Evolution: The Pleasures of Pluralism
STEPHEN JAY GOULD

New York Review of Books, June 26, 1997  

¶1 Charles Darwin began the last paragraph of The Origin of Species (1859) with a famous metaphor about life's diversity and ecological complexity:

It is interesting to contemplate an entangled bank, clothed with many plants of many kinds, with birds singing on the bushes, with various insects flitting about, and with worms crawling through the damp earth, and to reflect that these elaborately constructed forms, so different from each other, and dependent on each other in so complex a manner, have all been produced by laws acting around us.

He then begins the final sentence of the book with an equally famous statement: "There is grandeur in this view of life...."

¶2 For Darwin, as for any scientist, a kind of ultimate satisfaction (Darwin's "grandeur") must reside in the prospect that so much variety and complexity might be generated from natural regularities--the "laws acting around us"--accessible to our intellect and empirical probing. But what is the proper relationship between underlying laws and explicit results? The "fundamentalists" among evolutionary theorists revel in the belief that one overarching law--Darwin's central principle of natural selection--can render the full complexity of outcomes (by working in conjunction with auxiliary principles, like sexual reproduction, that enhance its rate and power).

¶3 The "pluralists," on the other hand--a long line of thinkers including Darwin himself, however ironic this may seem since the fundamentalists use the cloak of his name for their distortion of his position--accept natural selection as a paramount principle (truly primus inter pares), but then argue that a set of additional laws, as well as a large role for history's unpredictable contingencies, must also be invoked to explain the basic patterns and regularities of the evolutionary pathways of life. Both sides locate the "grandeur" of "this view of life" in the explanation of complex and particular outcomes by general principles, but ultra-Darwinian fundamentalists pursue one true way, while pluralists seek to identify a set of interacting explanatory modes, all fully intelligible, although not reducible to a single grand principle like natural selection.

¶4 The first part of this article outlined the general fallacies of ultra-Darwinian fundamentalism, especially in the light of new theories and discoveries in the core disciplines of developmental biology, paleontology, and population genetics.¹ In this second and concluding part, I shall analyze a prominent philosopher's influential but misguided ultra-Darwinian manifesto--Darwin's Dangerous Idea, by Daniel Dennett. I shall also take up the methodology of so-called "evolutionary psychology"--a field now in vogue as a marketplace for ultra-Darwinian explanatory doctrine. Evolutionary psychology could, in my view, become a fruitful science by replacing its current penchant for narrow, and often barren, speculation with respect for the pluralistic range of available alternatives that are just as evolutionary in status, more probable in actual occurrence, and not limited to the blinkered view that evolutionary explanations must identify adaptations produced by natural selection.


1.

¶5 Daniel Dennett devotes the longest chapter in Darwin's Dangerous Idea to an excoriating caricature of my ideas, all in order to bolster his defense of Darwinian fundamentalism. If an argued case can be discerned at all amid the slurs and sneers, it would have to be described as an effort to claim that I have, thanks to some literary skill, tried to raise a few piddling, insignificant, and basically conventional ideas to "revolutionary" status, challenging what he takes to be the true Darwinian scripture. Dennett claims that I have promulgated three "false alarms" as supposed revolutions against the version of Darwinism that he and his fellow defenders of evolutionary orthodoxy continue to espouse.

¶6 Dennett first attacks my view that punctuated equilibrium is the dominant pattern of evolutionary change in the history of living organisms. This theory, formulated by Niles Eldredge and me in 1972, proposes that the two most general observations made by palentologists form a genuine and primary pattern of evolution, and do not arise as artifacts of an imperfect fossil record. The first observation notes that most new species originate in a geological "moment." The second holds that species generally do not change in any substantial or directional way during their geological lifetimes--usually a long period averaging five to ten million years for fossil invertebrate species. Punctuated equilibrium does not challenge accepted genetic ideas about the rates at which species emerge (for the geological "moment" of a single rock layer may represent many thousand years of accumulation). But the theory does contravene conventional Darwinian expectations for gradual change over geological periods, and does suggest a substantial revision of standard views about the causes of long-term evolutionary trends. For such trends must now be explained by the higher rates at which some species branch off from others, and the greater durations of some stable species as distinguished from others, and not as the slow and continuous transformation of single populations.

¶7 In his second attack, Dennett denigrates the importance of nonadaptive side consequences ("spandrels" in my terminology) as sources for later and fruitful reuse. In principle, spandrels define the major category of important evolutionary features that do not arise as adaptations. Since organisms are complex and highly integrated entities, any adaptive change must automatically "throw off" a series of structural byproducts--like the mold marks on an old bottle or, in the case of an architectural spandrel itself, the triangular space "left over" between a rounded arch and the rectangular frame of wall and ceiling. Such byproducts may later be co-opted for useful purposes, but they didn't arise as adaptations. Reading and writing are now highly adaptive for humans, but the mental machinery for these crucial capacities must have originated as spandrels that were co-opted later, for the brain reached its current size and conformation tens of thousands of years before any human invented reading or writing.

¶8 Third, and finally, Dennett denies theoretical importance to the roles of contingency and chance in the history of life, a history that has few predictable particulars and no inherent directionality, especially given the persistence of bacteria as the most common and dominant form of life on Earth ever since their origin as the first fossilized creatures some 3.5 billion years ago.² Bacteria are biochemically more diverse, and live in a wider range of environments (including near-boiling waters, and pore spaces in rocks up to two miles beneath the earth's surface), than all other living things combined. The number of E. coli cells in the gut of each human being exceeds the total number of human beings that have ever lived. Moreover, if recent reports of Martian fossil bacteria are true, then bacterial domination may be interplanetary or universal, and not merely earthly.

¶9 These three concepts work as pluralistic correctives to both the poverty and limited explanatory power of the ultra-Darwinian research program. Punctuated equilibrium requires that substantial evolutionary trends over geological time, the primary phenomenon of macroevolution, be explained by the greater long-term success of some species versus others within a group of species descended from a common ancestor. Such trends cannot be explained, as Darwinian fundamentalists would prefer, as the adaptive success of individual organisms in conventional competition, extrapolated through geological time as the slow and steady transformation of populations by natural selection. The principle of spandrels, discussed at greater length later in this article, stresses the role that nonadaptive side consequences play in structuring the directions and potentials of future evolutionary change. Taken together, punctuated equilibrium and spandrels invoke the operation of several important principles in addition (and sometimes even opposed) to conventional natural selection working in the engineering mode that Dennett sees as the only valid mechanism of evolution.

¶10 My third pluralistic corrective to traditional theory does not invoke other principles in addition to natural selection, but rather stresses the limits faced by any set of general principles in our quest to explain the actual patterns of life's history. Crank your algorithm of natural selection to your heart's content, and you cannot grind out the contingent patterns built during the earth's geological history. You will get predictable pieces here and there (convergent evolution of wings in flying creatures), but you will also encounter too much randomness from a plethora of sources, too many additional principles from within biological theory, and too many unpredictable impacts from environmental histories beyond biology (including those occasional meteors)--all showing that the theory of natural selection must work in concert with several other principles of change to explain the observed pattern of evolution.

Some say this misses Gould's point: "What is special about the spectacular diversity of the Burgess Shale fauna is that these weren't just new species, but whole new phyla! These were radically novel designs!" I trust this was never Gould's point, because if it was, it was an embarrassing fallacy of retrospective coronation.

It would be--but since I have never even hinted at such a silly view, why bother to point out how stupid I would be if I ever had?

¶12 In an even more unfair example, Dennett conjectures about what I might believe (but I don't, and he cites nothing to support his supposition), and then seems to pretend that I hold such a view by attacking someone else who truly does:

Is it likely that Gould could be so confused about the nature of algorithms? As we shall see in chapter 15, Roger Penrose, one of the world's most distinguished mathematicians, wrote a major book (1989) on Turing machines, algorithms, and the impossibility of Artificial Intelligence, and his whole book is based on that confusion. This is not really such an implausible error, on either thinker's part [I have now become an explicit supporter of the idea--]. A person who really doesn't like Darwin's dangerous idea often finds it hard to get the idea in focus.

2. False characterization. Dennett depicts me as constantly and explicitly claiming to invent one scientific "revolution" after another. No characterization appears more frequently throughout the chapter, and none could be so false. "Gould," he states, "has gone from revolution to revolution. So far, his declarations of revolution have all been false alarms." We then learn that "the spandrel revolution (against panadaptationism) and the exaptation revolution (against preadaptationism) evaporate on closer inspection," and that "Gould's attempted revolution against gradualism was actually his first." In a final summary, we learn that "Gould's self-styled revolutions ...all evaporate."

¶13 Of course, I am never quoted directly as making any personal claim about a "revolution," self-styled or otherwise--and for a good reason: I have, quite consciously and on principle, never used the word to describe my work. I have too much admiration for my dear friend, the late Thomas Kuhn, and I have also too often watched foolish and ambitious colleagues trying to wrap themselves in a caricature of his "paradigm" in order to proclaim their own revolutions, ever to engage in such a farce myself.

¶14 Dennett says that I started all this revolution-mongering in my first paper on punctuated equilibrium, published in 1972. In that work, by contrast, I paid homage to Kuhn's influence, and explicitly disavowed any such fatuous intention: "We have no desire," I wrote (with Niles Eldredge) in 1972,

to enter the tedious debate over what is, or is not,...a paradigm [in Kuhn's terminology]. In using the neutral word picture, we trust that readers will understand our concern with alternate ways of seeing the world that render the same facts in different ways.

The details of Dennett's discussions also rest largely on ridicule and mischaracterization. For example, he denigrates the principle of spandrels by mockery, not argument.

Gould wants to convince us that adaptation is not "pervasive," so he needs to have a term for the (presumably many) biological features that are not adaptations. They are to be called "spandrels." Spandrels are, um, things that aren't adaptations, whatever they are.

But I have always defined spandrels precisely as one particular and eminently testable kind of nonadaptive structure--an architectural byproduct, or what Darwin called, in his own favorite and pluralistic example of nonadaptation, a "correlation of growth."

Speciation is not always an extension of gradual, adaptive allelic substitution to greater effect, but may represent, as Goldschmidt argued, a different style of genetic change--rapid reorganization of the genome, perhaps nonadaptive.

This statement still strikes me as entirely reasonable, and I see no reason to modify it today. I said that some instances of speciation--probably only a small percentage--proceed in this rapid mode. "Not always" surely means "not all the time, but usually." It cannot mean "never"--as Dennett charges in claiming that I attribute "any" speciation event to the opposite mode of saltation. In other words, I am trying to carve out a modest space for an unconventional mechanism at low frequency--and Dennett charges me with advocating a revolution based on universal occurrence for this unorthodox mode.

3. High density of error. A fair test can be made for a nonprofessional's grasp of scientific material by noting the frequency of factual errors in his descriptions of technical work. I do not claim that any of these minor mistakes produces great distortion, but Dennett's high density of errors, on easy points that only require accurate reading or copying, indicates an apparent indifference to the vital details that build the history of life.

¶18 Dennett's account of my book Wonderful Life includes the following errors in only four pages. He misstates the date of the Cambrian explosion by 70 million years--"a time around six hundred million years ago when the multicellular organisms really took off." The actual date is 530 million years ago. He then states (a serious error this time) that C.D. Walcott based his original Burgess Shale work on "literal dissection of some of the fossils"--when I emphasize in my book (as a major theme of my narrative) that Walcott failed precisely because he did not dissect the specimens, which he incorrectly interpreted as squashed absolutely flat. Dennett then says twice that most of the Burgess Shale species perished as rapidly as they arose--"most of them vanished just as suddenly"--when I clearly state that we know nothing at all about the manner of their dying, because they left no fossil record after the Burgess beds. Finally, Dennett lists eight of the wonderful Burgess creatures in a single sentence--and he spells three of their names wrong. I don't wish to harp on trivialities but Dennett could legitimately accuse me of disrespectful inattention if I listed Denet, Dawkuns, and Maynid Smith as the apostles of ultra-Darwinism.

4. Gratuitous speculation about motives. In Dennett's last few pages he apparently feels he must figure out why I could be so obtuse (since he doesn't regard me as stupid). "It might seem disingenuous," he writes, "for me not even to mention the obvious 'rival' explanations crying to be considered: politics and religion." Dennett has no clue about my political or religious views, and he has never bothered to ask me--but did lack of data ever derail the ultra-Darwinian game of adaptationist storytelling? So we first hear a little red-baiting. Then, in a surprise turnaround, I become a closet theist who secretly hates Darwinism because "Humanity cannot be special enough to matter if it is the product of merely algorithmic processes." As evidence, Dennett writes:

Gould often quotes the Bible in his monthly columns, and sometimes the rhetorical effect is striking. Surely, one thinks, an article with this opening sentence has to have been written by a religious man: "Just as the Lord holds the whole world in his hands, how we long to enfold an entire subject into a witty epigram."

I do often quote the Bible as great literature, but this time I was only citing a famous African-American folk song.


2.

¶19 The fallacy of Dennett's argument also undermines his other imperialist hope--that the universal acid of natural selection might reduce human cultural change to the Darwinian algorithm as well. Dennett, following Dawkins once again, tries to identify human thoughts and actions as "memes," thus viewing them as units that are subject to a form of selection analagous to natural selection of genes. Cultural change, working by memetic selection, then becomes as algorithmic as biological change operating by natural selection on genes--thus uniting the evolution of organisms and thoughts under a single ultra-Darwinian rubric:

According to Darwin's dangerous idea...not only all your children and your children's children, but all your brainchildren and your brainchildren's brainchildren must grow from the common stock of Design elements, genes and memes.... Life and all its glories are thus united under a single perspective.

But, as Dennett himself correctly and repeatedly emphasizes, the generality of an algorithm depends upon "substrate neutrality." That is, the various materials (substrates) subject to the mechanism (natural selection in this case) must all permit the mechanism to work in the same effective manner. If one kind of substrate tweaks the mechanism to operate differently (or, even worse, not to work at all), then the algorithm fails. To choose a somewhat silly example that actually played an important role in recent American foreign policy, the cold war "domino theory" held that communism must be stopped everywhere because if one country turned red, then others would do so as well, for countries are like dominos standing on their ends and placed one behind the other--so that the toppling of one must propagate down the entire line to topple all. Now if you devised a general formula (an algorithm) to describe the necessary propagation of such toppling, and wanted to cite the algorithm as a general rule for all systems made of a series of separate objects, then the generality of your algorithm would depend upon substrate neutrality--that is, upon the algorithm's common working, regardless of substrate (similarly for dominos and nations in this case). The domino theory failed because differences in substrate affect the outcome, and such differences can even derail the operation of the algorithm. Dominoes must topple, but the second nation in a line might brace itself, stay upright upon impact, and therefore fail to propagate the collapse.

¶20 Natural selection does not enjoy this necessary substrate neutrality. As the great evolutionist R.A. Fisher showed many years ago in the founding document of modern Darwinism (The Genetical Theory of Natural Selection, l930), natural selection requires Mendelian inheritance to be effective. Genetic evolution works upon such a substrate and can therefore be Darwinian. Cultural (or memetic) change manifestly operates on the radically different substrate of Lamarckian inheritance, or the passage of acquired characters to subsequent generations. Whatever we invent in our lifetimes, we can pass on to our children by our writing and teaching. Evolutionists have long understood that Darwinism cannot operate effectively in systems of Lamarckian inheritance--for Lamarckian change has such a clear direction, and permits evolution to proceed so rapidly, that the much slower process of natural selection shrinks to insignificance before the Lamarckian juggernaut.

¶28 Again, I applaud this development. If this principle were advanced in conjunction with the recognition that a putative evolutionary origin does not necessarily imply an adaptive value at all, then evolutionary psychology could make a substantial advance in applying Darwinian theory to human behavior. But the advocates of evolutionary psychology proceed in the opposite direction by twisting the observation that the behavior of modern humans may not necessarily have adaptive value into an even more dogmatic, and even less scientifically testable, panadaptationist claim. Evolutionary universals may not be adaptive now, they say, but such behaviors must have arisen as adaptations in the different ancestral environment of life as small bands of hunter-gatherers on the African savannas--for evolutionary theory "means" a search for adaptive origins.

¶29 The task of evolutionary psychology then turns into a speculative search for reasons why a behavior that may harm us now must once have originated for adaptive purposes. To take an illustration proposed seriously by Robert Wright in The Moral Animal, a sweet tooth leads to unhealthy obesity today but must have arisen as an adaptation. Wright therefore states:

The classic example of an adaptation that has outlived its logic is the sweet tooth. Our fondness for sweetness was designed for an environment in which fruit existed but candy didn't.

This ranks as pure guesswork in the cocktail party mode; Wright presents no neurological evidence of a brain module for sweetness, and no paleontological data about ancestral feeding. This "just-so story" therefore cannot stand as a "classic example of an adaptation" in any sense deserving the name of science.

The thousands and thousands of genes that influence human behavior--genes that build the brain and govern neurotransmitters and other hormones, thus defining our "mental organs" [note the modularity claim]--are here for a reason. And the reason is that they goaded our ancestors into getting their genes into the next generation [the claim for adaptation in the EEA]. If the theory of natural selection is correct, then essentially everything about the human mind should be intelligible in these terms [the ultra-Darwinian faith in adaptationism]. The basic ways we feel about each other, the basic kinds of things we think about each other and say to each other [note the claim for universality], are with us today by virtue of their past contribution to genetic fitness.

Wright's closing sermon is more suitable to a Sunday pulpit than a work of science:

The theory of natural selection is so elegant and powerful as to inspire a kind of faith in it--not blind faith, really.... But faith nonetheless; there is a point after which one no longer entertains the possibility of encountering some fact that would call the whole theory into question.

I must admit to having reached this point. Natural selection has now been shown to plausibly account for so much about life in general and the human mind in particular that I have little doubt that it can account for the rest.

¶33 This adaptationist premise is the fatal flaw of evolutionary psychology in its current form. The premise also seriously compromises--by turning a useful principle into a central dogma with asserted powers for nearly universal explanation--the most promising theory of evolutionary psychology: the recognition that differing Darwinian requirements for males and females imply distinct adaptive behaviors centered upon male advantage in spreading sperm as widely as possible (since a male need invest no energy in reproduction beyond a single ejaculation) and female strategies for extracting additional time and attention from males (in the form of parental care or supply of provisions, etc.). (In most sexually reproducing species, males generate large numbers of "cheap" sperm, while females make relatively few, "energetically expensive" eggs, and then must invest much time and many resources in nurturing the next generation.)

¶34 This principle of differential "parental investment" makes Darwinian sense and probably does underlie some different, and broadly general, emotional propensities of human males and females. But contrary to claims in a recent deluge of magazine articles, parental investment will not explain the full panoply of supposed sexual differences so dear to pop psychology. For example, I do not believe that members of my gender are willing to rear babies only because clever females beguile us. A man may feel love for a baby because the infant looks so darling and dependent, and because a father sees a bit of himself in his progeny. This feeling need not arise as a specifically selected Darwinian adaptation for my reproductive success, or as the result of a female ruse, culturally imposed. Direct adaptation is only one mode of evolutionary origin. After all, I also have nipples not because I need them, but because women do, and all humans share the same basic pathways of embryological development.

¶35 If evolutionary psychologists continue to push the theory of parental investment as a central dogma, they will eventually suffer the fate of the Freudians, who also had some good insights but failed spectacularly, and with serious harm imposed upon millions of people (women, for example, who were labeled as "frigid" when they couldn't make an impossible physiological transition from clitoral to vaginal orgasm), because they elevated a limited guide into a rigid creed that became more of an untestable and unchangeable religion than a science.

¹ "Darwinian Fundamentalism," The New York Review, June 12, 1997, pp. 34-37. (back)

² See my recent book Full House (Crown, 1996) for an account of this. (back)

³ See such technical works as J. Barkow, L. Cosmides, and J. Tooby, The Adapted Mind: Evolutionary Psychology and the Generation of Culture (Oxford University Press, 1992); and D.M. Buss, The Evolution of Desire (Basic Books, 1994); and, especially, for its impact by good writing and egregiously simplistic argument, the popular book of R. Wright, The Moral Animal: Why We Are the Way We Are: The New Science of Evolutionary Psychology (Random House, 1994). (back)