By William A. Dembski


A review of Steven Hecht Orzack and Elliott Sober, eds., Adaptationism and Optimality, in Cambridge Studies in Philosophy and Biology (New York: Cambridge University Press, 2001), 416 pages, ISBN: 0 521 59166 X, $75.00 (hard), $28.00 (soft).




According to Darwinism, biological evolution proceeds without discernible plan or purpose. To be sure, biological evolution produces things that look planned or purposed. But what underlies Darwinian evolution ultimately is a blind mechanical process -- the Darwinian mechanism of natural selection.


Darwin himself argued that natural selection, though not capable of purposive action, is nonetheless capable of producing the appearance of purpose in nature. Indeed, he ascribed remarkable skill to natural selection. In the Origin of Species he wrote: "Natural selection picks out with unerring skill the best varieties." Darwin elaborated: "Natural selection is daily and hourly scrutinising, throughout the world, every variation, even the slightest; rejecting that which is bad, preserving and adding up all that is good; silently and insensibly working, whenever and wherever opportunity offers, at the improvement of each organic being in relation to its organic and inorganic conditions of life."


In this way natural selection accomplishes all of biology's design work without being an actual designer. Natural selection is not a watchmaker per se but a blind watchmaker, to use Richard Dawkins's apt phrase. Natural selection does not operate with actual plans or purposes. It does not look into the future and deliberate about what biological structures and functions might be worth innovating. Rather, natural selection is an opportunist that takes advantage of useful variations that arise as organisms reproduce.


For this reason, even though the word "design" appears in the biological literature, one is more apt to find reference to "adaptation," especially within the field of evolutionary biology. Natural selection takes advantage of useful variations. What makes them useful is that they help fit or adapt organisms better to their environments. Monkeys climbing on trees and in danger of falling find it useful to have a tail that grasps branches. Natural selection will tend to favor such tails, and as a consequence monkeys have evolved that can swing from their tails. Such tails were not "designed" in the strict sense of the word. Rather, they are adaptations. So runs the Darwinian story.


In Adaptationism and Optimality Steven Orzack and Elliott Sober want both to test and to flesh out this story. They want to test it because there is some dispute within the evolutionary biology community whether the traits we observe in biology actually are adaptations and thus principally the result of natural selection or whether some other factors might be involved. Further, they want to flesh out this story by providing an analytically precise account of what it means for a trait to be an adaptation.


On both counts their project is commendable. It is all too easy to attribute an organism's traits to natural selection. Since an organism with some trait is presumably surviving and reproducing, the trait must surely contribute to the organism's benefit. From there it's easy enough to spin a tale about how it does so. Alternatively, if the trait is plainly hampering the organism, one can argue that it is vestigial -- that in time past it did benefit the organism, but that now it no longer does, and that natural selection is in the process of removing the maladaptive trait (perhaps by driving the organism extinct).


Orzack and Sober rightly want nothing to do with such stories. They want analytic precision. Accordingly, they define adaptationism as the claim that "natural selection is the only important cause of the evolution of most nonmolecular traits and that these traits are locally optimal." This brings us to the second key term in their title: optimality. The optimality that Orzack and Sober have in mind is not the global optimality of engineering design but the local optimality of trial-and-error tinkering.


In engineering design, the goal is to attain some prespecified function or collection of functions as efficiently and elegantly as possible. As a consequence, the engineer must frequently re-engineer things from the ground up. Darwinian evolution doesn't work that way. Instead of re-designing from scratch, evolution must take pre-existing materials and tinker with them. The optimality in this case is local. Rather than asking what is the best system for achieving a function überhaupt (and thus raising the possibility of re-designing the system from scratch), Darwinian evolution asks how much one can improve a system's function merely by tinkering with it.


Now from Orzack and Sober's perspective the important thing about local optimality is that it admits quantification and thus enables one to assess the degree to which natural selection is responsible not just for a trait but for its optimal performance. In their essay for this collection (Adaptationism and Optimality is an anthology with twelve essays), Orzack and Sober compare the respective contributions of natural selection and phylogenetic inertia in accounting for the stability of traits. Phylogenetic inertia refers to the influence of an ancestor on a descendent. The basic idea here is, If it ain't broke, don't fix it. Phylogenetic inertia keeps traits as they are but doesn't make them better (except by accident). How can we tell whether a stable trait owes its stability to natural selection and thus is locally optimal? Alternatively, how can we tell whether the trait's stability is merely a matter of phylogenetic inertia maintaining the status quo? Orzack and Sober lay out some techniques from Bayesian decision theory to answer these questions.


Orzack and Sober's essay is typical of the collection. Other mechanisms compete with natural selection to account for the emergence, stability, and vanishing of traits; these include genetic drift, phylogenetic inertia, and developmental constraints. The point of these essays is to provide analytic tools for teasing apart the role of natural selection from these other mechanisms, apply these tools to actual case studies, and to reflect philosophically on the role of adaptationism in evolutionary biology. This aim is consistent with the editors' respective disciplines: Orzack is a biologist and Sober is a philosopher.


The essays vary in technical sophistication. Peter Godfrey-Smith's "Three Kinds of Adaptationism" is eminently readable. It  provides a sufficiently nice overview of the key philosophical issues connected with adaptationism that I would like to have seen it appear much earlier in the collection (it is essay eleven out of twelve). I recommend reading it first. By contrast, Ilan Eshel and Marcus Feldman's "Optimality and Evolutionary Stability under Short-term and Long-Term Selection" is going to be tough sledding for those without the requisite technical background in mathematical game theory and evolutionary biology. This essay attempts the important task of mathematically teasing apart the effects of natural selection in cases where a population's genes are merely reshuffled (short-term evolution) from those where a population's genes are transformed, as via mutations (long-term evolution). Without an analytically tractable distinction here, one is at best justified maintaining that natural selection is a conservative force in biological evolution, not the creative force that Darwinists claim.


Have Eshel and Feldman in fact provided an analytically tractable distinction for the two types of evolution? The failing of this essay is typical of those in the collection. Tools are developed (in some cases quite sophisticated tools) to assess some aspect of the adaptationist claim that natural selection is the primary mover in evolution. But the tools give at best partial results. And where they give results, they apply to picayune cases.


Consider the sorts of cases to which the tools are applied: sex ratios in wasp populations, fur length as a function of environmental temperature, and flower coloration in response to bee-pollination. These cases are entirely typical of the book, and none of those treated is any grander. To be sure, there is nothing wrong in applying analytic techniques to assess the role of natural selection in such modest cases. And the essays in this collection do make some genuine progress here. But sex ratios, fur length, and flower coloration are hardly what makes biological evolution the interesting theory that it is. Biological evolution is interesting because it purports to tell us how we got sex, fur, and flowers in the first place.


Granted, the aim of the essays in Adaptationism and Optimality is more modest -- it is to try to lay out some methodologies for accurately assessing the role of natural selection in tractable  contexts. My concern is not that the volume fails to do what it aspires to do. Nor is it that the volume aspires to do too little. Rather, my concern is what an outsider should make of this work in the broader cultural discussion over biological evolution.


It seems to me that this volume should raise two worries. First, if high-powered mathematical methods are required to assess the role of natural selection in highly constrained cases like sex ratios, fur length, and flower coloration, what hope have we that these methods can be extended to the really interesting questions of how sex, fur, and flowers themselves evolved? I'm frankly pessimistic, and this volume provides no grounds for overturning that pessimism. Orzack and Sober's call for analytic precision to study adaptationism is much needed, but if the analytic methods described in their volume cannot be extended to large scale evolutionary changes, then their project will be of very limited interest.


The other worry will be of considerably greater concern to Darwin skeptics (who, if we believe George Gallup, constitute a majority of the United States population). The point of Adaptationism and Optimality is to tease apart the effects of the Darwinian mechanism of natural selection from other material mechanisms (like genetic drift, phylogenetic inertia, and developmental constraints). Embedded in this project, however, is a presupposition, namely, that evolution is the result of material mechanisms.


The problem with material mechanisms, however, is just that: they are mindless material mechanisms that do what they do irrespective of intelligence. To be sure, mechanisms can be programmed by an intelligence. But any such intelligent programming of evolutionary mechanisms is not properly part of evolutionary biology as the contributors to Orzack and Sober's volume conceive it. But what if biology's adaptations are not exclusively the result of material mechanisms but also require intelligence, where the intelligence in question is not reducible to such mechanisms?


It's this possibility -- the possibility of intelligent design -- that Orzack and Sober's volume completely sidesteps. To be sure, it's a possibility that intelligence may not be a basic creative force in biology. Darwin after all may have been right, and natural selection may be all he cracked it up to be. But whether intelligence is dispensable from evolutionary biology needs to be established and not merely presupposed. The methods described in Orzack and Sober's volume do nothing to tease apart the effects of real intelligence from natural selection. This in my view is the book's greatest deficiency.