The Selfish Phenotype
Organisms often act selfishly, i.e. so as to benefit themselves. They compete with other organisms for food and mates; their behavior is anything but altruistic. Human organisms are a case in point. But they don’t always act selfishly—sometimes they act so as to benefit others at their own expense, as in raising their young. They exhibit kin altruism. That is to say, the individual organism is not always beneficent in relation to itself: sometimes it acts to benefit organisms not numerically identical with it. It has been pointed out that the same is not true of genes: they always act to preserve and benefit themselves. No gene ever predisposes an organism to act in such a way that that gene goes out of existence—or else it would go out of existence! There are no genes that benefit other alien genes, no altruistic genes. There are genes for altruism in organisms (kin-directed altruism) but not genes that sacrifice themselves for other genes. When an animal nurtures its offspring at its own expense it is benefitting its own genes, but genes don’t do anything comparable in relation to other genes. Genes are invariably selfish, while organisms are only intermittently so. Genes are selfish according to biological law, while organisms are selfish only in so far as that benefits their genes.
That is the orthodox picture and its logic is inescapable, but closer analysis reveals that it misses something important. This is that the organism’s actions serve also to reproduce its own phenotype—an organism similar to itself. If the parent organism sacrifices itself for its offspring, its survival comes into question; but its phenotype marches on. Not, to be sure, its entire phenotype, since its genes mix with those of another organism with a different phenotype, but more so than if it had no offspring at all. It uses the other organism in order to ensure that a partial copy of itself is passed into the next generation. It shares space with another organism’s phenotype in order to carry on its journey through time. So we can say that the phenotype survives in cases of individual self-sacrifice—just as the genes survive in such a case (the latter is a means to the former). Thus the phenotype is a selfish entity too. The phenotype never acts so as to benefit an alien phenotype—as it might be, an antelope’s phenotype acting so as to benefit an elephant’s phenotype—but rather it always acts so as to propagate instances of itself. The individual organism promotes the survival of its own phenotype by benefitting its own offspring. It doesn’t benefit itself qua individual organism, but it does benefit the suite of traits we call its phenotype (including its extended phenotype). To use philosophical terminology, it doesn’t benefit this token of its phenotype (viz. itself) but it does benefit the type of its phenotype (hence the use of “type” in “phenotype”). It helps produce more tokens of this type—otherwise known as babies. The type is more abstract than the token, less of a concrete particular; it is more of a universal or general kind. If evolution is a path through the space of all possible phenotypes, then the individual organism acts so as to maximize the chances of a particular such path. Accordingly, phenotypes can be viewed as self-perpetuating entities, leading to copies of themselves (or partial copies), and hence “selfish” in the technical sense. They are like genes in this respect and unlike the individual organisms that contain them—those dispensable temporary tokens. They are immortal in the way genes are immortal and individual organisms are not.
It might be thought that there is a significant asymmetry between genes and phenotypes, namely that genes promote themselves not just entities like them. It isn’t that the organism is a survival machine for similar genes in future generations but for these particular genes. Organisms produce tokens of their phenotypic type in future generations, but genes qua tokens sail through to the next generation and beyond—they get to survive not just genes similar to them. But brief reflection shows that this is false: the physical particulars that constitute genes in one body do not themselves survive in a future body. They disintegrate along with the organs and cells of that body once death supervenes. They become food for bacteria and victims of entropy. What survive are numerically distinct physical entities that replicate the originals. It is just as if I were to make a copy of a coin and then melt down the original: the type of the original survives but not the token. Likewise, these particular strands of DNA go the way of all things—into dust and disorder—but their type persists into the future in the form of numerically distinct material particulars (ultimately atoms). The old type is embodied in a new token. So when we say that an organism’s genes survive in its offspring we speak loosely: we don’t mean those specific bits of matter but their type or form. A copy of x is not x—it is something just like x. It is qualitatively identical (or close to it) but not numerically identical. So it is not that my genes are immortal in the sense that those very physical particulars will go on indefinitely (as elementary particles do); rather, their specific type travels on down the generations—copies of my originals not the originals themselves. Genes replicate themselves; they don’t just survive as is. Genetic perpetuation is genetic reproduction. But then, they are not different from phenotypes: they too are types not tokens, kinds not instances of kinds. The individual organism acts altruistically so as to benefit genes just like its own (but not literally its very own, i.e. the ones sitting inside its body); and similarly it acts so as to perpetuate a body like its own by producing and preserving a (partial) copy of its phenotype. Instances of a phenotype have no immortality, and nor do instances of a particular type of gene—they will perish with the organism’s particular body. Not that anyone I know of has ever claimed otherwise, but it is good to be clear about the logic and ontology of the situation: tokens come and go, whether of genes or phenotypes, but types go on in perpetuity (or until extinction). So genes are not selfish in the sense that they are solely concerned with their own survival qua physical particulars—that being impossible under normal conditions—but rather in the sense that they are hell-bent on producing copies of themselves. But the same is true of phenotypes—body plans, brain designs, types of psychology—and so do not differ logically from genes. Both are selfish as to type but not as to token. Organisms areselfish as to token, at least a lot of the time, since they are very concerned to make sure things go their way and the devil take the hindmost. I want me to survive not someone just like me! I don’t much care if my twin lives to carry my genes and my phenotype; I am more concerned about this individual thing called “I”. I don’t even much care if my genes and phenotype survive as long as I do. That is true honest-to-God selfishness, not the watered down kind exhibited by genes and phenotypes. It is as if the gene is saying to itself, “I don’t care if I survive so long as there are copies of me in the future”, and similarly for the phenotype; but the self is concerned precisely with itself—that with which it is numerically identical not just anything qualitatively similar to it. Sure, I may be glad if copies of me survive, but I don’t confuse this with my own immortality. Nor do the genes make this confusion, being well aware of the distinction between numerical and qualitative identity. Compared to me, my genes are quite selfless beings—as is my phenotype. Wanting my particular body to persist into the future (say by cryogenesis) is not the same as wanting a body similar to mine to persist into the future. The latter is the kind of “selfishness” proper to genes and phenotypes not the former.
We can then say that phenotypes are selfish in the way genes are selfish: both (metaphorically!) want copies of themselves to go on down the generations. Neither is more selfish than the other (though both pale in comparison to me). That is, phenotypes that successfully produce copies of themselves are favored by natural selection, while phenotypes that fail to cut it in the reproductive stakes are apt to fall by the wayside. Ditto for genes: the genes that do well are the ones that produce survival machines (bodies and brains) capable of winning the reproduction wars. To do that the gene doesn’t need physically to hop into the next organism, like a coin going from pocket to pocket; it just needs to replicate itself in that organism—copies not persisting originals. To put it differently, the operative unit is the type not the token: it is what gets passed on or not. The type needs a token if it is to have real-world impact, but the token is a dispensable entity in the wider scheme of things—it can safely go the way of all flesh. The unit of natural selection is really an abstract type not a concrete token. It is closer to the meme than is sometimes realized: the meme produces copies of itself but its original can disappear without a trace. The brain that housed the first meme of a jingle, say, can be reduced to pulp and yet the meme itself survives in copies of the original token. The meme is what is common to its embodying tokens not any of these tokens individually. In principle it can be realized differently in states of the brain in different individuals, and it is certainly not identical to any one of them. The same could true of genes, though it appears not to be in the actual world: that is, tokens of the same gene type could in principle be realized in different chemical configurations, though the chemical composition would have to duplicate the functional properties of the gene. You could in principle replace DNA with some other molecule and have the same gene type so long as it functioned to build the same body (“the prosthetic gene”). Genes could be multiply realized, as philosophers say. So it is wrong to identifygenes or memes with certain physical configurations, though no doubt there are lawlike correlations. Both belong at a more abstract theoretical level. What is selfish in biology, then, is really fairly abstract—gene types not gene tokens (ditto for phenotypes). This is really implicit in the notion of a copy: nothing is a copy tout court but only in certain respects. The gene is not copied with respect to its physical context or its time of existence or its movement through space; it is copied with respect to its informational content and its internal molecular architecture. It is the same for copies of paintings: not the way the light was shining on the original or its historical period or what the original painter was thinking at the time it was created, but rather a subset of its properties concentrated in its shapes and colors. The logical form of “copy” is given by “x is a copy of y with respect to R”—that is, relative to certain selected features. The selected features are what matters to the copy not the totality of its properties. It is the same with genes and phenotypes: general features that are relevant to what these things do, not the particularity of the individual instance (a scar on the body, a little jiggle in the DNA). The ontology of biology is thus more abstract than might be supposed, less tied to physical particulars (in this it resembles the ontology of psychology).
Let’s rank the biological world on the score of selfishness. Individual organisms are often genuinely selfish but not invariably so (including humans); genes are invariably selfish, but their selfishness does not concern their individual survival as bits of matter, but rather the survival of copies of themselves; phenotypes are selfish in the manner of genes, type not token, copy not original; memes (also part of the biological world) are always selfish, being concerned to spread themselves through as many minds and brains as possible, rather like a selfish parasite; species and groups are not selfish at all, never acting for their own good or anyone else’s. As to selflessness, only individual organisms manage this feat, serving the interests of both genes and phenotypes (and I suppose memes). In nature, it’s mainly every man for himself.
 Let it be noted that organisms like humans can be literally selfish, culpably so, while genes are selfish only metaphorically, and not culpably. The metaphor can readily be cashed, but I won’t do that now (see Dawkins).
 Species make no effort to avoid extinction, though it is not wrong to say that they can perform collective actions, e.g. moving to another continent. Nor do they replicate themselves. Nor do they aid any other species at their own expense. They are neither selfish nor selfless. You couldn’t write a book called The Selfish Species (or The Selfless Species).