this. Take the example that Andreadis mentions: “The 2007 work of the Plomin lab indicates that the six single-nucleotide polymorphisms defined genomic regions that contribute the most to IQ – itself a population sorting tool rather than an intelligence indicator – influence IQ by a paltry 1%.” Two points are relevant here: first, suppose you are deciding which of two embryos to implant, and behavioral geneticists told you that one embryo has all six SNP correlated with IQ while the other has none. If your concern is to have a child with the highest probability of a higher IQ than its sibling, and this is the entirety of the information at your disposal, then you should choose to implant the embryo with polymorphisms identified by Plomin’s lab. There is no certainty here, for we are dealing with probabilities, but this is the only rational bet.
Second, numbers are not necessarily a showstopper here. Suppose that each allele we find correlated with higher IQ sorts the population by only a small fraction of an IQ point, and after 30 more years of investigation we find a total of 1,000 alleles, each with a paltry effect size. As a prospective parent you are asked by a genetic counselor to choose between an embryo that has 821 of the 1,000 alleles discovered or an embryo that has only 434 of the 1,000. Knowing nothing else, clearly the former is the correct choice for improving the chances of higher IQ. Of course, this is to assume that their individual contribution is causally independent. A more refined analysis might look for synergistic or antagonistic relations amongst alleles. True, we may not know what other effects these 821 polymorphisms have, but, in the absence of other information, there is no reason to assume that any effects will be deleterious as opposed to advantageous (or neither). Moreover, investigators will surely examine the polymorphisms for other correlations to well being such as disease and longevity, to see if there are any deleterious consequences associated with these polymorphisms. Clearly, this is a huge amount of data to crunch, but researchers are developing increasingly sophisticated statistical techniques to tease out small linkages, and computing capacity is improving all the time.
We should not forget that there are aspects of behavioral genetics that are relatively new, and so their power and scope are just now coming into focus. We have known for decades, for instance, that there is a heritable component to intelligence, but the search for genetic correlates is something that has become possible only recently. And the biological sciences are progressing at such an extraordinary rate that any research paper discussing traits like virtue or intelligence is likely to be subject to the “gotcha” problem where evidence is not as current as it might be. To wit: Arnhart and Andreadis suggest some of the scientific evidence in the target paper is stale. Of course, one should use the best and most up-to-date evidence available, so corrections along these lines are welcomed. A similar point applies to the Plomin (2007) research cited by Andreadis, which is now dated by research lead by Avshalom Caspi that found a 4-point IQ difference in a candidate “gene for” IQ:
It is reasonable to ask whether FADS2 is a “gene for” IQ. There was no overall main effect of genotype on IQ. However, breastfed rs174575 C-carriers scored 4.1 IQ points higher than GG homozygotes (104.0 vs. 99.9, P = 0.02). This finding suggests that under human ancestral conditions, when all infants were breastfed, genetic variation in FADS2 could have influenced individual differences in intelligence.[9]
Notice that if there were three more genes with independent population sorting potential like FADs2, we could potentially sort embryos by a whole standard deviation in IQ! Notice, too, that behavioral genetics does not ignore environmental influences, for this study demonstrates gene-environment interaction par excellence.
Arnhart raises the question of whether there is sufficient evidence for the claim that “there are genetic correlates of virtue that are clear, strong, and open to biotechnological manipulation.” To take this as the standard would be like President Kennedy saying to Wernher von Braun in 1960: “I am not going to authorize a massive expenditure to research and develop rockets to send a dozen astronauts to the moon unless you can provide clear and strong evidence of at least one person having already visited the moon.” A more reasonable standard is, of course, the one that was applied—namely, whether there was some reason to suppose that with considerable effort the probability of successfully sending humans to the moon would increase dramatically. Here, von Braun’s case would merely show plausibility in light of what was known in 1960. Likewise, the GVP proposal is for a research program designed to discover whether there are genetic correlates of virtue that are clear, strong, and open to manipulation. Thus, the case for the GVP is only circumstantial. Very little work on the heritability of virtues has been done. It should be emphasized that it is not as if scientists have looked and failed to find indications of heritability. So, the old saw applies: absence of evidence does not imply evidence of absence.
Ethics
Several forum contributors suggested that there could be terrible unforeseen effects with genetic engineering. Thus, Bronstein mentions that Belyaev’s experiment with selectively bred foxes demonstrates that there are possible unforeseen consequences of genetic manipulation. However, no reason is given as to why this is a problem indigenous to the case of moral enhancement. If one is worried about unintended consequences of genetic manipulation, then it is equally applicable to enhancing our intelligence, longevity, or our immune system. But if this is the case, then it seems the commentators overlooked the assumption, explicitly stated more than once in my original paper, that the relevant hurdles had been passed with non-moral traits. The paper focuses on reasons that are indigenous to the case of moral enhancement.