The final chapter of The Selfish Gene (Dawkins 1976) explored the analogy between genetic and cultural evolution. Cultural traits, Richard Dawkins suggested, evolve by a process of natural selection in which there is preferential proliferation of traits with properties that promote their own transmission. “We need a name for the new replicator,” he wrote, “a noun which conveys the idea of a unit of cultural transmission, or a unit of imitation. ‘Mimeme’ comes from a suitable Greek root, but I want a monosyllable that sounds a bit like ‘gene.’ I hope my classicist friends will forgive me if I abbreviate mimeme to meme.” Dawkins concluded his discussion with: “However speculative my development of the theory of memes may be, there is one serious point which I would like to emphasize once again. This is that when we look at the evolution of cultural traits and at their survival value, we must be clear whose survival we are talking about.” He entertained the possibility “that a cultural trait may have evolved in the way that it has, simply because it is advantageous to itself.”
The “meme” has exhibited admirable powers of replication and persistence in the thirty years since its conception, but its cultural spread pales before that of the monosyllable it was chosen to imitate. In the first half of this chapter, I will consider the diverse meanings that have become associated with that simple meme, the “gene.” Not every scientist means the same thing when they refer to a gene, and these differences in nuance can be a source of confusion. In particular, I will discuss Dawkins’s explicit definitions of the selfish gene and, in the guise of the strategic gene, propose what I believe to have been Dawkins’s implicit definition. We can think of the changing and diversifying concepts of the gene as an example of memetic evolution. The second half of this chapter will use the discussion of the “gene” in the first half to illuminate the status of the “meme” as a putative replicator subject to cultural selection.
The Danish plant breeder Wilhelm Johannsen (1909, 124) introduced das Gen (plural die Gene) into the German language as a shortening of Darwin’s “pangene” to replace the ambiguous, polysemic Anlage (unit). He then used the singular “gene” in an English-language address to the American Society of Naturalists in December 1910, published the following year. Johannsen’s intent in this address was to take issue with the common “conception that the personal qualities of any individual organism are the true heritable elements or traits!” The rediscovery of Mendelism had shown that “the personal qualities of any individual organism do not at all cause the qualities of its offspring; but the qualities of both ancestor and descendant are in quite the same manner determined by the nature of the ‘sexual substances’—i.e., the gametes—from which they have developed. Personal qualities are then the reactions of the gametes joining to form a zygote; but the nature of the gametes is not determined by the personal qualities of the parents or ancestors in question.” Thus, Johannsen (1911) made a crucial distinction between phenotype (observable traits) and genotype (heritable factors).
In Johannsen’s view, the mistaken notion of the inheritance of personal qualities was reinforced by the persistence of an outdated vocabulary. “It is a well-established fact that language is not only our servant, when we wish to express—or even conceal—our thoughts, but that it may also be our master, overpowering us by means of the notions attached to the current words. This fact is the reason why it is desirable to create a new terminology in all cases where new or revised conceptions are being developed. . . . Therefore I have proposed the terms ‘gene’ and ‘genotype’ and some further terms, as ‘phenotype’ and ‘biotype,’ to be used in the science of genetics. The ‘gene’ is nothing but a very applicable little word, easily combined with others, and hence it may be useful as an expression for the ‘unit-factors,’ ‘elements’ or ‘allelomorphs’ in the gametes, demonstrated by modern Mendelian researches” (1911, 132).
From this beginning, Johannsen’s “gene” has had an illustrious history, as have “genotype” and “phenotype” but not “biotype.” But “gene” itself conveys little information, consisting as it does of only four letters and a single syllable when spoken. The factors accounting for its success as a meme are probably those identified by Johannsen—that it was “a very applicable little word, easily combined with others”—and the historical contingency that the word was used to represent a set of ideas and concepts that had high memetic fitness. If “gene” is a meme, it is a rather uninteresting one. The interesting memes are the shifting concepts of the units of inheritance for which gene was a convenient label. The memetic history of “gene” is interesting only insofar as it provides a marker for the propagation of these more amorphous ideas and concepts as they have undergone constant reformulation.
“Gene” has never had a single meaning, but has always had different meanings for different people, and often different meanings for a single person, depending on context. For each person who added “gene” to their vocabulary, the word had a meaning that was derived from explicit definitions either read or heard, inferences from how the word was used, and reformulations of the concept within their own minds. This private definition of “gene” was then translated into new definitions and new uses in conversation and writing that were perceived by other minds and incorporated into new private definitions. My intention in stating the obvious is to point out what must be true of most memetic transmission: there is some degree of continuity in the propagation of ideas from mind to mind, but it lacks the high fidelity of the propagation of genes from generation to generation.
Johannsen, of course, had his own conception that he wished to convey to others. “As to the nature of the ‘genes’ it is as yet of no value to propose any hypothesis; but that the notion ‘gene’ covers a reality is evident from Mendelism. . . . We do not know a ‘genotype,’ but we are able to demonstrate ‘genotypical’ differences or accordances. . . . Genotypes can be examined only by the qualities and reactions of the organisms in question” (1911, 133). Genes were known by their phenotypic effects. Johannsen was dismissive of attempts to localize genes. “The question of chromosomes as the presumed ‘bearers of hereditary qualities’ seems to be an idle one. I am not able to see any reason for localizing ‘the factors of heredity’ (i.e., the genotypical constitution) in the nuclei. The organism is in its totality penetrated and stamped by its genotype-constitution. All living parts of the individual are potentially equivalent as to genotype-constitution” (154).
Johannsen’s “applicable little word” soon gained wide currency among geneticists, especially among those who believed, contrary to Johannsen, that the gene corresponded to a physical structure on chromosomes. One might say there had been memetic recombination that attached “gene,” as a label, to an alternative concept of the unit of inheritance. Supporters of the chromosomal theory however continued to define the gene operationally as that which was responsible for a heritable phenotypic difference. Alfred Sturtevant, one of the first to map genes to chromosomes, commented: “We can . . . in no sense identify a given gene with the red color of the eye, even though there is a single gene differentiating it from the colourless eye. So it is for all characters. . . . All that we mean when we speak of a gene for pink eyes is, a gene which differentiates a pink eyed fly from a normal one—not a gene which produces pink eyes per se, for the character pink eyes is dependent on the action of many other genes” (1915, 265).
Much of twentieth-century experimental genetics was engaged in making inferences about the physical nature of genes from observations of differences in the physical characteristics of organisms. These studies led to a definition of the gene as a stretch of DNA that was responsible for specifying the amino acid sequence of a protein. Thus, the operational definition of a gene—a gene is known by its effects on outward form—began to shift as genes came to be viewed as tangible elements with defined chemical properties. The existence of a gene is now often inferred from properties of a DNA sequence without any information about the gene’s phenotypic effects and without any observation of differences among DNA sequences. But this definition of the gene, as a protein-encoding stretch of DNA, is more recent than its definition as that which is responsible for a phenotypic difference, and it is not surprising that the modern molecular definition has not fully supplanted the older operational definition.
Experimental geneticists invoke genes to explain observed phenotypic differences. A pink-eyed fly differs from a red-eyed fly because the former possesses a gene for pink eyes inherited from both parents whereas the latter has inherited at least one gene for red eyes. In a similar way, evolutionary biologists often invoke genes to explain hypothetical phenotypic differences in an attempt to understand the nature of adaptation by natural selection. An ornithologist might wish to understand why males of some species help to raise offspring (dads) while males of other species put all their efforts into seeking additional copulations (cads). She might posit a gene for being a cad and ask under what circumstances it would invade a population in which most males behave as dads. To paraphrase Sturtevant, all that we mean when we speak of a gene for being a cad is a gene that differentiates a cad from a dad—not a gene that produces caddish behavior per se, for caddishness results from the action of many genes.
The use of genes to explain both differences among individuals and invariant features of organisms has led to an unfortunate confusion in public discourse over claims that genes cause behavior. Consider a contentious example. Behavioral geneticists are interested in differences among individuals and look for genetic factors that might explain why some people, but not others, engage in violent acts. Perhaps violent offenders are unable to control their impulses because they carry a mutation in the gene encoding the enzyme monoamine oxidase. The explicit comparison is between the behaviors of individuals with different variants of the gene. Evolutionary psychologists, on the other hand, are interested in species-typical behaviors that they view as adaptations to enhance the survival and reproduction of individuals. Therefore, they seek explanations for why we have evolved a genotype that makes us more likely to engage in violence under some circumstances, but not others. Perhaps young men are predisposed to violent behavior when they control few resources in societies in which there are large differences between rich and poor. The implicit comparison is between the reproductive fitness of individuals in the present world (or in a past environment in which the behaviors were adaptive) and the reproductive fitness of individuals in alternative worlds in which genes respond differently to the environment. Thus, behavioral geneticists ascribe the observed difference between offenders and nonoffenders to a genetic difference, whereas evolutionary psychologists ascribe the same difference to environmental factors. Yet both are castigated as “genetic determinists” by those who reject biological explanations of human behavior.
How then did Dawkins define the eponymous protagonist of The Selfish Gene? Dawkins (1976) recognized that there is “no universally agreed definition of a gene. Even if there were, there is nothing sacred about definitions. We can define a word how we like for our own purposes provided we do so clearly and unambiguously. The definition I want to use comes from George Williams. A gene is defined as any portion of chromosomal material that potentially lasts for enough generations to serve as a unit of natural selection” (1966, 30). Thus, the gene could be a longer or shorter unit than the protein-encoding gene recognized by molecular biologists. When defined in this way, Dawkins believed that the gene must be recognized as “the fundamental unit of natural selection, and therefore the fundamental unit of self-interest” (35).
Dawkins’s recognition of the gene as the unit of selection has not met with universal acceptance, partly because different scientists have different implicit definitions of the gene. David Sloan Wilson and Elliot Sober (1994), for example, have argued that the gene deserves no special status because it is merely the lowest level of a nested hierarchy of units of selection: genes, cells, individuals, groups, species. In this view, genes are nested within cells, cells within individuals, and individuals within groups; and natural selection can act at all levels of the hierarchy. Thus, there may be adaptations for the good of individuals and groups as well as of genes. The placement of the gene at a level of the hierarchy below the cell implicitly defines the gene as a material object located within cells. But this is not Dawkins’s concept: “What is the selfish gene? It is not just one single physical bit of DNA . . . it is all replicas of a particular bit of DNA distributed throughout the world. . . . The key point . . . is that a gene might be able to assist replicas of itself which are sitting in other bodies. If so, this would appear as individual altruism but it would be brought about by gene selfishness” (1976, 95). For Wilson and Sober, the gene is a material object that resides within cells, whereas for Dawkins the gene is a piece of information distributed across multiple levels of Wilson and Sober’s hierarchy. I cannot resist suggesting the usit—pronounced “use it”—as an applicable little term to represent the disputed unit-of-selection whatever that may be or mean.
The continued debate between gene-selectionists and hierarchical-selectionists identifies an ambiguity in meaning even when a gene is defined as a rarely recombining stretch of DNA. A gene could refer to the group of atoms that is organized into a particular DNA sequence—each time the double helix replicates, the gene is replaced by two new genes—or it could refer to the abstract sequence that remains the same gene no matter how many times the sequence is replicated. In the previous chapter, I called these concepts the material gene and the informational gene. Dawkins referred to something like the informational gene when he described the selfish gene as “all replicas of a particular bit of DNA,” but I believe that he neither wanted nor intended this definition. If all humans came to share the same DNA sequence, the theory of the selfish gene would not predict universal benevolence. A selfish gene does not “care” about all replicas of its sequence, but only about some of its replicas in a smaller group of related individuals. The reason why is tied up with the dynamics of genetic replicators.
Mutations create new informational genes that are modifications of existing informational genes. Each mutation occurs as a change to a single material gene. Therefore, the mutation must, at first, be a rare variant in the gene pool and its material copies will interact with each other only in cells of the same body or bodies of close relatives. Such interacting groups of material genes correspond to the strategic genes (or units of adaptive innovation) of the previous chapter. For such a mutation to increase in frequency by natural selection, its phenotype must benefit the transmission of its copies from this small group of individuals relative to the transmission of alternative informational genes by other small groups of related individuals. If the new informational gene increases in frequency because of this phenotype, strategic genes of its type will encounter each other in interactions between distantly related individuals but will still promote the nepotistic phenotypes that ensured their informational gene’s success when rare. For this reason, a mother will favor her own child over a twelfth cousin even though the child and twelfth cousin are identical for most of their informational genes. Genetic competition is predicted in populations of genetically identical individuals.
A material gene has dual roles. It can be expressed—that is, its sequence can be transcribed into a messenger RNA that is translated into a protein—and it can be replicated as copies of itself. The essence of adaptation by natural selection is that a material gene’s phenotypic effects, how it is expressed, influence the probability that the material gene, or its replicas, will be copied. The extent of the strategic gene is determined by how many replication cycles separate the material genes responsible for a phenotypic effect from the material genes that benefit from an increased probability of being copied. Thus, the strategic gene is not a fixed entity but one that can evolve to encompass more, or fewer, material copies of an informational gene.
Take for instance a large, well-mixed population of single-celled phytoplankton. Once a cell divides, the daughter cells separate and never interact again, except by chance. Each material gene is subject to selection solely on how its own expression influences its own replication. In this case, the strategic gene is limited to a single material gene. Now consider a cod fish in which multiple individuals of both sexes spawn simultaneously. A sperm and egg fuse to form a zygote that develops into a large multicellular individual that may itself contribute eggs or sperm to zygotes of the next generation. Zygotes become widely dispersed by ocean currents so that there are no preferential associations among kin. A single material gene in a zygote gives rise to replicas in all the cells of an adult cod. Material genes in a cod’s heart and brain never replicate, yet their expression promotes the replication of their replicas in the cod’s gonads. In this example, the strategic gene is spread throughout the body of a single fish. Finally, consider a beehive. Material genes that are expressed in the hive’s sterile workers promote the replication of their copies in the ovary or sperm storage organs of the queen bee. So in our third example, the material copies of the strategic gene are spread among the members of the hive. When Dawkins discusses selfish genes, it is in the sense of the strategic gene that his genes should be understood.
Dawkins proposed that memes play a role in cultural evolution that is analogous to the role played by genes in biological evolution. If so, memes should display features that promote their own replication. Such features could be interpreted as adaptations “for the good of” the meme itself. The remainder of this chapter will interrogate the analogy between genes and memes. I will employ a vague definition of a meme as “a mental item that is borrowed from one person and passed on to another.” There are many things that could be considered to be memes, but my focus will be on the transmission of ideas, and I will use the meme of the “gene” in counterpoint.
Rather than asking directly who benefits from memetic transmission, let’s consider who benefits from communication. Many acts of communication are committed because a sender wants to produce some change in a receiver. Such acts can be considered propaganda, from the Latin for propagation. An item of propaganda, a propagandum, is a device designed by a propagandist to achieve a change in the actions of a receiver. The propagandum has served the purpose of the propagandist if the receiver acts in the desired manner. To achieve this purpose it isn’t necessary that a receiver pass the propagandum on to others. If there is no chain of transmission, then the propagandum does not qualify as a meme. Its effects are not advantageous to the propagandum, but only to the propagandist.
Sometimes a propagandum is designed to be passed from one receiver to another because this increases the propagandum’s efficiency as an agent of mass persuasion. If the propagandist is successful in achieving ongoing transmission, the propagandum then qualifies as a meme. The propagandum serves its designer’s ends if receivers act in some desired manner and if receivers pass the propagandum on to others to affect their behavior. The features that promote the propagandum’s transmission benefit the propagandist but can also be said to benefit the propagandum considered as a meme. But the features that effect a change in the behavior of receivers, while they benefit the propagandist, need not benefit the meme.
A propagandist’s designs may misfire. A propagandum could fail to achieve the propagandist’s ultimate purpose of changing behavior, but could succeed in the subsidiary purpose of being propagated from mind to mind; or, the propagandum could continue to propagate from mind to mind after it no longer serves the propagandist’s goals. Once a chain of memetic transmission exists, there is no selection on a propagandum to serve its original designer’s ends, although the propagandum, as meme, may continue to serve these ends if the fidelity of transmission is sufficiently high.
Each step in a chain of transmission is an act of selection, insofar as the transmitter chooses to transmit one meme rather than another (or no meme at all). We can think of any feature of a meme that has predisposed successive transmitters to “want” to pass it on as an adaptation of the meme to enhance its own transmission: such adaptations can appeal to either the conscious motivations or the unconscious motivations and biases of transmitters; and such adaptations can be intended features, consciously selected by a propagandist, or they might be unintended features that arise from the interaction of “random” mutation with differential replication during the chain of transmission. Thus, the adaptive features of memes may be the products of “intelligent design,” “natural selection,” or a combination of the two.
Whose interests then are served when a meme is transmitted? We can look at this from the perspectives of individuals or memes. From the first perspective, we need to consider the interests of the transmitters at each step in the chain. If an individual consciously chooses to transmit a meme, then the meme must serve some perceived interest of theirs. I write perceived interest because individuals can be mistaken about what will promote their true interests. For example, a meme may be a propagandum serving the actual interests of some other individual earlier in the chain. (By the interests of an individual, I mean here their own self-defined goals in life.)
Does taking the second perspective, of viewing culture through the lens of a meme’s metaphorical interests in its own transmission, contribute anything that could not be obtained from the first perspective? A meme’s-eye view could be justified if it were shown that there are features of memes that promote a meme’s own interests without serving the interests of any of the meme’s transmitters. Such features might appeal to quirks of nervous systems that are better considered as unconscious biases rather than sources of personal motivation. A preference for a meme’s-eye view might also be defended if the features that make a meme likely to be passed on had accumulated in many steps over the course of memetic transmission.
Johanssen introduced “gene” to clarify the distinction between genotype (gene) and phenotype (trait). Can a similar distinction be made for a science of memetics? There are two principal kinds of things we observe that provide evidence about the nature of memetic transmission. The first are communication acts including sounds, texts, actions, and artifacts. The second are insights from introspection when we register a communication act, when we integrate the content of a communication act into our private set of concepts, and when we emit communication acts. Introspection may be an unreliable guide because unconscious aspects of our motivations are hidden and our conscious perceptions may be partial, inaccurate, and misleading. Communicative acts appear closer to the concept of genotype (things transmitted) whereas the conscious and unconscious effects of these acts on our internal state appear closer to phenotype (effects that influence what is transmitted). In the history of genetics, the phenotype was observed and the genotype inferred. But this relation is reversed for memetics. Memes are observed, but their effects are inferred.
The phenotype–genotype distinction works fairly well for genes, but there are many unresolved problems in its application to memes. For example, let’s suppose that there is a body of lore preserved and updated by the Medieval Guild of Propagandists about what techniques are effective in changing public opinion, and that this lore is passed from master to apprentice. An apprentice uses techniques of proven efficacy to design propaganda, and then a propagandum’s success in public persuasion influences whether the apprentice passes on the technique to his own apprentices when he becomes a master. From the perspective of the techniques as memes, propaganda are meme-products that influence a technique’s probability of transmission, but these items may also function as memes in their own right. A propagandum may be both “memotype” and “phemotype.”
The gene has a material definition in terms of a DNA sequence that maintains an uninterrupted physical integrity in its transmission from generation to generation. Memes also have a physical form in their transmission from one individual to another, sometimes as sound vibrations, as text on paper, or as electronic signals relayed through a modem. When these “outward” forms of a meme are perceived, they elicit changes in a nervous system that constitute the meme’s “cryptic” form. The material basis of the cryptic form is probably unique to each nervous system colonized by the meme. Memetic replication, then, has nothing like the elegant simplicity of the double helix.
If the material form of memes is problematic, might it be more appropriate to define memes purely in terms of information? But what are the memes in our evolving concepts of “the gene”? These concepts have been reformulated and recombined with other ideas at each step in the chain of transmission. How can one identify the “nuggets” of ideas that remain unchanged during this process and thus persist “for enough generations to serve as a unit of natural selection”? Dawkins argued that “selfishness is to be expected in any entity which deserves the title of a basic unit of natural selection.” His definition of the gene qualified as such an entity because it possessed three properties that “a successful unit of natural selection must have . . . longevity, fecundity, and copying fidelity.” Unlike his careful definition of the gene, Dawkins was somewhat vague about the definition of a meme, simply stating that this was a “unit of cultural transmission, or unit of imitation.” Is there some way to define a meme so that it possesses the properties that would qualify it as a unit of natural selection and hence deserve the “selfish” label?
Consider the 215 pages of the first edition of The Selfish Gene. Dawkins’s slim volume contains many ideas influenced by older texts and itself has influenced ideas expressed in newer texts (including this one). Can The Selfish Gene be parsed into a set of selfish memes each displaying longevity, fecundity, and copying fidelity? But do we similarly parse the genome into discrete genes? Dawkins’s definition of the gene did not specify boundaries between genes. The gene was a piece of chromosome sufficiently short to last long enough, without recombination, to function as a unit of selection. But this definition meant that there were many different, overlapping ways in which a chromosome could be divided into genes. Could the same approach work for memes?
Dawkins’s principal interest was in the phenotypes of organisms rather than of genes. Just as he failed to specify precisely how to divide a chromosome into genes, he did not specify how to divide the phenotype up into individual adaptations due to individual genes. I believe this approach was justified for his purposes. As long as all parts of the genome have the same rules of inheritance, what is good for one part of the genome is good for all parts of the genome, and the genome itself can be considered as an adaptive unit. Highly complex adaptations require a long genetic text. There are two widespread solutions to this problem that one can call the asexual and the sexual solution. In the asexual solution, an entire genome replicates as a unit and does not recombine with other genomes. Thus, the whole genome behaves as a single Dawkinsian gene. In the sexual solution, two entire genomes come together for some length of time, then separate into two new genomes after exchanging interchangeable parts, with each new genome receiving one of each part. The sexual genome is an ephemeral collective of many Dawkinsian genes, but the rules of Mendelian inheritance ensure that what is good for one part is good for all, at least for the time that the genes are temporarily associated. (I leave to one side the complexities that arise when the “rules” are broken and there is conflict within the genome.)
Neither the sexual nor the asexual solution seems to apply to most complex memetic “texts.” Ideas recombine freely to generate each new text, and there is no well-defined exchange of interchangeable parts. One idea can be adopted from a text and the remainder abandoned. Therefore, the adaptations of memes will be adaptations for the good of the individual rarely recombining ideas. Some of these ideas may be so simple—for example, the idea that the gene is a part of a chromosome—that they can exhibit few, if any, adaptations for their own transmission. I see little value in treating such ideas as selfish, just as there is little value in treating a single nucleotide as selfish. Such ideas serve the utility of propagandists or (perhaps) of larger nonrecombining meme complexes into which they become incorporated. The place to look for sophisticated adaptation and selfishness will be in coherent ideologies, large “asexual” meme complexes that are transmitted as a unit with high fidelity of transmission. Richard Dawkins would identify the world’s great religions as the prime examples and would argue that the free recombination of ideas is important if ideas are to serve our ends rather than their own.
These are some of the problems I see in defining memes and thinking of memes as selfish. And yet, in the four decades since I first read The Selfish Gene, no section of the book has stayed more in my mind than its final chapter. In the intervening period, I have disseminated the selfish meme many times in conversation, and here I am spreading the meme in print. The meme of the “meme” is a tenacious beast, at least for those minds that are vulnerable to its charms. The current chapter is a work of propaganda. I wish to communicate ideas that I hope will influence your own concepts of genes and memes. If I am effective, you may pass on these ideas in modified form to others. In pursuit of these ends, I have crafted phrases to grab your attention, and have worked on clarifying concepts in my own mind. This process has involved testing numerous alternatives against the standards of what I think will be effective and what will form a coherent whole with the rest of the chapter. I have read and revised my text innumerable times. I think of the final product as expressing my own intentions rather than serving the ends of the ideas it tries to communicate. But am I fully autonomous in this process? Many ideas have competed for inclusion during the course of writing, but only some have made it into a final version that has nothing like the form and content that I intended when I first sat down to write. It is only in retrospect that I know what I mean. The final version contains the ideas that have grabbed my attention. It has sometimes seemed that they have been using me for their ends. What fraction of these ideas are my own and what fraction have been borrowed from others? The web of intellectual influence is complex, and it is unclear whether I ever have a truly original idea.
Darwin’s disciple, the Oxford biologist-cum-psychologist George Romanes, wrote in Darwin, and After Darwin:
Quite apart from any question as to the hereditary transmission of acquired characters, we have in this intellectual transmission of acquired experience a means of accumulative cultivation quite beyond our powers to estimate. For . . . in this case the effects of special cultivation do not end with the individual life, but are carried on and on through successive generations ad infinitum. . . . [In] this unique department of purely intellectual transmission, a kind of non-physical natural selection is perpetually engaged in producing the best results. For here a struggle for existence is constantly taking place among “ideas,” “methods,” and so forth, in what may be termed a psychological environment. The less fit are superseded by the more fit, and this not only in the mind of the individual, but through language and literature, still more in the mind of the race. (1895, vol. 2, 32)
Richard Dawkins exhorted us to ask: “fit” in what sense and for whom?