I do not deny that modern English usage favors the negative connotations — hence my rationale for this introduction. But the positive meanings remain current, and certainly sanctioned both historically and linguistically. I began with Elihu's statement because, when I first studied the Bible as a teenager, this passage confused me. I did not yet know the positive meaning of constraint, and therefore couldn't figure out why Elihu, although practically bursting from his need to speak, felt so constrained that he dared not do so (even though he did so in the very next verse!). Of course, Elihu meant “constraineth” in the opposite and positive sense that his need would force, or constrain, the desired result — and that his words would pour forth in a definite and channeled direction, not as a random spewing.* This active sense [Page 1027] of “constraint” as promoting change in particular directions marks the positive meaning that motivates this chapter and raises an important issue for Darwinian theory in providing a best general case and opportunity for positive interaction with the functionalist precepts of natural selection.
The first (empirical) positive meaning of channeling
Orthodox Darwinian functionalists have often reacted by arguing “why the big fuss, we know (and both admit and use) the concept of constraint already” to evolutionists of a structuralist bent who claim that a properly formulated version of “constraint” should evoke great interest and provoke substantial reform. I would accept this jaded reaction if the version of “constraint” proffered by structuralist critics stayed within the negative meanings described above. But positive meanings of constraint — and I will outline two different constructions of positivity in this section — can lead to important extensions of evolutionary theory by questioning and reformulating what I have called the second branch, or the second tripod leg, of Darwinism's essential triad of indispensable arguments: the functionalist attribution of effectively all substantial evolutionary change to natural selection.
I agree that negative constructions of constraint do not seriously challenge this major precept (while supplying some interesting subtleties and wrinkles that orthodox functionalists can use and appreciate) for the following set of interconnected reasons: Mainline Darwinism is a functionalist theory of “trial-and-error externalism” (in R. C. Lewontin's phrase). The organism “proposes” by generating variation, ultimately by mutation (and subsequently by distribution in sexual recombination for organisms traditionally deemed “higher”), among members of populations. This variation acts as raw material — the “chance” in Monod's famous metaphor of “chance and necessity”; the “error” in Lewontin's “trial-and-error” — for a causal process of natural selection (the “necessity” in Monod's pairing; the “trial” of Lewontin's joining). That is, the organism proposes, and the environment (interacting with the organism) disposes.
The organism's generation of variation provides the internal component of evolution; the environment's process of selection marks the external contribution. These internal and external factors play strikingly different roles in Darwinian theory — a contrast well epitomized (as noted above) in Monod's phrase “chance and necessity.” The internal component can only supply raw material and does not establish the rates or vectors of change. This claim — that variation provides potential, but not direction — sets a fundamental postulate of Darwinian mechanics and philosophy. Natural selection, the external component, carries full responsibility for the direction — and also, ultimately, for the modes and rates — of evolutionary change.
As discussed at length in Chapter 2, Darwin's central insight that variation must be “isotropic” — particularly, that it be copious in amount, small in extent, and undirected towards adaptive configurations — underlies his brilliant grasp of what selection requires from variation to permit a functionalist theory to operate in principle, and also to dominate the causes of evolutionary change. If variation is truly isotropic in Darwin's hypothesized sense, then selection [Page 1028] gains free rein (and reign) as the cause of change — in sharp contrast to nearly all competing 19th century evolutionary theories, with their stress on internally generated directionality.
To epitomize the theoretical importance of constraint in a single sentence, the concept of non-isotropy in variation may be roughly synonymized with notions of “constraint” — that is, with claims that internal factors restrict the freedom of natural selection to establish and control the direction of evolutionary change.
The crucial importance of constraint in evolutionary theory therefore centers upon this potential challenge, particularly upon the nature and scope of the restrictions. In Chapters 4 and 5 — the longest section of this book's first half, because the subject consistently commanded the principal attention of Darwin's most cogent critics — I considered both pre and post-Darwinian theories of evolutionary internalism, as rooted in the common claim that any significant non-isotropy of variation would reduce, or even cancel, the creative role of natural selection by identifying potent internal forces of evolutionary change in either directed or saltatory variation. These two persistent bugbears of “quick” and “channeled” received their most memorable (and influential) joint expression in the challenging metaphor of Galton's polyhedron (see pages 342–351 of Chapter 5).
Of course, no sophisticated Darwinian ever denied some limited domain of validity to the concept of internally generated “constraints.” (As I have emphasized throughout this book, validation in natural history rarely follows the criterion of “never in principle for this would violate nature's laws,” as favored in some constructions of the so-called exact sciences, but rather the standard of “conceivable in principle, but not occurring often enough to matter,” as followed in historical sciences that formulate most basic judgments by analysis of relative frequencies.) Rather, Darwinian functionalists have tended to admit certain kinds of constraints, and have then tried to limit their modes of occurrence and domains of action in such a manner that the central principle of Darwinian theory — the control of evolutionary change by natural selection — will not be threatened.
In short, and to summarize these few pages of argument in a paragraph, orthodox Darwinians have not balked at negative constructions of constraint as limits and impediments to the power of natural selection in certain definable situations. But they have been far less willing to embrace positive meanings of constraint as promoters, suppliers, and causes of evolutionary direction and change. This distinction follows logically from the basic premises of Darwinian functionalism, because the admission of a potent and positive version of constraint would compromise the fundamental principle that variation (the structuralist and internalist component of evolution) only proposes, while selection (the functionalist and externalist force) disposes as the only effective cause of change.
In considering how structural constraints might limit the power of natural selection to adapt each feature of an organism to each local environment, we recognize that some modes will rank as “benign” for Darwinian [Page 1029] functionalists; others as less benign but not subversive to orthodox theory; whereas still others — particularly the positive modes that promote, and do not just limit, evolutionary change — do pose a deeper theoretical challenge, and have therefore set the major battleground of a subject that has, of late, become both highly confusing and maximally contentious in the literature of evolutionary biology (Gould, 1980c; Alberch, 1982; Maynard Smith et al., 1985; Stearns, 1986; Antonovics and van Tienderen, 1991; Schwenk, 1995; Duboule and Wilkins, 1998; Eble, 1999).
The most benign category does not restrict the organism's potential for reaching a best-adapted overall phenotype for a particular environmental background — and therefore only counts as “constraint” with respect to unrealizable and idealized abstractions. To cite two examples, so called “trade offs” preclude separate optimization of each part because natural selection works upon the entire organism as a totality. The best-adapted whole cannot evolve as
a simple summation of separately optimized parts because, in an integrated structure that must function as a single coherent entity, the “perfection” of some parts can only be achieved at the expense of others. Therefore, to cite an old conjecture for illustrating the obvious, the optimal size for a human brain at birth may be too big to allow the passage of a neonate through the birth canal. But such structural constraints, imposed by a selected whole upon individual parts not subject to independent optimization in any case, do not challenge, but rather affirm, the central Darwinian postulate that selection works on organisms.
Secondly, mechanical limits (also structural or formal in character) obviously preclude certain solutions that might offer abstract advantages in adaptation. Zebras could avoid feline predators by flying away, but even if genetic variation existed (as it almost surely does not!) for constructing a supernumerary pair of limbs in wing like form, zebras clearly exceed permissible weight limits under the venerable Galilean principle of declining surface to volume ratios in large creatures.
I cite both these examples tongue-in-cheek because no one would view such obvious, and evidently “benign,” classes of structural constraint as challenges to Darwinian adaptationism (or even as particularly interesting in any intellectual sense). Darwinian functionalism works by local adaptation of integral organisms to immediate environments. Neither biomechanical optimization part by part (prevented by “trade-offs” or integral constraints), nor putatively advantageous configurations outside the limits of mechanical possibility (physical or formal constraints), poses any challenge to the tenets of Darwinian functionalism.
In a less benign, theoretically relevant (although ultimately not debilitating), and widely discussed category, limitations based upon absence of sufficient variability to provide raw material for natural selection (and usually called genetic or developmental constraints) do operate widely in nature. (Since natural selection “makes nothing” by itself, but can only operate upon raw material supplied by an independent process of variation — a statement familiar enough to rank as a “mantra” among Darwinian evolutionists — a [Page 1030] shortage of building stones can slow, or even derail, the construction of a well designed house.)
In the “consensus paper” of Maynard Smith and eight prominent colleagues associated with a wide range of views from Darwinian orthodoxy (Lande and Maynard Smith himself) to serious structuralist heterodoxy (Raup, Goodwin and Kauffman), “limitations on phenotypic variability” (1985, p. 269) became the nucleating point of agreement in their remarkable exercise in intellectual diplomacy. Therefore, “developmental constraint” in this sense of limitation in necessary raw material to fuel the workings of natural selection, has become (in a minimalist interpretation that I do not challenge) the canonical “base-line” or “common ground” definition for this important structural component in Darwinian theory. (In fairness, Maynard Smith et al., do acknowledge positive meanings of constraint as both legitimate and more interesting (see p. 1037), while advocating this “negative” definition as a minimal standard that all evolutionists can embrace, and that no Darwinian need regard as dangerously debilitating.)
Thus, to return to my previous and facetious example, zebra wings would not work for the reason cited above, but natural selection will presumably never encounter an opportunity even to attempt their construction because sufficient variability for a supernumerary pair of limbs presumably does not exist in the genetic and developmental systems of tetrapods. In a more meaningful category — representing a frustrating and unresolved issue that has permeated Darwinian discussion ever since the eponym himself — the limited range of realized phenotypes in some clades (with domestic breeds of cats vs. dogs as the classic example) may reflect a structural limit in variation, rather than a lack of selective opportunity or advantage.
An important functionalist principle of natural selection, frequently (and quite explicitly) emphasized by Darwin, holds that we may, in operational terms as the “null hypothesis” of our initial assumptions for empirical testing, treat populations as though they always possess sufficient variation to permit natural selection an unimpeded range of action. As a practical expression of this basic Darwinian belief, rates of evolutionary change fall under the control of natural selection, not of limitations (or superfluities) in raw material. Clades that either change slowly, or fail to generate many species, should be regarded as subject to little selective pressure, not as limited by intra-populational variation. If cats have developed far fewer varieties than dogs, then the differential selective efforts of human breeders, rather than any disparity in the ranges of available raw material, should explain the striking difference.* [Page 1031] Darwin wrote to Lyell in 1862 (quoted in F. Darwin, 1903, vol. 2, p. 338 — see pp. 330–341 for further discussion, and especially p. 341 for Darwin's comprehensive architectural metaphor for dismissing constraint as a theoretical challenge to natural selection):
Mere variability, though the necessary foundation of all modifications, I believe to be almost always present, enough to allow of any amount of selected change; so that it does not seem to me at all incompatible that a group which at any one period (or during all successive periods) varies less, should in the long course of time have undergone more modification than a group which is generally more variable.
Placental animals, e.g. might be at each period less variable than Marsupials, and nevertheless have undergone more differentiation and development than marsupials, owing to some advantage, probably brain development.
I label this negative category of constraint based on lack of sufficient variability as “less benign” because its operation does place a genuine damper, both actual and theoretical, upon the exclusivity of natural selection as the cause of evolutionary change. Darwin himself certainly read the issue in this light, as the above quotation indicates — for if his argument fails, and constraint often trumps selection as a regulator of evolutionary rate, then his resulting disappointment, in needing to recalibrate and downgrade the relative importance of natural selection, will evidently be severe.
But this far even the most devoted selectionist must proceed — for the logic of this “less benign” category cannot be gainsaid. The basic formulation of the theory of natural selection does require structural input of raw material by variation to fuel the functional outcome of evolutionary change by selection. (And since natural selection cannot, in principle, manufacture this necessary fuel for its own operation, lack of input can stymie output — just as the niftiest motor car can't move if you run out of gas in the middle of the Sahara, hopelessly far from the nearest petrol station in Timbuctu.) Therefore, one cannot brand limitation in raw material as an incoherent concept, or even an empirical rarity, a priori.
And yet, if selectionists can hold the line — as they generally attempt to do — at this negative definition of constraint, their theory, while deniably impacted, suffers no serious setback to its truly essential postulate that natural selection controls the direction of evolutionary change. At most, the negative forces of constraint may slow down, or even prevent, modifications. But so long as these structural factors do not operate in a positive sense — either to determine important variation in rates and extents of change or, more threateningly, to impact or set the actual direction of change — then the fundamental Darwinian rule still prevails: variation proposes, but only natural [Page 1032] selection can dispose. Internal forces supply the possibilities; but natural selection builds the pattern (because the possibilities nearly always exist in sufficient abundance to fuel the changes that natural selection might favor). Constraints impede, but do not direct.
Thus, a standard Darwinian “truce” accepts the notion, even the potential importance, of constraint as a negative force that can impede rates and amounts of change (and therefore cannot be dismissed as irrelevant to the generation of evolutionary pattern in this limited sense). But strict Darwinian functionalists generally t
ry to hold the line by denying importance, or even legitimacy, to positive definitions of constraint as causes of directionality in evolutionary change.
In one of the best examples I have ever encountered of the vital (almost morally enjoined) principle that the forebears of our current struggles demand our continuing study and respect — and that we often gain, as recompense for this fealty, substantial practical benefit in getting our own thoughts straight, and channeled in useful directions — the rich history of debate about Darwinian theory has brought the theme of positive constraints into sharp focus (see Chapters 4–5). This clarity emerges from the common emphasis placed by all major structuralist critics upon (1) the difference between positive and negative meanings of constraint (accompanied by specifications that only the positive meanings could pose serious difficulties for Darwinism), and (2) the parsing of positivity into two essential themes of speed, or enhancement of rates beyond the power of natural selection to instigate, and channeling, or the preferential (perhaps even requisite) flow of change in particular directions set by internal possibilities, even if natural selection must supply an initial impetus. We should also note the ironic sense in which this argument inverts the canonical roles of the two central components in Darwinian theory. In natural selection, an internal source of variation provides the impetus, whereas selection determines direction. In channeled change by constraint, natural selection supplies the impetus by “getting the ball rolling” (to use Galton's metaphor of the pool table), but the directionality of evolutionary change, or “where the ball rolls,” emerges from internal channels that, so to speak, “use” natural selection as their convenient source of power. In short, variation as raw material and selection as the shaper of change in Darwinism; vs. selection as raw power, and channeled variation for shaping in theories of positive constraint. Such an epitome, needless to say, remains far too simple to resolve nature's ways — but this formulation does embody a clear and useful conceptual dichotomy for clarifying our thoughts.
The Structure of Evolutionary Theory Page 164
