Saturday, July 11, 2015

Review of Jerry Coyne's Why Evolution is True

Jerry Coyne, biology professor at University of Chicago, is a vocal and popular advocate of neo-Darwinian evolution (what he himself often refers to as "Darwinism").  He is likewise severely critical of Darwinism's theoretical rival, intelligent design.  His 2009 book Why Evolution is True sets out to make a case for Darwinism and against intelligent design.  Coyne puts his thesis this way:
"This book lays out the main lines of evidence for evolution. For those who oppose Darwinism purely as a matter of faith, no amount of evidence will do--theirs is a belief not based on reason. But for the many who find themselves uncertain, or who accept evolution but are not sure how to argue their case, this volume gives a succinct summary of why modern science recognizes evolution as true" (p.xiv).
Coyne indeed accomplishes his goal of laying out lines of evidence in favor of evolution in his book.  In the first chapter, he lists six predictions of Darwinism.  The first three of these predictions are related to the fossil record and describe the features we should expect to see if organisms slowly changed over time from simpler forms to more complex forms.  For example, Coyne says we should find simpler organismal forms in older layers of rock; we should find cases of speciation in the fossil record; and we should find links (or transitional forms) between groups that diverged from common ancestors.  Unrelated to the fossil record, Coyne says we should observe a wide range of genetic variation in organisms (i.e. random mutations); we should find examples of "imperfect" biological systems, because evolution is blind or unguided; and we should see natural selection occurring in the present-day in the wild.

Coyne goes on in the subsequent chapters to discuss evidence in support of these predictions.  I found most of this discussion relatively uncontroversial, as I think there clearly is evidence in favor of the Darwinian thesis.

What I found more controversial was the inference Coyne wants his readers to make from this evidence.  As the title of his book suggests, Coyne wants to come down hard on the claim that Darwinian evolution is true.  He writes, "[E]volution is far more than a 'theory'....Evolution is a fact" (p.xii); he also claims evolution is an "indisputable fact" (p.209).

Although Coyne wants to come down hard on the certainty of Darwinism, he does slip in a few comments that weaken this certainty a bit:
"As we'll see, it is possible that despite thousands of observations that support Darwinism, new data might show it to be wrong. I think this is unlikely, but scientists, unlike zealots, can't afford to become arrogant about what they accept as true" (p.16).
"Now, when we say that 'evolution is true,' what we mean is that major tenets of Darwinism have been verified" (p.223).
These two quotes appear to me to more accurately capture the strength of the Darwinian thesis than the title of Coyne's book.  The main message of the book is that there is evidence in favor of evolution by random mutation and natural selection.  The further inference that the Darwinian thesis is true for all forms of life--that is, that Darwinian evolution sufficiently explains the origination of every life form on Earth--seems to me much less certain than Coyne often wants to claim. I say this for three reasons.

First, Coyne falls short on discussing a strong objection to the Darwinian thesis--namely, the Cambrian explosion.  Near the end of the book, Coyne says, "Evolutionary biology is teeming with questions and controversies" (p.223).  He then goes on to ask, "What caused the Cambrian explosion of life, in which many new types of animals appeared within only a few million years?" (ibid).  From here, Coyne does not attempt to answer this question; instead, he gives this response:
"Critics of evolution seize upon these controversies, arguing that they show that something is wrong with the theory of evolution itself.  But this is specious.  There is no dissent among serious biologists about the major claims of evolutionary theory--only about the details of how evolution occurred, and about the relative roles of various evolutionary mechanisms.  Far from discrediting evolution, the 'controversies' are in fact the sign of a vibrant, thriving field.  What moves science forward is ignorance, debate, and the testing of alternative theories with observations and experiments.  A science without controversy is a science without progress" (p.223).
Coyne's response here is pretty underwhelming.  He provides no specific explanations about how Darwinism is compatible with the rapid production of new life forms in the Cambrian period.  He does not even provide any references for interested readers to pursue.  In fact, Coyne gives the Cambrian such little discussion that it is not even listed in the index of his book!  Instead, Coyne reassures his readers that no serious scientist doubts evolution, then throws out a few trite statements about how controversy is healthy in science.  This would be fine if the title of his book were Why Controversy is Good for Science; but, his book is entitled Why Evolution is True, so he should be defending evolution against serious objections rather than falling back on scientific platitudes.

Furthermore, there is nothing "specious" about a critic arguing that a counter-example to a theory indicates a problem with the theory.  Argument by counter-example is utterly reasonable, normal, and effective.  I don't understand why Coyne would say such arguments are specious.  The proper response to a critic's counter-example is an explanation about how the theory can accommodate the apparent counter-example.  For example, critics of evolution present the Cambrian explosion of life as a counter-example to the Darwinian proposal of gradual, slight, successive changes over time.  The Cambrian explosion is not an event we would expect under the Darwinian thesis.  Therefore, the Darwinian must explain how the Cambrian event is ultimately compatible with the gradual, slight, successive modifications over time essential to the Darwinian thesis.  Coyne simply does not even attempt to do this.  His reply that "there is no dissent among serious biologists about the major claims of evolutionary theory" is a red herring--it changes the subject from the criticisms over specific Darwinian claims to the confidence some biologists have that Darwinism is correct.  It makes no difference to the Cambrian argument whether many scientists believe evolution is true, just as it made no difference historically how many learned people believed the sun revolved around the earth--what matters are the arguments, not the people who accept or reject them.  Thus, I claim Coyne fails to adequately address a strong objection to the Darwinian thesis, and the reader is left with insufficient reason to conclude with Coyne that Darwinian evolution is (in a strong sense) true.  (Note: for those interested in the significance of the Cambrian argument against Darwinian evolution, see Steven Meyer's 2013 book Darwin's Doubt and/or my review of it.)

Second, Coyne fails to discuss the particulars of random mutations at any length.  Specifically, he does not analyze the productive powers of random mutations in a quantitative way; that is, he does not discuss how probable and therefore numerous beneficial random mutations are, so the reader is not given much material with which to evaluate the plausibility of random mutations producing all of the genetic novelties observed in the history of life.

Let me record that Coyne accepts the familiar Darwinian picture of random mutations and natural selection:
"Evolution by a combination of randomness and lawfulness. There is first a 'random' (or 'indifferent') process--the occurrence of mutations that generate an array of genetic variants...and then a 'lawful' process--natural selection--that orders this variation, keeping the good and winnowing the bad..." (p.118). 
We see here that Coyne makes random mutations one-half of the evolution equation: first random mutations create novel biological structures, then natural selection deletes the bad, preserving the good.  Without the random mutation mechanism, natural selection would not have any new structures to winnow, and evolution would not happen.  Thus, random mutations are a vital piece of the Darwinian picture.  In order for Darwinian evolution to work, not only must genetic mutations be possible, but they must also happen often enough to produce new variations in life forms within restricted periods of time.

For example, Coyne discusses a transition from a land-dwelling, wolf-sized creature called Pakicetus to a fully aquatic, whale-like creature called Dorudon.  This sequence is supposed to be the path that evolved whales from land animals, and Coyne says all of the modifications took place within a window of 10 million years.  Coyne also states that this transition "did not require the evolution of any brand-new features--only modifications of old ones" (p.51).  Coyne further includes an illustration of this transition involving six creatures representing the step-wise modifications from land-dwelling features (like legs) to water-dwelling features (like fins).

I will record that the intermediate life-forms Coyne discusses in this transition do lend some credibility to the Darwinain view--namely, they are consistent with gradual, successive modifications over time.  However, my criticism is that he completely omits any quantitative discussion of the number and probability of the necessary mutations necessary for this transition to occur within 10 million years.  For example, how many engineering modifications are necessary to transform a land-dwelling, wolf-like creature into a fully aquatic, whale-like creature?  On the face of it, this would require quite a lot of changes, even if, as Coyne says, no new features were created.  If we could estimate how many engineering-like changes were required to turn a Pakicetus into a Dorudon, we could then estimate how many random mutations were necessary.  We could assume for the sake of argument that each successive modification required at least one genetic mutation, which would then tell us how many mutations must have occurred in that 10 million-year window for this transition to fit within the Darwinian model.  Furthermore, we could also estimate the prevalence of offspring for each of the transitional creatures.  This would tell us how many "trials" were possible within the 10 million-year period, since each offspring counts as one roll of the genetic "dice," so to speak.  Finally, we could estimate the probability that a single offspring (i.e. a single roll of the genetic dice) could produce a beneficial mutation.  Putting all of these pieces together, we could estimate the probability that a Pakicetus could evolve into a Dorudon through Darwinian mutation and selection.

Now, it is not at all obvious to me that these numbers would come out favorably to Darwinism.  In fact, I expect that the numbers would be unfavorable because random mutations are very rarely beneficial.  Coyne himself seems to agree that beneficial mutations are rare:
"Mutations are simply errors in DNA replication.  Most of them are harmful or neutral, but a few can turn out to be useful.  The useful ones are the raw material for evolution" (p.118).
Coyne points out that most are "harmful" or "neutral," but only a few are "useful."  He doesn't quantify this any further, so it's not clear whether "few" means 1 percent, 5 percent, 25 percent, etc.  He does not provide any footnotes either, so interested readers are left in the dark about how to evaluate the prevalence of beneficial mutations.  Likewise, readers are not given footnotes with references to learn more about the Pakicetus to Dorudon transition, so the book does not help clarify how many engineering changes were required, nor how large the breeding population sizes were for the animals in the transition.

Coyne's discussion of other transitions in the book likewise omits commentary and references pertaining to the quantification of random mutations.  Thus, I claim that Coyne's book suffers from a lack of thoroughness/completeness in substantiating the first half of the Darwinian equation--that of beneficial random mutations.

My third and final criticism of Coyne's book is that his objections to intelligent design (ID)--Darwinism's conceptual rival--are not very strong.   Let me begin with this excerpt from Coyne:
"How do we know that creationists are wrong when they say that selection can make small changes in organisms but is powerless to make big ones?  But first we must ask: What's the alternative theory?  We know of no other natural process that can build a complex adaptation.  The most commonly suggested alternative takes us into the realm of the supernatural.  This, of course, is creationism, known in its latest incarnation as 'intelligent design'"(p.136).
Here, Coyne seems to accept that Darwinian evolution and intelligent design are in competition with each other.  At the very least, he does not offer any alternatives to Darwinian evolution other than intelligent design.  Judging by Coyne's discussion of intelligent design, I take his analysis to ultimately collapse into a competition between naturalism and supernaturalism.   He of course classifies Darwinism as a naturalist position and intelligent design as a supernaturalist one.  Coyne argues that the supernatural nature of intelligent design renders the view untestable and unscientific; by contrast, Darwinism's naturalist commitments render it testable and therefore scientific. Coyne, predictably enough, holds that only theories that are testable and scientific deserve our rational consideration.  Thus, Coyne argues that intelligent design is essentially a non-option.  We could rephrase the same conclusion by saying that, for Coyne, the only possible theories that could explain the origin and diversification of life are naturalist theories.

Since supernaturalist theories like intelligent design are automatically off the table, and since Coyne admits that "we know of no other natural process that can build a complex adaptation," Darwinian evolution is effectively the only possible option.  For Coyne, then, Darwinism basically wins by "default."

Now, I should mention that Coyne attempts to maintain that Darwinian evolution is still subject to falsification:
"But we can still ask if there are adaptations that could not have been built by selection, and therefore require us to think of another mechanism" (p.136, Coyne's italics).
Since Coyne offers no other possible mechanism, his claim that we may need to think of a new mechanism strikes me as superficial--I think he wants to uphold the virtue of falsifiability for Darwinism by recording the possibility of another mechanism, but I don't think there is any conceptual space available for anything other than the mutation-selection mechanism.  At the very least, we know that, whatever new mechanism one might invent, Coyne would require that the mechanism be naturalistic if it would hope to compete with Darwinism.  Inherent in any naturalistic view is that explanations depend exclusively on chance and physical law; naturalistic views prohibit any kind of purposeful, intentional, or teleological forces.  To the extent that Coyne could permit an alternative mechanism to Darwinian evolution, that mechanism would be quite similar to the Darwinian view in that it would likewise depend on some combination of chance (like random mutations) and physical law (like natural selection).  This observation is why I say Coyne's appeal to a non-Darwinian alternative is merely superficial--I do not think Coyne's view substantively allows for a serious competitor to Darwinism.

Let me return to my claim that Coyne's preference for Darwinism over intelligent design is based on a more fundamental preference for naturalism over supernaturalism.  Suppose we agree with Coyne that Darwinian evolution is naturalistic and intelligent design is supernaturalistic.  Suppose further we agree with Coyne that naturalistic propositions are testable and supernaturalistic ones are not.  If I understand Coyne correctly, the crucial distinction between these two views is that naturalism is testable/falsifiable and supernaturalism is not.  Coyne's position seems to be that the truth of a proposition depends on the proposition's testability.  It is not necessarily the case that a testable proposition is automatically true, but it does seem to be the case that an untestable proposition, on Coyne's view, is automatically false.  I have two comments in response to this.

First, Why should we suppose that the truth of a proposition depends on its testability?  While we can easily agree that testability is desirable and convenient for rational creatures like us, on what grounds are we justified in supposing the world "out there" is under any obligation to make itself testable to us?  It seems anthropocentric to assume the facts of the world only exist in ways that we can test them.  I think it is entirely possible that some of the facts of the world are not open to scientific testability/falsifiability; I at least cannot think of any good arguments against this possibility.  Coyne, certainly, does not offer any such arguments in this book.  Furthermore, the fact that a proposition might be untestable does not seem to render that proposition either arbitrary or incoherent.  For example, the proposition that the past exists is untestable; for all we know, the entire universe just popped into existence right now replete with the appearance of age.  There is no way to test the proposition that the past exists, for any kind of test must presuppose the existence of the past in the first place, which would make it circular.  Despite this, the proposition that the past exists is not arbitrary or incoherent.  We do have positive reasons for thinking there is a past--namely, our experience of moving through time.  Our belief in the reality of the past is consistent with our experiences of the world.  Thus, even though the reality of the past is a scientifically untestable proposition, it is not automatically false.  I argue the situation with intelligent design is in a similar position (at least if we agree with Coyne that intelligent design is untestable).  If the supernatural flavor of intelligent design renders the view scientifically untestable, it should not follow that intelligent design is false.  We can still ask if we have positive reasons for supposing intelligent design is true, just like we might have positive reasons for supposing the past is real.  Although I do think we have positive reasons for supposing intelligent design is true, my purpose here is not to get into them.  I merely wish to establish that Coyne's grounds for rejecting intelligent design are problematic: intelligent design should not be discarded on the basis of non-testability.

Second, insofar as naturalism is in competition with supernaturalism, naturalism itself is ironically untestable.  Naturalism and supernaturalism are both philosophical world views, not scientific discoveries.  There is no way to scientifically test whether naturalism is true.  Suppose I define naturalism as the proposition "all real entities are physical entities."  This definition is logically identical to the claim "no non-physical entities exist."  I claim there is no way to test either definition of naturalism.  In order to test whether non-physical entities exist, a scientist would need some kind of device that is capable of measuring the existence of non-physical entities.  If non-physical entities are detected, naturalism would be falsified; if no non-physical entities were detected, supernaturalism would be falsified.  But, what kind of device would this "non-physical detector" be?  How would a scientist even begin to build such a thing?  I submit that no such device is possible.  As a result, both naturalism and supernaturalism are scientifically untestable.  These two views, however, are still philosophically evaluable.  This is to say, one can make philosophical arguments for or against these two views and infer that one is more rationally supported than the other, even though neither is empirically falsifiable.  My main message here is that, to the extent Coyne relies on testability to support Darwinism over intelligent design (which I claim is a large extent), this reliance on testability breaks down for him at the level of naturalism versus supernaturalism.  His preference for naturalism cannot endure his own standard of empirical testability.  Thus, it is arbitrary for him to discard intelligent design as untestable due to its supernatural commitments, for Darwinism is likewise untestable due to its naturalistic commitments.  Testability, therefore, cannot be a necessary standard for theory fitness.

If Coyne's standard for empirical testability ends up being flawed, as I argue it is, it stands to reason then that intelligent design (assuming it is in fact empirically untestable) is a realistic alternative to Darwinian evolution.  It does not follow from this that intelligent design is therefore true; but it does elevate intelligent design's plausibility in the debate over the origin and diversification of life.  In other words, Darwinian evolution would no longer win by default, but would have to stand or fall by virtue of the strengths and weaknesses of its arguments.  In order to succeed in this context, I think Darwinian advocates like Coyne would have to take intelligent design arguments far more seriously than, for example, Coyne does in this book.  Thus, I claim that Coyne's treatment of intelligent design in Why Evolution is True is underdeveloped.

In sum, I found Coyne's primary claim in his book--that evolution is true--insufficiently supported.  While he certainly did provide some lines of evidence in favor of Darwinian evolution, these lines of evidence are insufficient to establish the strong claim the book's title makes.  After reading several books by both Darwinian and intelligent design advocates alike, I have found that the competition between these two views is far more intense than Coyne acknowledges in the book.  His work would be far more compelling if he took intelligent design as a more serious challenge to evolution than he does, and if he would engage with the intelligent design arguments in greater detail. 


  1. This comment has been removed by the author.

  2. Ryan, I am now studying that book!

    Actually I have read your first and last paragraph, but I will save your review to look at it later.

    I think what books on evolution try to refute is not "intelligent design". True they claim so, but practically they do not, and cannot refute intelligent design. However, I think these books give us very good evidence to refute "Creationism". The view that all the species were created all at once in a young or old earth. I think there is good evidence to put these options aside.

    But what about intelligent design? In my idea, intelligent design should be discussed in philosophical circles, rather than scientific ones. I think a theist can happily concede that the whole process of evolution was guided by God.

    What about examples of bad design? Although they can be used as counterexamples for the argument from design [which is far from being a successful argument with or without these counterexamples], I think they do not work. In the very similar way that arguments from evil do not work. We are unaware of reasons and metaphysical limitations that God has for creation of the the organisms.

    We should put together different pieces of evidence from science, history, religious experiences, reports of miracles, etc., and then decide which theory is of more theoretical virtue.

  3. Thanks for the comment, Dara.

    I agree that books like Coyne's definitely set out to refute biblical creationism. As far as Intelligent Design goes, the authors do claim to refute it, but I agree that they rarely target it directly. Often the design argument is written off as a clandestine form of biblical creationism, so ID's unique philosophical structure is simply ignored. This is unfortunate because I think there is a really interesting argument to be had over Intelligent Design.

    If you're interested in the relationship between ID and science, I highly recommend Stephen Meyer's books, Signature in the Cell and Darwin's Doubt. Meyer provides a very reasonable discussion about the philosophy of science and how hard it has historically been to demarcate the boundaries of what's "scientific." He rightly points out, I think, that there's no way to avoid arbitrariness when setting the boundaries around what is and is not science. I happen to agree with Meyer that ID is an appropriate subject in scientific circles, but, like Meyer, I also think whether it is recognized in those circles or not is secondary to whether or not the view is true.

    I agree with you as well that the scientific evidence is not by itself sufficient to judge whether design is a good argument. I am deeply critical of methodological naturalism and "scientism." Those positions seem to me so obviously self-defeating.