The Structure of Evolutionary Theory

by Stephen Jay Gould

  Lower levels in a hierarchy cannot be deemed inherently simpler, either to conceive or to operationalize, than higher levels. If we had been brought up in an intellectual world that emphasized populations, rather than organisms, as primary entities, we would probably regard interdemic selection as maxi­mally simple, and organismic selection as an unwelcome complication. A pri­ori preference for lower levels represents a claim for reductionism, not parsi­mony. I do not say that such a preference therefore becomes invalid; I simply ask evolutionists to recognize the proper status of Williams's claim as an ar­gument about reductionism — and also to acknowledge that reductionism, as a cultural prejudice, may be far harder to defend than true parsimony, when properly invoked as a logical principle (though aspects of our preferences for parsimony may rank as cultural prejudice as well).*

  In Western science, which developed with such strong traditions for explanation by analytic division into constituent parts, claims for reduction have often been mistakenly advanced in the name of parsimony — most notably in biophilosopher C. Lloyd Morgan's early 20th century dictum that no human activity should be explained by a higher psychological faculty when a lower faculty suffices.

  This inappropriate invocation of parsimony did not disable Williams's argument because he usually proceeded beyond this theoretical point. That is, Williams extended his argument further by presenting direct evidence favor­ing the organismic mode in particular cases. He wrote: “This conclusion sel­dom has to rest on appeals to parsimony alone, but is usually supported by specific evidence” (1966, p. 19). But subsequent developments force us to [Page 554] consider one of the most troubling phenomena in the sociology of science — the principle of epigones and bandwagons.

  Williams himself did not abuse, rigidify, misconstrue, or unduly simplify his criteria — but his followers did, both early and often (to cite the classical principle for voting in Boston local elections), as Williams's “doctrine” be­came a dogma among his epigones. Few aspects of academic life can be more distressing and ironic than the common observation that a fine scholar often becomes a victim of his own success in this manner — but subtle po­sitions can be trivialized to sound bites in science as well as in political culture.

  “Genie or organismic selection only” became the bandwagon slogan of the late 60's and 70's. Combined with a strong preference, already established as the Synthesis developed, for hardline adaptationism in general (see previous section), this restriction set a predisposition strong and exclusive enough to be labeled as a dogma: interpret all substantial phenotypic characters as ad­aptations built by natural selection in the organismic mode (or lower). This dictum did not always function as a cleansing wind in a former stable, but all too often as a narrow and misdirected tunnel that carried a necessary reform too far. Moreover, many epigones used the dogma as a kind of linguistic game rather than a guide to research: “Can I tell a clever story to render this or that puzzling phenomenon as an organismic, rather than a group, adaptation?” For some evolutionists, the ability to spin such a tale, and to answer such a challenge as a theoretical affirmation, became the goal of a supposedly scien­tific effort. I have never witnessed a more distressing bandwagon in science, or seen any idea of such salutary origin pushed so far in the direction of thoughtless orthodoxy.

  (Pardon a personal incident, but I remember raising a question, early in my career, at a session of the first ICSEB meeting in 1973.1 asked a speaker, fol­lowing his formal presentation, if the dwarfed size of Pleistocene mammals on Mediterranean islands might have been favored by resistance to extinction afforded by the correlated effect of larger population sizes (than full-bodied hippos and elephants could have maintained in such small places). I hadn't thought the issue through, and I may well have been making a dumb sugges­tion, but the speaker's response floored me (and stunned me into silence at this ontogenetic stage of early diffidence). He said this and only this — and his words, with their intended dripping irony, still cut through me — “are you re­ally satisfied with a group selectionist argument like that?” He made no at­tempt to rebut my suggestion with any content whatever; the stigma of group selection sufficed for refutation.)

  As a final illustration of how reform, once established, can turn into the opposite phenomenon of rigidification, I interviewed Sewall Wright several times during the last decade of his life. He felt hurt by what he interpreted as his exclusion from the Modern Synthesis (beyond the ritualistic invocation of his name within the founding trinity of population genetics). “I was out of it,” he told me. He explained this passage into obscurity as the failure of a new generation of evolutionists to understand either his intended role for genetic [Page 555] drift, or his proposed mode for the operation of drift within his “shift­ing balance” theory of evolution.

  Most evolutionists of the 1960's viewed genetic drift only as a random force of evolutionary change — a prime anomaly under adaptationist harden­ing (or at least a factor relegated to the marginal role of efficacy only in tiny populations at the brink of extinction). Since genetic drift bore such a promi­nent association with Wright's name (extending to its original designation as the “Sewall Wright effect” in early days of the Synthesis), such demotion to marginality relegated the author to a similar fate under hardline adaptationism.

  This situation is surely unfair enough in itself, but — and now the irony — Wright had also participated in the adaptationist hardening of the Synthesis (see pp. 522–524), and his later interpretation of genetic drift invoked this concept primarily as an aid to an enlarged style of adaptationism, and not as a contrary force in evolutionary change (as he had originally argued). So if Wright had tried to be helpful in the service of orthodoxy, why did he become so misunderstood and relegated to the sidelines?

  Many reasons might explain Wright's fall into limbo, including the opaque character of his highly mathematical writing and the fact that he had invoked genetic drift as a nonadaptive force for change in his earlier work (see p. 523). But the major factor almost surely resides in a failure of evolutionists to un­derstand the multi-leveled character of his theory — the aspect that allows drift to serve as an input to an adaptive process. Wright told me (in an inter­view in 1981) that he originally intended to call his shifting-balance process the “two-level theory,” for the full process relies on essential components of both organismic and interdemic sorting. (Wright also told me that he re­garded “exclusive focus on individual selection” as the major error of the Synthesis.)

  In Wright's later formulation of “shifting balance,” drift enters into a cre­ative and selective process in the following manner: The founding population of a new species moves by selection to an adjacent adaptive peak in a larger landscape. The chief problem for adaptationism then intrudes: this initial peak will probably not represent the most favored spot in the landscape (for other unoccupied peaks probably stand higher), but how can daughter demes ever move to these better places? Valleys of lesser fitness surround this local peak. If evolution always proceeds towards adaptation, then the initial popu­lation must remain stuck on the first peak forever. But drift allows small groups to enter valleys, and to cross troughs into areas where selection may then draw populations up to a higher peak. When expanding populations, by this process, occupy several peaks in the landscape, a process of interdemic (interpeak) sorting can occur, eventually leading to a mean increase in adap­tation for the species as a whole. Thus, genetic drift does not operate as a ran­dom force against adaptation in Wright's mature theory, but as a source of variability for fueling a higher-level process of interdemic sorting. In other words, drift operates as part of a process that enhances adaptation through higher-level sorting. [Page 556]

  This argument seems clear enough in logic (its validity, or relative fre­quency, in the world of real populations raises a different issue that can only be resolved empirically). Why was Wright's theory misunderstood or, even worse, simply ignored? I suggest the following (and unfortunate) primary
reason: How could Wright's argument be grasped by an intellectual commu­nity now committed to the exclusivity of organismic selection? The concep­tual tools no longer exist under such a stricture: adaptation arises by “strug­gle” for reproductive success among genes or organisms; drift causes a population to depart from a place attained by such struggle. How then could drift possibly act as a helpmate to adaptation? To grasp Wright's view, one must allow for a higher level of interdemic sorting, and one must understand the logic of hierarchical models, with sorting operating at several nested levels.

  The intellectual space for viewing drift as an aid to interdemic selection doesn't exist in a context of exclusive commitment to selection at genic or organismic levels. Wright's idea then gets demoted to a status even lower than “merely wrong” in such a world; “shifting balance,” in Wright's own sense, becomes inconceivable, and therefore intellectually inaccessible. Erro­neous ideas can at least be expressed and made available to others with po­tentially different opinions. But the definitions of orthodoxy simply erased Wright's multiple-level theory — in much the same way that evolutionary sta­sis could not be recognized as interesting, or even grasped as a phenomenon at all, within a community committed to gradualism. When we think an idea through, and then reject the notion, we have at least made an intellectual decision (perhaps wrong, perhaps overly rigid). But when we maintain an unarticulated and unexamined commitment, and then use such a premise, al­beit unconsciously, to render interesting ideas inconceivable, then we have fallen under the spell of dogma.

  Sewall Wright — unlike Schubert, Wegener and a host of historical figures deemed tragic — lived long enough to witness his vindication (and to partici­pate mightily in his renewed respect by writing a four-volume mathemati­cal treatise, largely during his eighties — see Wright, 1978). But his period of unjustified eclipse should warn us all about the dangers of bandwagons and unexamined commitments.

  EXTRAPOLATION INTO GEOLOGICAL TIME

  A good flavor of the confidence, even the dogmatism, of the hardened synthe­sis, as presented at the Darwinian centennial celebrations of 1959 (see pp. 569–576), shines forth in Mayr's introductory proclamation from his 1963 book. Mayr pronounces the “complete unanimity” of competent professional opinion, the “colossal ignorance” of the “few dissenters,” and the conse­quent “waste of time” involved in any refutation of the intellectual stragglers:

  When we reread the volumes published in 1909, on the occasion of the 50th anniversary of the Origin of Species, we realize how little agreement [Page 557] there was at that time among the evolutionists. The change since then has been startling. Symposia and conferences were held all over the world in 1959 in honor of the Darwin centennial, and were attended by all the leading students of evolution. If we read the volumes result­ing from these meetings ... we are almost startled at the complete una­nimity in the interpretation of evolution presented by the participants. Nothing could show more clearly how internally consistent and firmly established the synthetic theory is. The few dissenters, the few who still operate with Lamarckian and finalistic concepts display such colossal ig­norance of the principles of genetics and of the entire modern literature that it would be a waste of time to refute them. The essentials of the modern theory are to such an extent consistent with the facts of genetics, systematics, and paleontology that one can hardly question their correct­ness (1963, p. 8).

  Later on, Mayr proposes a succinct definition of the Synthesis, emphasiz­ing all three legs (or branches) of the essential tripod (or tree) of Darwinian logic “The proponents of the synthetic theory maintain that all evolution is due to the accumulation of small genetic changes, guided by natural selec­tion, and that transpecific evolution is nothing but an extrapolation and magnification of the events that take place within populations and species” (1963, p. 586).

  I regard Mayr's balance of emphases as particularly revealing. His defini­tion includes two phrases. The first statement buttresses the two legs of Dar­win's tripod that receive most attention in this work — control of direction by external selection rather than internal constraint (with attendant gradualism of change), and operation of the process through differential reproductive success of organisms (implicit in the term “natural,” as opposed to some other form or level of selection). But the second, and longer, statement affirms Mayr's appreciation for the importance of the third leg — the complete suf­ficiency of microevolutionary theory to explain the entire history of life — so long as the earth's geological behavior sets a proper stage, and does not derail a full extrapolation of microevolutionary mechanics into all geological time and to the entire extent of phylogenetic change. For how can we celebrate the power and generality of a beautiful, sufficient and completely validated mechanism of change, established for the immediacy of an ecological here and now, if the same processes cannot render the broad pattern of life's his­tory as well?

  This theme of extrapolation becomes, in many ways, the most comprehen­sive issue of all. In Chapter 2, I distinguished two separate aspects of Dar­win's radicalism in proposing the theory of natural selection — a methodologi­cal pole to grant operational status to the study of evolution by asserting that observable events, however apparently trivial and inconsequential, can ex­plain change at all scales by extension; and an ideological pole to present a radical mechanism of evolutionary change for rendering all “higher” attri­butes of good design and organic harmony as side-consequences of a process [Page 558] working only by a struggle among organisms for personal reproductive suc­cess. Strict Darwinians must defend extrapolation as crucial at both poles — for observable events at small scale cannot generate the full panoply of phylogeny without such a principle; while daily happenings cannot accumulate into totalities if the background setting does not “behave” properly — if, for example, the ecological stage explodes every once in a geological while, fortu­itously dumping most of the improved and accumulated inhabitants into a vat of extinction.

  For these two reasons — to validate the entire methodological pole and to uphold the third leg of the Darwinian tripod supporting the ideological pole — the principle of extrapolation represents the key to the validity of the Synthesis as a fully general theory of evolution. And extrapolation — the es­sential Lyellian postulate that Darwin imbibed from his most important men­tor (see p. 94) — embodies two central aspects of what Lyell and his school called “uniformitarianism”: (i) the complete theoretical sufficiency of cur­rently acting small-scale changes to produce, by successive and imperceptible increments, the entire panoply of large-scale phenomena; and (ii) the proper “behavior” of the earth, with geological change sufficiently slow and steady that trends produced by gradualistic, accumulative natural selection will not be derailed often enough to yield a history of life patterned more by these geological upsets than by biological accumulations.

  For this last leg of the tripod, the Synthesis did not so much harden as be­come emboldened during its ontogeny. We have seen (pp. 514–518) how Haldane and Huxley, in the early days of the Synthesis, still respected (even feared) apparent paleontological exceptions to natural selection as the cause of trends. But the balance of power had shifted by the 1960's. Simpson (1944, 1953) and others had forged their “consistency argument” — the principle that all known phenomena of the fossil record can, in principle, be explained by modern mechanisms of genetics and selection (even though no direct proof of sufficiency can be derived from paleontological evidence). Paleontology had been tamed, taken in by the synthesis, and told to behave. (And I do in­tend “taken in” in the metaphorical sense as well, as I shall argue more ex­plicitly in Chapter 9.) Paleontology could retain the archives of actual phe­nomenology as its particular bailiwick in exchange for giving up the conceit of believing that the fossil record could say anything distinctive about the causes of evolutionary change.

  The distinguished panel on “the evolution of life” at the Chicago cen­tennial celebration of 1959 includ
ed Julian Huxley as chairman, Th. Dobzhansky, E. B. Ford, Ernst Mayr, Ledyard Stebbins, and Sewall Wright. After an orthodox discussion of mechanisms, Huxley shifted the topic to “the course, the process, of evolution as shown in the fossils” (Tax and Callender, 1960, volume 3, p. 127) — and paleobotanist D. I. Axelrod rose to present a summary in advance, characterizing the history of life as a stately process of unfolding to more and better: “I think most of us are in full agreement about the gradual change in time: increasing diversification; then, gradual transformation, [Page 559] so new categories gradually arise, first at smaller and then at higher levels” (in Tax and Callender, 1960, volume 3, p. 127).

  In the panel's stated agenda of 16 points, only one even hints at non-adaptive phenomena, and only as an adjunct to selectionist orthodoxy. Huxley addressed this topic midway through the discussion: “Natural selec­tion may lead to side effects, which at the same time are of no adaptive value but may later provide the basis for adaptive changes” (Tax and Callender, 1960, volume 3, p. 125). Vertebrate paleontologist E. C. Olson, the sympo­sium's lone (and very gentle) doubter (see pp. 574–576), ventured a percep­tive comment on this point. But his words, as Moses said of Pharaoh's chari­ots, promptly “sank as lead in the mighty waters.” The entire discussion of this topic occupies less than a page. Olson said: “This is the general area in which we can include events that are random with respect to the adaptive value of the genotype of populations. I refer to the simple matter of acci­dent — for example, the effects of a forest fire on a population ... This sort of side effect, the impact of accidents and other factors producing non-adaptive shifts, may cause very rapid changes and give completely new shape to the course of evolution. I think this is an extremely important evolutionary fac­tor” (in Tax and Callender, 1960, volume 3, p. 125).