The Evolutionary Definition of Selective Agency and the
Fallacy of the Selfish Gene
A FRUITFUL ERROR OF LOGIC
Science thrives upon the continuous correction of error. Most errors arise from inadequate knowledge of the empirical world, or (if grounded in a theoretical prejudice) at least persist because we have no means (conceptual or technological) to secure their empirical refutation. For example, we once lacked the technology to prove that buried organic matter might petrify, and that wood made of stone might therefore represent the remains of ancient plants, and not the power of rocks to mimic organic design by a process analogous to crystallization.
Only rarely, however, do professions get sidetracked by pursuing an extensive and long-lasting program of research initiated by an error in reasoning rather than an inadequacy of empirical knowledge. Yet I think that the gene-centered approach to natural selection — based on the central contention that genes, as persistent and faithful replicators, must be fundamental (or even exclusive) units of selection — represents a purely conceptual error of this unusual kind. In beginning with Williams's manifesto (1966) — based on a mode of thinking rooted in the brilliantly consistent, if limited, worldview of R. A. Fisher (1930), but immediately inspired by the remarkable work of W. D. Hamilton (1964) — and proceeding through the codification of Dawkins (1976), to numerous works both popular (especially Cronin, 1991) and technical (Dennett, 1995), this gene-based approach to selective agency has inspired both fervent following of a quasi-religious nature (see R. Wright, 1994), and strong opposition from many evolutionists, who tend to regard the uncompromising version as a form of Darwinian fundamentalism resurgent (see Gould, 1997'd), variously designated as ultradarwinism (Eldredge, 1995) or hyperdarwinism. [Page 614]
I shall show in this section that, while genes may be appropriately designated as fundamental replicators (under a defendable but nonexclusive strategy of research), replicators simply aren't units of selection or, for that matter, causal agents at all under our usual notions of mechanism in science. The misidentification of replicators as causal agents of selection — the foundation of the gene-centered approach — rests upon a logical error best characterized as a confusion of bookkeeping with causality.
We fall into another serious fault of reasoning when we accept the common conceptual taxonomy that relegates error itself to a purely negative category of unfortunate blunder. Some errors do lead only to blind alleys and wasted time. But others, as thoughtful scientists have always recognized, serve as essential prods and directors of progress through correction. Darwin's famous words, distinguishing harmful from salutary error, have frequently been cited in this context: “False facts are highly injurious to the progress of science, for they often endure long; but false views, if supported by some evidence, do little harm, for every one takes a salutary pleasure in proving their falseness” (from the Descent of Man). I prefer the stronger statement of the great Italian economist, Vilfredo Pareto: “Give me a fruitful error any time, full of seeds, bursting with its own corrections. You can keep your sterile truth for yourself.”
During my career in evolutionary science, no error has proven more fruitful in Pareto's sense than the gene-centered approach to selection. The central claim, clearly expressed, forced us to reconceptualize the entire domain of evolutionary causality. The outrageous character of such an ultimate reduction compelled us to rethink our subject by explicitly rejecting the oldest, most traditional and entirely commonsensical notions about our own bodies as agents. (Yet the reductionistic cast of the theory fit so well with conventional ideas about the goals of science that many biologists “caught the spirit” and “followed the program” despite its assault upon ordinary intuition.) Nevertheless, the theory could not work. However stubborn and heroic the attempt, explanation inevitably faltered upon the central logical error — especially when selection had clearly worked upon emergent properties of higher-level individuals, and no verbal legerdemain could recast the story in terms of genes as causal agents. If “Pareto errors” contain the seeds that burst their own boundaries, then such uncommon errors of fallacious reason (rather than absent fact) qualify best for this status. Empirical correction usually requires a period of waiting for new technologies or new discoveries (as the sources of resolution do not lie within the argument), but logical errors always carry the seeds of correction within the fruit of their own structure.
HIERARCHICAL VS. GENIC SELECTION
The fallacy of gene selectionism, and the consequent validity of the alternative (and opposite) hierarchical model of selection, can best be expressed in a series of seven arguments and vignettes — of different length, but all connected in a logical order, and all developed for the same import and purpose: [Page 615]
The distinction of replicators and interactors
as a framework for discussion
Both leading founders of modern gene selectionism as a general view of evolution (Williams, 1966; Dawkins, 1976) drew a crucial distinction between reproductive units of heredity, and entities that interact with the environment to bias the transmission of reproductive units into the next generation. Williams viewed nearly all evolution as proceeding via genes as reproductive units, with adaptation of organisms (the interacting entities) construed as a result — a duality that he usually labeled (1966, p. 124 for example) as “genie selection and organic adaptation.” Dawkins (1976) agreed entirely, and drew a more colorful and explicit distinction between “replicators,” considered as units of selection and identified as genes — and “vehicles,” considered as merely passive repositories built by replicators for their own purposes, and identified as bodies of organisms. In other words, both Williams and Dawkins invoked a criterion of replication to identify genes as the active and fundamental agents of natural selection.
In his 1980 review on “Individuality and selection,” David Hull formalized this distinction in a manner that has — quite usefully and properly in my view — organized the professional discussion on units of selection ever since. Hull (1980, p. 318) defined a replicator as “an entity that passes on its structure directly in replication”; and an interactor as “an entity that directly interacts as a cohesive whole with its environment in such a way that replication is differential.” Hull then defined selection with reference to both attributes: “a process in which the differential extinction and proliferation of interactors cause the differential perpetuation of the replicators that produced them.”
Hull insisted that a causal account of selection must include both concepts (1980, pp. 319-320): “Evolution of sorts could result from replication alone, but evolution through natural selection requires an interplay between replication and interaction. Both processes are necessary. Neither process by itself is sufficient. Omitting reference to replication leaves out the mechanism by which structure is passed from one generation to the next. Omitting reference to the causal mechanisms that bias the distribution of replicators reduces the evolutionary process to the 'gavotte of the chromosomes,' to use Hamilton's propitious phrase.” Later, Hull (1994, pp. 627-628) continued to espouse this view: “According to the terminology I prefer, there are no units of selection because selection is composed of two subprocesses — replication and interaction. Selection results from the interplay of these two subprocesses. Genes are certainly the primary (possibly sole) units of replication, whereas interaction can occur at a variety of levels from genes and cells through organisms to colonies, demes, and possibly entire species.”
I shall argue in this section that the causality of selection resides in interaction, not in replication, and that the hierarchical model almost automatically prevails once we accept this analysis of causality. Moreover, Hull's intuitions ran in this direction from the start, for even while he insisted upon the “relevancy” of both replication and interaction, Hull always acknowledged that [Page 616] the classical argument for multiple levels of selecti
on only invokes interactors. He wrote in his original paper (1980, p. 325): “In most cases when biologists argue that entities more inclusive than single genes function in the evolutionary process, they have interaction in mind.” And Hull (1994, p. 628) directly followed his defense of duality (quoted just above) with this sentence: “The units-of-selection controversy concerns levels of interaction, not levels of replication.” I shall defend and develop Hull's intuition in the rest of this section. Only interactors can be deemed causal agents in any customary or reasonable use of this central term. Replicators are important in evolution, but in a different role as items for bookkeeping. Replicators are not causal agents. If causality resides in interactors, and interactors at several levels rank as legitimate evolutionary individuals, then the hierarchical theory of selection becomes unassailable as a coherent logical structure, subject to the ultimate scientific test of empirical verification (or invalidation) in nature.
Faithful replication as the central criterion for the gene-centered
view of evolution
As noted above, both Williams and Dawkins chose to define units of selection as replicators rather than interactors. I shall explain under argument three why I am confident that they made the wrong choice — thus committing the fruitful “Pareto error” discussed at the outset of this section. Having thus decided, and correctly understanding that selection can only work on “individuals” as previously defined, what replicating individuals would Williams and Dawkins then designate as units?
We all know that they chose genes as fundamental — and effectively exclusive — replicators, and therefore as the unit of selection in Darwinian theory (in maximal contrast with the hierarchical theory of multiple, simultaneously acting levels, as defended in this book). I will discuss the stated reasons for their choice, but I cannot know the deeper motivations of their philosophical and psychological preferences. I strongly suspect that they, and all defenders of strict gene selectionism, feel drawn to the traditional reductionism of science. They understood that Darwin himself went as far as he could in this direction, by breaking down the Paleyan edifice of highest-level intentionality (God himself) to the lowest level then practical — organisms struggling for reproductive success (see Chapter 2). They also recognized that this breakdown had produced revolutionary consequences for Western thought, particularly in reconceptualizing all perceived natural “benevolence” (especially the good design of organisms and the harmony of ecosystems) as a side-consequence of struggle for personal success among lowest-level individuals, rather than as an explicit intention of a loving and omnipotent deity. I imagine that the more thoughtful gene selectionists then worked by analogy, reasoning that if they could break causality down even further, below the level of the organism, similarly interesting, and perhaps revolutionary, consequences might follow. I can't gainsay either the intuition or the ambition — but I can fault the resulting argument for an erroneous choice of both category and of level. [Page 617]
If a search for ultimate reduction below the Darwinian body set the deeper motivation for choosing genes as units of selection, what particular rationales did proponents of this theory offer? Both Williams and Dawkins began by arguing that the conventional unit of Darwinian theory — bodies of organisms — cannot properly occupy this role because organisms lack a key feature that genes possess. The bodies of sexual organisms disaggregate in reproduction, making only half an appearance (so to speak) in the genetic constitution of offspring. How can something so ephemeral be a unit of selection? But genes pass faithful copies of themselves into future generations, and therefore maintain the integrity required of an agent of natural selection in their definition.
Both Williams and Dawkins advance the same argument in three steps: (1) the unit of selection must be a replicator; (2) replicators must transmit faithful, or minimally altered, copies of themselves across generations; (3) sexual organisms disaggregate across generations and therefore cannot be units of selection, but genes qualify by faithful replication. Williams developed this argument in his first book (1966), and continues his verbal defense to this day, despite remarkable movement, as we shall see, towards the position advocated in this volume. But Williams still employs the language of gene-selectionism, particularly in the identification of genes as “units of selection” by virtue of faithful replication (so different from Hull's pluralistic view that the definition of a unit must include both replication and interaction): “These complications are best handled by regarding individual [i.e. organismic] selection, not as a level of selection in addition to that of the gene, but as the primary mechanism of selection at the genic level. Because genotypes do not replicate themselves in sexual reproduction (cannot be modeled by dendrograms), they cannot be units of selection” (Williams, 1992, p. 16).
Dawkins (1978) advances the same argument, with the same designation of genes as units of selection: “However complex and intricate the organism may be, however much we may agree that the organism is a unit of function, I still think it misleading to call it a unit of selection. Genes may interact, even 'blend' in their effects on embryonic development, as much as you please. But they do not blend when it comes to being passed on to future generations.”
In a later book (1982, p. 91), Dawkins affirms the terminology of genes as units of selection, by making a strong link to his favorite subject of adaptation: “The whole purpose of our search for a 'unit of selection' is to discover a suitable actor to play the leading role in our metaphors of purpose. We look at an adaptation and want to say, 'It is for the good of...' Our quest... is for the right way to complete that sentence ... I am suggesting here that, since we must speak of adaptations as being for the good of something, the correct something is the active, germ-like (sic, but clearly a misprint for the intended 'germ-line') replicator.”
Dawkins's extended defense of genes as the unit of selection invokes a set of related criteria bearing unmistakable concordance with primal virtues of our culture, another extrascientific reason for the argument's appeal — namely, faithfulness, (near) immortality, and ancestral priority. Dawkins enlarges the [Page 618] basic argument about faithfulness — sexual organisms disaggregate across generations but genes transmit accurate copies — into a paean about genetic immortality compared with the tragic transiency of our personal lives:
It does not grow senile; it is no more likely to die when it is a million years old than when it is only a hundred. It leaps from body to body down the generations, manipulating body after body in its own way and for its own ends, abandoning a succession of mortal bodies before they sink in senility and death. The genes are the immortals, or rather; they are defined as genetic entities, which come close to deserving the title. We, the individual survival machines in the world, can expect to live a few more decades. But the genes in the world have an expectation of life, which must be measured not in decades but in thousands and millions of years. In sexually reproducing species, the individual is too large and temporary a genetic unit to qualify as a significant unit of natural selection. The group of individuals is an even larger unit. Genetically speaking, individuals and groups are like clouds in the sky or dust storms in the desert. They are temporary aggregations or federations. They are not stable through evolutionary time (1976, p. 36).
Dawkins then commits one of the classical errors in historical reasoning by arguing that because genes preceded organisms in time, and then aggregated to form cells and organisms, genes must therefore control organisms — a confusion of historical priority with current domination (see Chapter 11, and Gould and Vrba, 1982, for a full discussion of this common fallacy). But Dawkins's argument collapses for many reasons, most notably the issue of emergence. A higher unit may form historically by aggregation of lower units. But so long as the higher unit develops emergent properties by nonadditive interaction among parts (lower units), the higher unit becomes, by definition, an independent agent in its own right, and not the passive “slave” of
controlling constituents. In advancing this false argument, Dawkins closes with a statement that can only compete with some choice Haeckelian effusions for the title of purplest prose passage in the history of evolutionary writing:
Replicators began not merely to exist, but to construct for themselves containers, vehicles for their continued existence. The replicators which survived were the ones which built survival machines for themselves to live in ... Survival machines got bigger and more elaborate, and the process was cumulative and progressive ... Four thousand million years on, what was to be the fate of the ancient replicators? They did not die out, for they are past masters of the survival arts. But do not look for them floating loose in the sea; they gave up that cavalier freedom long ago. Now they swarm in huge colonies, safe inside gigantic lumbering robots, sealed off from the outside world, communicating with it by tortuous indirect routes, manipulating it by remote control. They are in you and me; they created us, body and mind; and their preservation is the ultimate rationale for our existence. They have come a long way, those replicators. [Page 619] Now they go by the name of genes, and we are their survival machines (1976, p. 21).
One might dismiss this rhetorical flourish as harmless enthusiasm. But we must also recognize that, however extended the metaphors, Dawkins's images do accurately express his false theory of selective agency — for if genes can be depicted as exclusive units of selection, then they become the causal agents of evolution; and if bodies are Darwinian ciphers both for their transiency and by their lethargy relative to the “lean and mean” genes living within, then bodies might as well be described as inert and manipulated repositories (“lumbering robots”).
The Structure of Evolutionary Theory Page 98