Tuesday, October 09, 2007

Ethnic Genetic Interests Revisited   posted by DavidB @ 10/09/2007 02:17:00 AM

[Added 10 October: The comments facility on this post has been temporarily suspended due to obscene or offensive comments. Anyone with serious comments can email them to DavidBGNXPtemp@hotmail.co.uk , and I will try to reply to and/or post the comments in due course.]

[Added 14 October: No-one has taken me up on this offer, but my attention has been drawn to a comment elsewhere. See the Note at the end of the post, below the fold.]

In 2005 I commented on Frank Salter's theory of Ethnic Genetic Interests, as set out in his book On Genetic Interests: Family, Ethny and Humanity in an Age of Mass Migration (Frankfurt, Peter Lang, 2003). My post on 'Ethnic Genetic Interests: Part 1' (EPG1) summarised the theory, and stated some basic objections to it. 'Ethnic Genetic Interests: Part 2' (EPG2) covered more technical issues. A post on 'Interracial Marriage: Salter's Fallacy' (IMSF) dealt with a specific topic.

Razib has informed me that Salter's book is available in a new edition (2006) from Transaction Publishers. The main text does not appear to have revised, but the word 'ethny' in the title has been changed to 'ethnicity'.

The new edition also has a short Introduction commenting on reactions to the original edition. These include my own posts. For the small number of people who may be interested in such matters, I have transcribed the relevant passage from the Introduction below the fold, and added a few counter-comments of my own.

In general, I think my posts can still speak for themselves, and I see no reason to make any fundamental change. I will however take the opportunity (below the fold) to correct an oversimplification. I should also say that although I disagree with Salter's conclusions, I recognise the ingenuity and thoroughness with which he has pursued his ideas.

I also take the opportunity now to emphasise something I should have done previously: that differences in gene frequencies between human populations are in general fluid and transitory. Since Salter's aim is to preserve existing differences, this is fatal to his doctrines. Let us consider how such differences arise.

First, they may be due to natural selection still in progress. If an allele is generally advantageous, it will sweep through the world population in a relatively short time. But the process does take some time (of the order of a few hundred generations), depending on the strength of the selective advantage, rates of migration and intermarriage, and so on. Some of the differences between populations are therefore likely to be due to the different stages that selection has reached in different places. Differences of this kind are essentially transitory, and will be eliminated by ongoing selection. The process is unstoppable, unless selective conditions change. All that is needed is a very small amount of gene transfer (introgression) between different populations, and selection will do the rest.

Second, differences in gene frequencies may be due to genetic drift or founder effects in populations isolated from each other. In large populations (e.g. continental races) genetic drift is ineffective unless the alleles concerned are selectively neutral, and very slow even when they are. It is generally assumed (by Salter among others) that many of the genetic differences between human geographical races are of this kind, having arisen when populations were much smaller and more isolated than they are now. If so, then even moderate amounts of migration and intermarriage will narrow and eventually eliminate the existing differences. No feasible amount of 'Salterism' can halt the process, though it may slow it down.

Finally, differences may be due to variation in selective conditions, such as climate, diet, and exposure to infectious diseases. Genetic differences in skin colour, metabolism, blood groups, and the immune system are probably of this kind. Such differences are often clinal and cut across conventional racial groupings, though in some cases (e.g. resistance to malaria) different genes may be responsible in different areas. It is a matter for legitimate concern that increased migration and intermarriage may weaken the adaptedness of populations to their distinctive selective regimes. However, compared to the huge changes that are taking place in selective regimes for other reasons, this seems a very minor concern. Moreover, since differences in selective conditions cut across racial or ethnic boundaries, the resulting differences in gene frequencies are only weakly correlated with Salter's coefficients of 'ethnic kinship'.

I conclude that even if Salter's doctrine of ethnic genetic interests were sound in principle (which it isn't), in practice it would be either unnecessary or futile.

A correction

In EPG1 I commented that:
Salter's doctrine is profoundly anti-eugenic. For Salter, it is in the interest of an individual to preserve and promote the gene frequencies of his own ethnic group, whether the existing gene frequency is good, bad or indifferent, as judged by qualitative criteria. So, for example, it is in the interest of American blacks to promote their own gene frequencies against those of American whites, even if in some respects it would be better for blacks themselves to change those gene frequencies. The doctrine of genetic interests is inherently backward-looking and conservative.

In EPG2 I said:
For Salter, it is by definition adaptive for an individual to maximise the number of copies of his own genetic material, regardless of its selective value. So, for example, it is adaptive for an individual with Huntingdon's [sic: should be 'Huntington's'] Disease to maximise the number of other Huntingdon's sufferers.

Salter's theory implies, for example, that African populations who have migrated to temperate countries should seek to preserve their 'distinctive' African genes - even those, like the genes responsible for sickle-cell anemia, which are actively harmful in their new environment.

In defence of Salter it has been pointed out (though not by Salter himself) that in a few passages he does explicitly allow for the promotion of 'eugenic' genetic change, for example (p.89 of both editions):

Phenotypes can lose inclusive fitness through individual alleles in their genomes becoming maladaptive, due to mutation or to changes in the environment. Those dysfunctional alleles no longer serve the interests of the majority of the genes comprising the genome and thus the individual's genetic interests would be preserved or increased by substitution of maladaptive alleles.[i.e. their replacement by more adaptive alleles].

There are a few other such concessions.

My posts therefore oversimplified Salter's doctrine, and I apologise for the oversight. In mitigation, I would plead that the relevant passages do not occur until nearly 100 pages into the book, they are not given much emphasis, and they are hedged around with qualifications. This is perhaps not surprising. They go against the general tendency of Salter's doctrine, and are difficult to reconcile with his emphasis on the overriding importance of genetic continuity as such. If we accept - as Salter does accept in these few passages - that the maladaptiveness of genes can override the 'ultimate' interest of genetic continuity, where does it end? Pursued to a logical conclusion, the implication of Salter's concessions is that the ultimate test of whether the frequency of a gene should be increased or decreased is its effect on fitness, not its current frequency within the 'ethny'. We would therefore be left with the absurd result that the only genes to which Salter's original doctrine of genetic interests applies are those that are selectively neutral!

I now transcribe the passage from Salter's new Introduction which responds to my original posts:

From the Introduction to the 2006 edition

I [Salter] now turn to two technical points. One commentator has identified a terminological error that needs to be corrected (David B. 2005c [EPG2], point 1). Throughout On Genetic Interests I define an individual's genetic interests as consisting in his or her distinctive genes, some of which are found in kin and fellow ethnics. Such genes exist, but kinship generally consists of shared frequencies of genes, even when the genes concerned are not unique to any one individual. 'Distinctive genes' should be read as 'distinctive gene frequencies' or 'distinctive allele frequencies'. The same commentator thinks that making such a change weakens the emotive force of genetic interests (evidence that even some critics are impressed by it). 'Why anyone should consider this a "vital", "fundamental", or "ultimate" interest is beyond me. Why so much fuss about shifting a gene frequency from, say, 50 percent to 60 percent?' But kinship found within ethnies is homologous with that found within families. Are we to believe that parents do not have a vital, fundamental, or ultimate interest in their children? The point is that Hamilton's theory of inclusive fitness still applies whether genetic interests consist of unique genes or gene frequencies. If any branch of genetics is likely to have emotive force it is surely that which explains the evolutionary and social significance of kinship, whether in family or tribe. The same author [David B.] makes many other criticisms, including some thoughtful philosophical comments, but none are as decisive as that concerning distinctive genes. An example is his criticism of the argument I make in Chapter 8 (section 8.1) that exogamy, marrying outside the ethny, can reduce fitness (David B., 2005a [IMSF]. This is an interesting argument, but one made previously by Grafen, which I discuss on pp.262-5.

Comments on Salter's Comments

"The same commentator thinks that making such a change weakens the emotive force of genetic interests (evidence that even some critics are impressed by it)."

In the cited post I didn't refer to 'the emotive force of genetic interests', so the 'evidence' is purely in Salter's imagination. My objection was to the absurdity of Salter's own emotive rhetoric, when we consider what the pursuit of his 'genetic interests' really involves, namely a minor tweaking of population frequencies for alleles which often have no adaptive significance in any case.

"kinship found within ethnies is homologous with that found within families."

I'm not sure what exactly Salter means by 'homologous', but in fact there are many important differences between the 'kinship' of randomly selected co-ethnics and the ordinary kinship of close relatives. I pointed out some of the differences here. I should however correct or clarify Note 2 to that post. I would now stress that 'relatedness' must always be measured relative to the gene frequencies of a specified population. (This is clearer in Sewall Wright's original correlational concept of relatedness than in the more recent and widely used concept of 'identity by descent'.) The average relatedness between randomly selected members of a population, relative to the gene frequencies of that population, is always zero (apart from sampling error). We therefore should not expect altruistic behaviour to evolve between random members of a population, so far as interactions within that population are concerned.

"Are we to believe that parents do not have a vital, fundamental, or ultimate interest in their children?"

It depends what is meant by 'interest'. Salter's own concept of 'genetic interest' seems to me to be mystical twaddle, so I do not believe that parents have an interest in their children in Salter's sense. Of course, it is undeniable that human individuals do literally care for their offspring. Humans have been selected for an unusual degree of parental care. If they do not have children of their own, they may by default care for their nieces or nephews, or even adopt unrelated children or pets. What I deny is that this specific evolved tendency for parental care derives from, or can be extended to, some general concern for 'genetic interests'. There is nothing to suggest that people usually have any concern for the genetic prospects of their more remote relatives, let alone of random 'coethnics'.

"Hamilton's theory of inclusive fitness still applies whether genetic interests consist of unique genes or gene frequencies."

Salter's frequent references to Hamilton seem to me to be little more than name-dropping. There is no real connection between Hamilton's theory and that of Salter. Hamilton's theory of inclusive fitness applies to the effects of particular genes (alleles). If an allele increases the inclusive fitness of its bearers, then we expect it to increase in frequency in the population until it is fixed. It does not (metaphorically) say to itself, 'Hey, we're up to 60 percent, we can stop now.' In Hamilton's theory an allele has no 'interest' of any kind in maintaining the existing frequency of that allele in the population (unless it is already fixed). Moreover, many of the alleles which differ in frequency between populations are selectively neutral, in which case they do not affect inclusive fitness, and are not covered by Hamilton's theory. Under Salter's theory, in contrast, individuals are exhorted to defend the existing allele frequencies of their own ethnic group, regardless of the selective value of those alleles (apart from the occasional concessions mentioned earlier).

"This is an interesting argument, but one made previously by Grafen."

No, it wasn't. I would be delighted to find myself in agreement with Alan Grafen, who is one of the most brilliant evolutionary theorists currently working, but on consulting the cited work of Grafen I do not find my own argument. The purpose of Grafen's paper [Grafen 1990] is to consider whether organisms have evolved mechanisms for recognizing kin. As a part of this project, he considers the claim (known as Genetic Similarity Theory) that individuals can detect the degree of general genetic similarity of other individuals to themselves. He further considers whether assortative mating is due to such detection of similarity, and whether mating with relatives would be advantageous from the point of view of inclusive fitness. On this last point, he discusses some pros and cons, and sagely concludes: 'Assortative mating in humans therefore seems unlikely to be explained by genetic similarity theory because it probably does not succeed in creating substantial relatedness between mates, it would be an unnecessarily costly way of mating with a relative, and it is dubious whether it is advantageous to mate with a relative anyway'. In reaching this conclusion he appears to accept that mating with relatives would, other things being equal, bring an increase in inclusive fitness, but thinks that this needs to be balanced against other considerations, such as inbreeding depression.

In contrast, my own position is that mating with relatives (in this case, co-ethnics) in itself brings no increase in an individual's inclusive fitness at all, and that it is a fallacy to suppose otherwise. The key point to grasp is that, on neutral assumptions, each inbred mating replaces two outbred matings (one for each partner). I claim no great originality for this point, because (as I mentioned in my post) it is essentially the same as Dawkins's '10th Misunderstanding of Kin Selection', which I discussed here, with a correction here, which is the idea that 'Individuals should tend to inbreed, simply because this brings extra close relatives into the world'. Perhaps Grafen himself has fallen into this fallacy, but it would be more charitable to suppose that his discussion of the issue was simplified for the sake of brevity, as it was peripheral to his main concerns.

As Dawkins recognises, there could be special cases in which inbreeding does increase inclusive fitness. In his discussion of interracial marriage, Salter puts some emphasis on parental care, arguing that this has a greater fitness benefit if offspring are more closely related to their parents, as they would be through inbreeding or breeding within the 'ethny'. This is an intricate point, but my tentative view is that Salter's position is largely but not entirely misconceived.

In all species, there is a trade-off between the number of offspring produced and the amount of resources (including parental care) devoted to each offspring. The general evolutionary principle (usually attributed to David Lack, though he had several precursors) is that for each species there will be an optimum level of parental investment, at which the number of offspring who survive to maturity is maximised (subject to refinements over the timing of reproduction, as expressed in Fisher's 'Malthusian Parameter'). The optimum level is determined primarily by the ecological circumstances and habits of the species. As far as I can see, in general it does not make any difference to the optimum whether the breeding system favours inbreeding or outbreeding (except perhaps through side-effects such as the amount of competition between siblings). The optimum level, as determined by ecology, etc, maximises the reproduction of all the genes of the organism equally, including any genes that promote inbreeding or outbreeding. Such genes are therefore neutral with respect to the optimum level. It may however still be said that a gene for parental care would 'prefer' parental investment to be directed towards inbred offspring, who will on average contain more copies of that gene. I think this is true to the extent that if a parent has both inbred and outbred offspring, its genes for parental care would have an 'interest' in discriminating in favour of the inbred offspring, other things being equal. Of course, this interest would be offset by any reduction in the viability of inbred offspring due to inbreeding depression. In any case, I do not think that this limited preference for inbred offspring, where both inbred and outbred offspring already exist, should be confused with a general tendency to mate with relatives, and thereby to produce inbred offspring. This is an instance of the fallacy pointed out by Dawkins.

I should also stress that even if there is some validity in Salter's comments on parental care, this would not affect my main criticism of his arguments against interracial marriage, which is that interracial marriage does not in general reduce the net 'genetic interests' of the participants, as fallaciously assumed by Salter.

Ref: Grafen 1990: Alan Grafen, 'Do animals really recognize kin?', Animal Behaviour, 1990, 39, 42-54.

Note added on 14 October:

My attention has been drawn to a critique of the above post by J. W. Holliday. I won't give a link, as the critique is personally abusive, but those who are interested will be able to find it easily enough. I recommend it to connoisseurs of paranoid fantasy.

I won't respond in detail. As a famous philosopher once said, comments on comments on comments (etc) are subject to the law of diminishing fleas.

I will however pick up a few points. I emphasised in my post above that differences between populations due to genetic drift will be eroded and eliminated even by modest amounts of migration. Holliday appears to doubt this, but it is a well-established result of population genetics. The strength of genetic drift is inversely related to the size of the population. In a very large population, such as a continental race with a population of hundreds of millions, genetic drift in the population as a whole is extremely slow. Even low rates of migration - say, one in a thousand per generation - would be orders of magnitude more effective in changing gene frequencies. Indeed, subject to certain qualifications, even an exchange of one migrant per generation is sufficient to stop two populations drifting apart. Anything more than this, and the gene frequencies in the populations will tend to converge.

Differences in gene frequencies due to differing selective conditions are another matter. Migration will tend to reduce them, but if selection is strong enough some differences may be maintained. I have never disputed this.

Holliday assumes that I (or rather gnxp, which in his fantasies has a policy on such matters) advocate 'mass migration/miscegenation'. In fact, as I said in EGI1, 'there are sound arguments against large-scale, uncontrolled immigration '. I just don't think Salter's doctrines are 'sound arguments', and I believe it would be positively harmful to the case for immigration control if it became tainted with Salterism. As for 'miscegenation', I gave my considered views on the effects of interracial marriage here. In my 'Salter' post on the subject (IMSF) I was simply concerned to show that Salter's main argument against interracial marriage is a fallacy. Holliday concedes that it is a fallacy, but imagines that he can get Salter off the hook by pointing out that my assumption of 'neutrality' is unlikely to be strictly true in practice. Well, of course. It is an idealisation which simplifies the arithmetic for purposes of presentation. But Salter's own approach remains fallacious even if we relax the assumption.

Holliday rejects my conclusion that "the implication of Salter's concessions is that the ultimate test of whether the frequency of a gene should be increased or decreased is its effect on fitness, not its current frequency within the 'ethny'." But Holliday's counter-argument is vague and unquantified. In Salter's system, an individual's EGIs are measured by multiplying the number of his 'co-ethnics' (other than close kin) by the relevant degree of ethnic 'kinship'. Any gene substitution which increases the fitness of the co-ethnics will therefore increase the individual's EGIs unless by making the substitution the degree of 'kinship' is reduced by a larger factor. This is extremely unlikely to be the case, since ethnic kinship is averaged over a large number of different genes (probably thousands). The effect of any single gene substitution on the degree of kinship would therefore be very small, and would be outweighed by even a modest fitness advantage (of the order of one in a thousand).

It should also be noted that if we replace 'distinctive' genes, i.e. those with different gene frequencies between two populations, by genes that have equal frequencies in the two populations, then the negative kinship between the populations will also be reduced. I have pointed out previously that for consistency we should always count negative as well as positive EGIs, in which case the net effect of a gene substitution on EGIs is zero (assuming equal population sizes), and any fitness benefit of the substitution is pure gain.

But as EGIs are nonsense anyway, I think this is like arguing about whether chimaeras bombinate in the void.

Finally, Holliday stresses that Salter recognises the inevitability of genetic change and evolution even if his policy recommendations (essentially, preventing migration and intermarriage) were implemented. I agree that it is possible to find passages of this kind in his book. The fact remains that the overwhelming emphasis of his doctrine is on defending existing genes and gene frequencies as they are, and he does not (in my opinion) explain why some changes are to be cheerfully accepted while others are to be fought tooth and nail. If an author were to write a book making a passionate plea for freedom of speech as an 'ultimate' and 'overriding' value, but inserting sporadic unexplained concessions such as 'except for libel', 'except for obscenity', 'except for religious abuse', and so on, we would not form a high opinion of his or her consistency and intellectual rigour.