Skip to content

Random thoughts about God and evolution

As some may know, I am writing a couple of book chapters to try to sell a proposal to a publisher for The Nature of Classification, a book I am coauthoring with Malte Ebach. I bring the philosophy and he brings the knowledge. However, this means I am not devoting much in the way of brain cells to extraneous blog posts right now. And I feel that I am letting you down. You need something to disagree with from me, or else you may as well just go read PZ.

So I have been thinking of the relationship of theism to evolution lately. As you know, I have a paper in long gestation arguing that theism and evolution – the whole schtick of Darwinian evolution by random mutation and natural selection and stochastic drift, etc., not some cut-down version in which God intervenes in an otherwise clockwork process from time to time – are mutually compatible. One day my coauthor may complete his half and we can submit it. But here are the issues as I see them:

There is the deist option: Deism has several meanings but the one that is most relevant here is that God set up the world with its laws and initial conditions, pushed the Big Red Button, and retired, permanently. It then mechanically evolved, astronomically and biologically, and here we are. The problem with this, is that this god is not the god of theism, who has a plan. This deity is not providential, but the deity of Epicurus and Lucretius:

For all the gods must of themselves enjoy

Immortal aeons and supreme repose,

Withdrawn from our affairs, detached, afar:

Immune from peril and immune from pain,

Themselves abounding in riches of their own,

Needing not us, they are not touched by wrath

They are not taken by service or by gift.

For the deist option to work, evolution must be predictable and inevitable, if we are to be the plan of the deity. This is the view put forward by Simon Conway Morris and those who concur with him; life has a “deep structure” that means that all evolving pathways must converge upon similar solutions, and one of those solutions is, well, us. So the deist option involves two constraints: either god is not providential or his providence consists in creating a world in which humans are inevitable, and if the latter, evolution is predictable and limited in its outcomes.

But many think that evolution is not predictable. In Gould’s famous metaphor, if you rerun the tape of life, each time you would get a different outcome. This is counterfactual thinking, of course – we cannot test this in the large, although we can do limited experiments. There is a problem with those, though; they tend to be very limited. If a selective breeding experiment consistently tricks on a single outcome, is that because the prior state of the initiating genome has limited options, or because there is only one way to solve that “problem”? There’s nothing very deep about finding that the weighted clown stands upright each time; it is hard to make it do anything else. This doesn’t imply that every object in the world will tend to do one and only one thing, let alone stand upright.

The motivation for a “periodic table of life”, to use George McGhee’s term from the book linked above, seems to be physicism. This is a label I just invented (or probably reinvented) to denote biologists who think biology has to be just like physics. The logic runs roughly: since biology is a science, and sciences have deep regularities that are unique to them, biology must have deep regularities that do not reduce to other sciences. Ergo there are deep regularities in evolution. Now it is very likely that there are deep regularities in biology: they are at the very least, regularities of physics. The reason why we do not have solutions involving compounds of neon is that neon doesn’t form compounds under biologically realistic conditions. And there are abstract regularities in biology: Mendel’s law and natural selection are two examples. Both of these involve abstracting away from the physical properties involved in any one case and giving a level of description and explanation that are independent of the substrate properties (see Goode and Griffiths 1995, for example).

But that there are some purely biological constraints that are neither abstract nor reducible to physics, a view sometimes called structuralism, is a hard position to maintain, and that is needed to make the deist view consistent with a purely Darwinian evolution. By “Darwinian”, of course, I mean the modern set of theories, not some slavish devotion to selectionism or Darwin’s own views, although I will argue that he is pretty acute on this topic. I shall call the position that Conway Morris and collaborators adopt the convergence option.

Now, is the deist option available for ordinary theists? Can a Christian be a Darwinian? By this I do not mean, can a Christian compartmentalise their beliefs from their science, nor can they trim either their science or their religion, but are the two positions even compatible? Can one be, to steal a phrasing, fully theist and fully scientist? Is homoousias possible? Are science and religion of one substance? Obviously some think they are mutually incompatible, while others think they are totally compatible. But what I am concerned with is this: can one rationally hold that the core teachings of theism are true, and yet not modify in any way the results and best hypotheses of science? The “rational” part is crucial here. There are all kinds of irrational ways, such as the God of the Gaps, or simply science denialism like creationists.

What is a “core” teaching of theism? We cannot simply specify that theism equates to a literalistic interpretation of the Bible, the Q’uran, or the writings of Ayn Rand. We need to appeal to some philosophical theology. Generally, I take theism to mean at least the following things:

1. There is a deity independent of the physical/created world.

2. That deity is a providential deity, with a plan that it implements towards an end.

3. That deity has the ability to do this, and has so done. That is, the deity has intervened in the physical world.

4. The outcomes are therefore, in some sense, designed.

Now, several things mitigate against theism from the scientific domain.

One is that if you have a purely physical explanation for an outcome X, then the explanatory role for god in producing X is no longer necessary. For example, we do not need to appeal to God’s actions to produce babies in explaining fetal development from a fertilised zygote. A purely physical account is all that we require. This is the Excluded God from the Gaps argument. Once a gap has been filled by knowledge, God is not needed.

Another is that if there are uncaused outcomes, either by randomness or by spontaneous generation (in the broader than biology sense), this, too, mitigates against a deity that is supposed to be competent and sufficient to explain outcomes. The First Cause argument, where God is needed to set things in motion (literally, in Aristotle’s version), becomes otiose. So if evolution involves random outcomes and causes, or if the universe is such that you cannot specify a determinate set of boundary conditions that will always cause a certain outcome even if you are god or Laplace’s Demon, as quantum mechanics currently suggests, then to accommodate the theist creator with science means that either unguided causation has to go or god has to be constrained by the laws of physics. Either way, this is not something a traditional philosophical theist would want to adopt.

Is the convergence option acceptable? Apart from prior metaphysical prejudices that it must be true, I fail to see how it can be supported by modern science. Biology is the science of the contingent, and the empty “spaces” in structure that are touted as evidence for it are a lot less convincing the more broadly one considers biology. Conway Morris, for instance, thinks that multicellularity is inevitable once life evolves, but basically this is because he thinks that life is roughly synonymous with cellularity, and once you have single units, of course multiple unit ensembles become a formal possibility. This needs no structuralist to tell us that. But is life necessarily cellular? Ours happens to be, but the necessity is unproven. Moreover, are eyes going to be inevitable? Well maybe if you get something like animals with nervous systems, but plants “sense” light fine without eyes, and so I see no necessity there, either. Animal-centrism is question-begging. So is vertebrate-centrism, tetrapod-centrism and so on. If you phylogenetically bracket the outcomes, then of course some outcomes are relatively likely, but that was what we were debating in the first place, whether the phylogenies are necessary. Conway Morris thinks phylogenetic thinking is wrong; I think he does it all the time and doesn’t even realise it.

So the deist option is out for a traditional theist, and the convergence option is at best unconvincing and at worst question-begging. No, actually at worst it is complete nonsense, but I think that some aspects of convergence are true, but only if you can bracket the options phylogenetically, which is to say, on the basis of contingent historical development.

Therefore, I reject both options as accommodations between theism and science. In the next of this series, I will suggest another way how a theist can rationally accommodate the two. It is not, I suspect, the only way to do so, but all we need for it to be a rational position is one option that works. However, it may not be something that the ordinary Christian or Muslim could appreciate or fully accept.


Conway Morris, S. 2003. Life’s solution: inevitable humans in a lonely universe. Cambridge: Cambridge University Press.

Conway Morris, S. 2008. The deep structure of biology: is convergence sufficiently ubiquitous to give a directional signal? West Conshohocken, PA; Edinburgh: Templeton Foundation; Alban.

Goode, R., and P. E. Griffiths. 1995. The misuse of Sober’s selection of/selection for distinction. Biology and Philosophy 10 (1):99-108.

Ruse, Michael. 2000. Can a Darwinian be a Christian? : the relationship between science and religion. New York: Cambridge University Press.


  1. The unpredictability of evolution is very overplayed. Evolution is more than one thing. Some of those things are inherently very predictable and others very unpredictable. For example, I predict that on a planet with life forms that follow Darwinian patterns (imperfect replication, over production, offspring resemble parents, etc.) there will emerge niches and multiple different organisms will converge adaptively into those niches (or the convergence will happen and define the niches, if you like … an important difference but one that does not matter for my prediction). That’s a pretty strong prediction. I see no reason to not accept its likelihood.

    (Sorry, I know this is a bit OT but … well, just sorry.)

    • John S. Wilkins John S. Wilkins

      Yes, but consider this: evolution may traverse any pathway that is physically realiseable. And that is a vast space of possible states.

      The properties you mention are highly abstract – what is the mass or entropy of imperfect replication or resemblance?

  2. Absent the contingent fact that we find ourselves at the end of a history replete with theism, why would anybody even be talking about gods? It isn’t as if a concept like deity falls naturally out of considerations of quantum field theory. The theism problem is merely unnecessary. A philosopher has no business taking an idea seriously simply because it is deeply meaningful to the majority of the human race.

    I’m enough of a Kantian to want an account of the nature if not the source of the categories I use. Since the God concept, at a minimum, includes such features as purpose, I need some story that will make sense of a metaphysical principle having characteristics that, so far as I can see, only make sense as something pertaining to animals. Of course if God really is the gaseous vertebrate of the old atheistical joke, there isn’t a problem; but the trouble with this solution is that it replaces an incoherent metaphysics with incredibly unlikely science fiction. I can accept that one does not need HOX genes in order to qualify as able to act, but can a being without a head and a tail or some other sort of gradient or direction be said to do anything?

    • John S. Wilkins John S. Wilkins

      I agree that one would not now find a reason to posit a deity. But all ideas and social institutions have a history, and neither is it the philosopher’s job to assert that an institution or idea should be rejected just because it has a history (that particular fallacy is called the Genetic Fallacy, by the way). I am interested only to know if a theist (of the kind I defined) can rationally believe in science, real science, and still be a theist. I have no interest in showing that theism is true or warranted, and indeed I think it isn’t. But I can take someone else’s views seriously enough to explore them, and maybe in the process, uncover something interesting about ideas in general.

      It turns out, on my account, that inherent teleology fails to be needed even by a theist. That might be of interest, don’t you think?

  3. My point is not that the concept of deity should be rejected because it has a history, but that it should be rejected because at this point all it has is a history.

    The trouble with a purely logical analysis of notions such as theism is that logic is a very weak instrument indeed, a fact which keeps the paranoids in business. More powerful methods are necessary to rule out even the most absurd and contrived hypotheses. In particular, it should be legitimate to point out how goofy certain hypotheses really are. That’s why I favor a certain degree of intransigence. How else do you convey the message that the problem with theological reasoning is that it presumes a legitimacy it doesn’t deserve?

    In practice, I have trouble remaining true to these impolite principles. I like to argue too much. And I can’t imagine what would be left of traditional theism without teleology so I’ll stay tuned.

    • My point is not that the concept of deity should be rejected because it has a history, but that it should be rejected because at this point all it has is a history.

      Affirm the consequent much? Theism has meaning to theists in the moment. Asserting that this is just a matter of historical inertia very conveniently spares you the work of having to engage that meaning on anything but your own terms. JW wrote that his project here was to take theists *seriously* before coming to pat conclusions about compatibility, which would mean making at least a token effort to get past the confirmation bias of caricature. What interest would you have in thwarting this effort? Is it just an altruistic desire not to see him waste his time on something you’ve already demonstrated to be obviated?

      I also think it would be a mistake to imagine that the only reason a person would believe in god(s) would be to explain the origin or teleology of the cosmos. There is a great deal of religious doctrine and practice that is largely if not wholly unconcerned with these questions. One of the reasons why the Argument from Moribundity is a dicey one.

      • Obviously nobody is going to keep people from arguing about God, and I have no special reason to want them to stop doing so unless they are philosophers. I’m not hostile to religions such as Christianity, I just think they’re erroneous. Where the burden of proof lies, however, is a real issue. A great many theological arguments depend upon the presumption that it is prima facie reasonable to talk about gods at all–Pascal’s Wager is the obvious example. My objection to theological arguments is based on pragmatics, not logic, i.e. on a different set of rules about what’s legitimate in discourse.

        Note that I’m specifically talking about versions of God that assume that God is the sort of being that acts or does things and has purposes. Notions of absolute or necessary being and other extremely abstract ways of construing god are another matter. For example if you figure that anything that is always present is a necessary being, it is perfectly reasonable to think of the Universe as God as the Stoics apparently did, assuming the eternity of the World.

        Chris, I understand that you don’t want to rule out ideas presumptively. I can sympathize with that. It’s just that after years of reading modern theological writings, I’ve found very little that doesn’t strike me as warmed over versions of 19th Century idealism.

      • I suppose that’s a fair comment, but even 19th C. Idealism needs to be taken seriously, yes? We may not follow every particular of Schopenhauer, but he’s still a giant, and deserving of some deference. Nor were the neo-Kantians anybody’s fools.

  4. Barry Rountree Barry Rountree

    Back when Zoe and I were having an extended email conversation, the idea of Universe-As-Turing-Machine came up. This fulfills your four requirements of theism, I think, with the twist that the deity doesn’t really care that bits of Her machine think they are conscious. Much closer in spirit to Epicurus, of course, but there is a plan, and the plan is designed….

    See Jürgen Schmidhuber’s “A computer scientist’s view of life, the universe, and everything”, DOI 10.1007/BFb0052071.

    Or, slightly more formally (IIRC):

    Brian Hayes, “Debugging the Universe”, International Journal of Theoretical Physics, 42:2, DOI 10.1023/A:1024447300346

    • John S. Wilkins John S. Wilkins

      My concern with that is that it conflates the thing doing the simulation with the thing being simulated (okay, maybe we are in the Matrix, but assuming the universe is real). The universe may be a machine, and it may be such that any complete representation of it may have to be the universe itself, but I do not think it is a Turing machine. That is because I do not think information is an objective physical property.

      • P.S. I think John’s objection deals well with my somewhat cynical “…whatever that might mean”.

      • Barry Rountree Barry Rountree

        No, no, the deity is not creating a simulation of anything. It simply wants to know the solution to Rado’s sigma function for some modest value (say, 42) and creates a machine to compute that. (Think of each cell on the tape being some Planck-sized volume of space, perhaps.)

        This assumes that God finds non-recursive functions as interesting as I do: to what better use could a universe be put?

      • John S. Wilkins John S. Wilkins

        The problem that arises with this, Barry, is that it denies one of the posits of theism; that the diety has a plan it can implement. On traditional theism, that implies that God is able to plan out in every last detail, and with complete foresight, the outcomes of any physically realised world. So why does it need the universe to predict? What need has God of a Turing Machine?

        I did not spell that out above, though, and I am going to appeal to a limited deity, which my coauthor and I call a “neo-Leibnizian” deity, so you can help yourself to this sort of deity, I guess. But it doesn’t go to the traditional theist view.

    • The concept of the Universe as a Turing machine actually shows an ignorance of the history of meta-logic/meta-mathematics. Turing conceived his machine as a solution to one of the problems in Hilbert’s programme for the foundations of mathematics; as Turing’s original paper stated he is making a contribution to the solution of the “Entscheidungsproblem”, i.e. decidability. The question is, “are all mathematical statements decidable”? That is given a correctly formulated mathematical statement does a mechanical process exist, an algorithm, that will tell us if the statement is true or false. Turing hypothesised that only those statements that are Turing computable are decidable. To claim that the Universe is a Turing machine is to claim that the Universe only consists of Turing computable statements, whatever that might mean. Unfortunately there are valid, correctly formulated mathematical statements that are not Turing computable, i.e. decidable, the continuum hypothesis being the most famous example.

      • Barry Rountree Barry Rountree

        The concept of the Universe as a Turing machine actually shows an ignorance of the history of meta-logic/meta-mathematics.

        Alas, what little expertise I have in the area comes from reading Turing’s original paper (included, with the rest of his significant work, in Copeland’s excellent biography), the reformulation in the first (?) chapter of Kleene’s _Metamathematics_, everything written about Rado’s Sigma function through ~2002 (which can fit into a modest binder), and a couple of indifferent grad classes in the Theory of Computation (as well as the papers cited above).

        I’m not sure if the above is sufficient to elevate me from “interested amateur” to “hobbyist”, but I certainly won’t claim anything beyond that (at least not until I finally finish reading Davis’s _Computability and Unsolvability_ that I’ve been dragging around the country with me for a decade now).

        So looking at the history of metamathematics /after/ Kleene’s 1950 reformulation, you find a body of research due to proto-computer-scientists wondering about the physical behavior of this theoretical construct. Several decks of punch cards are sacrificed to the cause, and the most influential (to the best of my knowledge) has been Rado’s question: given an n-state Turing machine and a program that halts, what is the maximum possible number of steps (or 1s on the tape)?

        n=1…4 has been decided (and the solutions to 3 and 4 warranted their own publications). Last I heard, n=5 was undecided. The best answer for n=6 is:

        3.187 × 10^{10566} 1s, 3.809 × 10^{21132} steps.

        Those are universe-sized numbers generated by a half-dozen if-then-else statements. Finding the solution for n>7 might require constructing a universe.

        If you’re willing to swallow the idea that consciousness &ct can arise through chemicals interacting via a handful of natural laws, I don’t think it’s too much of a leap to consider consciousness arising from cells on a tape interacting via a handful of rules. Thus, the papers cited in my initial post.

        [I’m afraid some of the confusion may have been caused by my never having seen “The Matrix”.]

  5. Great post!

    I worry, though, about the supposed completeness of a “purely physical explanation”. This is the sort of thing that often slides into arguments casually without being properly examined.

    Explanations in the sciences are never “fully physical”, they involve not only the forces that act on physical objects but presuppose the absolute validity of mathematics, for example. They also presuppose things like Hume’s uniformity of nature, something which seems (to him) almost miraculous. We can somehow rely on induction without having any rational basis for doing so.

    Explanatory gaps abound in such explanations, but they tend to be more abstract and philosophical than practical. Here, the deist/theist has plenty of room to move around (though, as Hume said, it’s unclear as to why their *particular* ideas have any more evidential weight than some other bizarre hypothesis).

    • John S. Wilkins John S. Wilkins

      Any explanation in terms of physical causation and systems is, in my view, a “physically complete explanation”, even if it does appeal to mathematical entities. All scientific explanations will be in the same boat WRT math, so I am happy to call them purely physical.

      The reality of mathematical objects (I’m a nominalist, BTW) is not going to change this issue one way or the other, unless you think that Platonism implies theism, which I’d really like to see an argument for.

  6. Nick Smyth wrote:

    …but presuppose the absolute validity of mathematics…

    I’m not sure what you intend by this half statement but one does not presuppose the validity of mathematics but prove it and as far as I know the mathematic systems used in the physical sciences have been proved valid!

  7. DiscoveredJoys DiscoveredJoys

    Interesting post. I’m not yet convinced that you have covered the preliminary forms of accommodation fully – but I guess I’ll have to see what you post next…

    It seems to me that should you prove that some form of accommodation is valid you will also have shown that some other forms – ones which may be more common – are not valid.

    • John S. Wilkins John S. Wilkins

      It’s more the case that I think the usual suspects, such as God intervening in the production of variants, as Asa Gray supposed (and to whom Darwin responded at the end of the Variation), fail. There are almost certainly other forms of accommodation that I have not thought of that are coherent, if this one is. That is, if my accommodation works, it is also a proof of concept for other possible forms of accommodation.

  8. This is an excellent post. Thank you for writing it. You said very clearly a lot of things I have long believed, but often have trouble expressing properly.

    Two points. First, you write, “That deity is a providential deity, with a plan that it implements towards an end.” If we are talking about traditional theism, then it is not just some plan in the abstract that we are discussing. Rather, it is a plan that specifically involves human beings. This is why people like Conway Morris and Ken Miller place so much weight on the convergence argument in their writing.

    Second, you write, “It is not, I suspect, the only way to do so, but all we need for it to be a rational position is one option that works. However, it may not be something that the ordinary Christian or Muslim could appreciate or fully accept.” I don’t know what you mean by an “option that works.” What constitutes working? Is the only standard that we not violate any laws of logic? That seems like a low standard. If an ordinary Christian or Muslim could not accept your argument then how successful can we really say your reconciliation has been?

    • Atheist.pig Atheist.pig

      Rather, it is a plan that specifically involves human beings. This is why people like Conway Morris and Ken Miller place so much weight on the convergence argument in their writing.

      I don’t know about Morris but Ken Miller doesn’t have that view. He agrees with Gould that if the tape were rewound it would come out different.

      • But Miller believes creatures with human-like intelligence would evolve a second time, and that is what is relevant to the argument.

  9. Atheist.pig Atheist.pig

    Well then its not a problem since people like Dennett believe that also.

  10. Nathan Nathan

    I wonder if you’re familiarity with Francis Collins’s book “The Language of God” – he tries to reconcile theism with Darwinian evolution. I’d be interested to hear your comments on his approach.

  11. AK AK

    I’m wondering if I’ve missed something here: you seem to be letting the ID people create the frame: a predictable clockwork universe that could be created in just such a fashion that it would evolve into our present world. When it doesn’t “Divine intervention” takes place in a supernatural event.

    But AFAIK the actual quantum universe is (as far as we know) unpredictable: when a wave function collapses the outcome can be predicted only statistically.

    Why not assume that God is actually controlling _every_ collapse of _every_ wave function? Since such collapses are behind most (probably all) mutations one way or another, He is then capable of guiding evolution in an ongoing fashion.

    Thus, his “plan” would exist at a high level, with the details constantly being adjusted through control of the wave function collapses.

    So in place of the medieval idea of angels flapping their wings steering the planets around we have bazillions of tiny angels specifying the results of wave function collapses.

    (Note that re-running the “tape” could produce a very similar outcome, although different animals might do different things, and the same mutations might have to be introduced in different places and times.)

    Of course what this notion does to “free will” I’m not sure: much of detailed neural activity is driven by activity at such a small scale that randomness (if it actually is random) in wave function collapse plays a significant role in whether/when a neuron will fire an action potential. I doubt we can assume that it all evens out: the brain is extremely non-linear.

  12. gregwrld gregwrld

    What if God is just a dilettante, bored enough to create a universe to amuse itself but not interested enough to stay away from its opium pipe (or whatever)? Every now and then it looks in on things, shuffles a few bits around and goes back to its pipe again. It sort of has a plan but changes it on whim for a laugh. In the meanwhile it’s creatures mistake it for something more noble when it truth it only cares as far as it is amused.
    Seems just as likely as any other explanation (of a religious nature).

    • John S. Wilkins John S. Wilkins

      Yes, there are an indefinite number of solutions that do not involve taking science and naturalistic explanations seriously. That’s not what we are addressing here.

Comments are closed.