Selfish Genes and Selfish Beasts
It is sometimes supposed that selfish gene theory implies that animals are always biologically selfish—that all animals carry “genes for selfishness”. That is a misunderstanding because the whole point of the theory is to allow for kin-related altruism: an animal will act against its own individual interests in order to further the interests of its offspring (or other relatives), since they carry its genes. The selfishness of the genes ensures the unselfishness of the beasts whose genes they are. Of course, some degree of selfishness is built into the theory because an animal must preserve itself if it is to propagate its genes—it needs to survive long enough to reproduce, hopefully several times. But the thrust of the theory is that animals always act to ensure the survival of their selfish genes, which can be either in them or in other animals. Selfish genes are not “genes for selfishness” but genes for altruism—granted that other animals share their genes.
To drive this point home, consider the following thought experiment. Suppose an animal were to transfer all of its genes into its offspring, leaving none behind. The DNA is physically transferred from parent to offspring—not merely copied but itself sent elsewhere. After reproduction the parent is left with no DNA and so cannot reproduce again. Nothing like this happens on earth, of course; we are imagining a mere logical possibility. What would selfish gene theory predict about the behavior of the parent in relation to the child? Well, there is no need for the parent to conserve any resources for further acts of reproduction, so it will not neglect its offspring in order to save up for future offspring. Will it behave selfishly in relation to itself? No, because there are no selfish genes in it to program such behavior—from their point of view the animal is now quite useless (if it behaves selfishly). What the genes will program is total altruism in relation to the animal’s offspring, because that is the only place where its genes reside now. This hypothetical species will lay down its life for its offspring, no questions asked, because the genes program behavior that serves their interests: genes for anything else will not survive given the competitive realities of animal life. That is, selfishness in any form or measure will not exist in animals that lose their genes at reproduction to their offspring. So the theory predicts a complete lack of selfishness in beasts of the type described. The parent will always provide dinner for its offspring even if that means starvation for itself (unless its staying alive is in the interests of the offspring). That is the logic of the theory, not concern for the interests of the individual. Selfish genes make unselfish beasts.
Of course, genes are not really selfish—that is just a dispensable metaphor used to sum up the underlying structure of natural selection. Only beasts can be selfish—beings with desires, needs, interests. But they aren’t selfish according to the selfish gene theory; and indeed you would be hard put to find an instance of an actual animal that behaves selfishly in relation to its offspring. If it did, its genes would be less likely to survive into future generations: selfishness within families would be strongly selected against. So there is nothing, according to the theory, in the biological world that is really selfish: not the selfish genes and not the beasts they construct and program. Selfishness only arises in relation to genetically remote animals: here the genes will program selfish behavior because they have no interest in preserving other genes. But this is not because animals care only for themselves; on the contrary, they care for any animals with which they share their genes. In the imaginary case this can go so far as to make animals act without any regard at all for their own wellbeing. If we came across a planet populated by animals of the kind described, we would marvel at the daily feats of extreme altruism performed on that planet—no matter how “selfish” the genes are there. Given the underlying principles of biological evolution, it is only a contingent fact that animals ever behave selfishly, i.e. in their own best interests. Altruism is the basic rule. 
 I should emphasize that the altruism in question is confined to cases of genetic overlap; it does not extend to strangers. However, it would be possible to imagine whole populations that act altruistically in relation to each other if they were sufficiently genetically related. The logic of the theory is always that genes program behavior that favors their survival prospects—it doesn’t matter who contains those genes. The correlation between family membership and altruism is itself contingent: if strangers had more genetic overlap with an animal than its own family members, then the strangers would be the recipients of that animal’s altruism (remember we are here dealing only with imaginary cases).