Is natural selection science (or philosophy)?

I intervened on one of the many current threads against about Intelligent Design at BioLogos yesterday in response to the oft-repeated claim that evolution is not random because natural selection is not random. I suggested that, as per my last post, part of the rational limitation of science ought to be the recognition that it can only construct theories about repeatable regularities, whereas it can merely observe and list the contingent – and natural selection is firmly in the latter category.

I concluded my comment thus:

If science is the study of the repeatable, what makes natural selection any more a scientific process, than is contingent history – which is as much as to say a product of providence?

That it drew considered responses from Merv Bitkofer, Sy Garte and GD (all Hump persons, of course) did not surprise me – but it also more interestingly received a “like” from Joshua Swamidass.

Let me expand my thinking here. Natural selection is not random in the ontological sense of “arbitrary”, because it has causes. But science does not (when properly handled) deal in ontological randomness anyway: only in epistemological randomness – that humans cannot discern the causes of some process enough to predict its outcomes. “Chance”, in science, means only “causes unknowable.” And since the causes of natural selection are not knowable in advance, they are by the proper scientific definition random. So what kind of definition can see selection as “non-random”?  Clearly only a definition from outside the methodology of science.

In this way, I suggested, natural selection is directly comparable to human history. There are so many contingencies involved in human affairs that no human can predict even the immediate future confidently. Merv helpfully cited the example of “psychohistory” in Isaac Asimov’s sci-fi “Foundation Trilogy” (though I must modestly claim priority in this application from 2013). History is intrinsically about contingencies, which is why attempts to create scientific theories of history have failed, and failed on a grand scale too, if one considers overarching theories of history like that of “scientific” Marxism with its baleful effects on the world. And that is why history is not considered a science, but a humanity.

This is not altered by the fact that historians can, after the event, unpick the causes of historical events, and even enumerate those that are common. In the first place, such hiroshima-bombing-enola-gaystudies seldom produce any repeatable principles which predict future outcomes. To this end, in a subsequent BioLogos comment I pointed out that even some apparently incontrovertible historical truth, like “The Allies inevitably won World War II because they possessed atomic weapons and used them first,” was almost immediately rendered unrepeatable by the nuclear race with its doctrine of Mutually Assured Destruction. The whole Cold War was waged on the basis that having the Bomb and using it first would decidely not guarantee victory.

In real life, of course, no single factor is usually so overwhelmingly decisive as unilaterally available nuclear weaponry. Even in retrospect historians can only achieve a best-guess at causes from the limited documentation available to them. Those best guesses are also at the mercy of personal and societal bias. For example, the dominance of Marxism in recent historical academia has tended to blind historians to the possibility of religious motivation, rather than political or economic motivation, for actions in mediaeval times (notably the Crusades – see for example Rodney Stark’s counterblast in God’s Battalions. Even his own biases reinforce the case that history is necessarily the variable human interpretation of contingencies, not the application of scientific principles).

But fallible interpretation is the case, in retrospect, even when such glaringly obvious causes as the dropping of Big Boy and Little Boy are known. It is not impossible that the real cause for Japan’s surrender was some traitor in the Emperor’s court, unrecorded in the available records, who had been undermining the Emperor’s resolve for months before the bomb was (in that scenario unnecessarily) dropped. As for prediction, Oppenheimer could not predict that his bombs would win the war – the Germans were working on nuclear weapons, and who could be sure that the Japanese, benefitting from German expertise, were not at that moment preparing to hit New York before Allied weapons were ready?

bearIt is certainly very plausible that a stark instance of natural selection like the white fur of polar bears accounts for their success: but we cannot know for sure that other adaptations to cold were not, in fact, more important. And in very few cases is selective advantage as clearcut as that – for every polar bear or Hiroshima there are thousands of species of similar insects in almost identical niches, or skirmishes with thousands of potentially decisive factors.

So there seems little scientific advantage from knowing that there are traits that, given the right contingent circumstances, have been implicated in assisting survival. So what? The historian has known since time immemorial that ambition, technology, feminine wiles, family jealousy, climate change and signs from heaven (2 Samuel 5:24) have all been factors in historical outcomes. They are all true causes, but being irreducible to scientific principle, they are not scientific causes.

signoI deliberately mentioned miraculous signs (as a marker for God’s activity) because, in the context of natural selection, theistic evolution must consider the role of providence as much as it does in any other aspect of nature. God, in Christian theology, is the Lord of history, and it can never be known in historical events just how the power of God interacted sovereignly with other contingencies to bring about the actual result.

When it comes to the reproducible regularities of true science, this can be ignored, by subsuming it to the faithfulness of God and calling it “law”. But contingency is another matter – and as I argued in the last post, contingency is something that science can only observe, not theorise upon. The scientific category “random” is a bucket into which all contingent causes, including the designs of God, must be thrown perforce because of methodological naturalism’s commitment to only repeatable, material efficient causes.

Natural selection depends on the concept of “evolutionary advantage”. But “advantage” is a value that is not measurable, and is therefore a philosophical, rather than a scientific, category. One can measure speed, or oxygen-carrying capacity, or even the whiteness of fur, and use them to proxy one’s (teleological) value-judgement of advantage.

One can even measure number of offspring, which gets closest to the usual definition of “fitness” as “differential reproduction”. But by reducing the meaning of advantage to that extent, the core distinction of natural selection from mere happenstance is lost. For a gene allele that becomes fixed in the population by neutral drift is somehow not considered to be selected, though it replaces all other variants according to the definition of “fitness”. And if kangaroos were to become extinct because Australia was struck by an asteroid, it would be rather specious to call it an example of non-random natural selection, presumably by seeking to argue that asteroids move in predictable orbits. Surely “fitness” should mean more than “not blasted to kingdom come”.

So, I suggest, “advantage” (and therefore “selection”) cannot be defined scientifically, but only as a philosophical concept tacked on to science. Or at least, if one argues that “evolutionary advantage” can be defined scientifically, then one must concede that it is equally possible to define “the good” in classical philosophy as a scientific category. After all, Plato would say that the pusuit of “the good” makes one “happy” rather than “miserable”, meaning nothing more than that it brings advantage to those who possess it.

In fact, though, since the old philosophers were relatively consistent in stating what virtues always constituted “the good”, and what vices detracted from it, then “the good” is far more rigorously defined than “fitness” has been by scientists, if all the latter means, white-pinion-spottedpractically speaking, is “reproductive success”: reproduction is, after all, just a numerical result, not a cause of anything. The causes of fitness, unlike the philosophical virtues, are infinitely malleable. Blackness favours industrialised peppered moths, but not polar bears (or even the thousands of insect species that did not acquire melanotic forms even though they rested on sooty trees). Aggression favours chimpanzees, but not their closest relative the bonobo. The only fixed point of definition is survival, and survival, under current knowledge, also results from neutral mutations and sheer dumb luck. It’s Contingency Café.

In summary, the argument is not that natural selection does not happen because God guides it, any more than one argues that there is no such thing as history because God is sovereign over it. But rather, I suggest that because, like history (for natural selection is history, no more nor less) natural selection is contingent, and not lawlike, it belongs outside science. And if one wants, for whatever reason, to retain it under the scientific umbrella, one has to justify why one does not also embrace other contingencies, such as divine choice, under that same umbrella of science.

Jon Garvey

About Jon Garvey

Training in medicine (which was my career), social psychology and theology. Interests in most things, but especially the science-faith interface. The rest of my time, though, is spent writing, playing and recording music.
This entry was posted in Creation, Philosophy, Science, Theology. Bookmark the permalink.

10 Responses to Is natural selection science (or philosophy)?

  1. pngarrison says:

    This has nothing to do with your post, but it was something I thought you would like to see and I have lost your e-mail address. (Mine is the same as before if you want to reply.)
    Evolution of Hoxa11 regulation in vertebrates is linked to the pentadactyl state

    Not open access, unfortunately, but the abstract is fairly informative.

  2. Sy Garte Sy Garte says:


    I really hate it when you make me think too hard, which happens a lot, so please try to show some consideration in future. All kidding aside, your idea is brilliant, and of course very convincingly stated. And as you know we have similar conversations before in various venues. But your biologos post and this one are the first time I understood your point that natural selection really belongs outside of science, and I appreciate and agree with your rationale. What I especially like is opening the door to other contingent unlawful (meaning not subject to mathematical analysis) areas of thought.

    But I think your analogy to historical knowledge is not quite accurate, in that while natural selection is contingent in terms of specific predictions regarding specific phenotypes, one can make some prediction about evolution based on NS that cannot be in fields like history. We know for example, that any phenotype that results in increased probability of producing offspring (which includes higher chance of survival to reproductive age) will be selected for and lead to evolution. We cannot say anything quite that definitive in historical analysis. Marx tried to (with not very good results) since the idea that the working class will work toward its own interests, might seem like valid logic, but as the present situation on this side of the pond shows, often doesnt hold. And so on.

    But there is a more important feature of evolutionary theory that I believe is more important in keeping evolution, and biology in general in the scientific fold (no matter how tenuously). The phenotype is the target of selection. But if there were no link between phenotype and genotype, which is provided by the genetic code and the translation system, it wouldnt matter if some bird had better eyesight thanks to some mutation. That improvement would die when he died, and the kids would be like everybody else. The real miracle at the heart of evolution (and biology) is the fact that all of life shares a system that can pass down genetic information based on how well the resulting phenotypes do. This is why translation (and replication) need to be almost (but not completely) error free.

    This doesnt answer the objection that it is impossible to predict which specific phenotypes and which specific mutated genotypes will result in fixation and evolutionary change. But I am not quite ready to accept that the contingent aspect of NS moves it entirely outside of science, I think it better to say that NS require a new approach to MN in science, as we have discussed before.

    • Jon Garvey Jon Garvey says:


      Raising the issue for thought is the main point, so it’s no disaster if my thesis has to be modified. To question how hard the boundaries of science are is the aim – and so to raise awareness that science cannot be entirely self contained, but will inevitably make use of philosophy etc.

      Your third paragraph also raises interesting questions on the law/contingency axis. Clearly the code and its mechanisms operate in a lawlike manner, but the errors on which evolution depends are, individually, contingent (and so on my argument outside science proper). Yet there are, statistically, known error rates which can be studied and formalised as laws, at least to some extent (whether or not the molecular clock stands up as a valid model).

      I still suggest that natural selection is a lot more difficult to pin down in that way. And I’m still not sure I should capitulate on the comparison of NS with history – it’s true that anything that aids survival and reproduction will be selected, but also that almost anything, in the right circumstances, might achieve that. I’m not convinced that “selection” and “survival” don’t simply mean the same thing, raising the old spectre of tautology. Yet certain traits will, generally, aid survival (such as retaining basic life functions and so on).

      But that’s true for history, too – using the warfare example on which I majored in the OP. One can say that anything that leads to victory in a war will be favoured, leading to the political dominance of the victor. That’s as specific an end as reproduction – but equally subject to contingencies, where what wins one war will lead to defeat in another. Yet, like biological variation, certain traits like having big armies, lots of money and strong leadership will tend to favour success.

      Does the rule “whatever leads to victory produces political dominance” constitute a scientific law of history, or just a rather tautological truism?

  3. swamidass says:

    Hey Jon,

    While I would not go so far as to say that NS is outside the purview of science, I would agree with its characterization as a type of “historical contingency”. I also agree that science calls things “random” that it cannot fully predict, but this should be overinterpreted as some deep statement about lack of patterns or “inability of God to have foreknowledge of”. In this sense, it ties in with a long line of though about providence and historical contingency.

    I will also point out that this dovetails nicely with Francis Collin’s proposal for God’s direction of evolution: historical contingency. He once supposed that God sent an asteroid to kill of the dinosaurs 65 million years ago to enable the rise of mammals, and then us. Of course, NS (and therefore evolution) are intimately shaped by historical contingencies like this at every stage. Each of these points is critical to explaining the full diversity of life we see (and not intrinsically biology either). Who knows how many of these God influenced by direct action? Science (and certainly genomes) tell us nothing about this (and therefore useless to the ID movement). Perhaps, even His manipulation here is required to bring us about.

    Any how, thanks for sticking it out on the latest flap with the DI. I’m glad that blow it is drawing to a close.

    • Jon Garvey Jon Garvey says:

      Hi Joshua. Thanks for commenting.

      Part, at least, of the purpose of the post was to insist that science (of course I really mean “scientists”) sticks within the limits of what it can actually say about randomness, which is (I have argued) that it involves unpredictable irregularities, aka contingency.

      Contingencies can be accurately observed or recorded by science (on the BL thread I cited the endless measurements of early taxonomists before any theory really existed to suggest lawlike process behind them), but everything beyond that – like seeking to explain their unpredictability – goes beyond science’s competence. And as you rightly say, commenting on God’s involvement or lack of it is such an encroachment on non-science.

      Donning the theological hat is necessary to comment on “Who knows how many of these God influenced by direct action?” I’ve increasingly come to believe that the way to think of providence is as God’s instrumental use of secondary causes, that is, nature is the instrument on which he plays the music of his sustaining and governance of the world. That is my conclusion from the biblical data, and it echoes the thinking of the theological-philosophical greats like Aquinas.

      If that is so, the question “What did God influence by direct action” becomes a strange thing to ask – rather like, “Which bits of the Mendelssohn did Menuhin play directly, and which were down to the Stradivarius?” If asteroids are within God’s providence, then the KT event was providential (and one may extrapolate to whatever else is sustained in being by him – a pretty universal extrapolation, on classical doctrine!)

      Yet (touching on the TE-creationist divide) contingency is still contingency (and so beyond the understanding of science) whether it comes through secondary causes or miracle. I have long maintained that chance is formally indistinguishable from miracle. Eddie was right to point out back at BL that to significant thinkers (notably Aquinas) there was a particular glory in acts done by God without “natural means”. That balances the claim from Charles Kingsley onwards that God’s use of natural means is a particularly glorious way to create. That fact is that God’s sovereign freedom allows for both, and we’re seldom in a position to determine which has occurred empirically.

    • Edward Robinson Edward Robinson says:

      Hello, Joshua.

      You wrote:

      “Who knows how many of these God influenced by direct action? Science (and certainly genomes) tell us nothing about this (and therefore useless to the ID movement).”

      I agree with your main statement, but not with your aside about ID.

      ID argumentation doesn’t rest anything on a determination of which things happened by direct divine action. ID tries to show that certain things wouldn’t have happened by chance, or even by chance plus natural laws, and therefore that design was a real causal factor in their occurrence. Whether that design was implemented by direct divine action, or through a long-range plan involving chains of natural causes, is not something ID methods claim to detect.

      Thomas Aquinas and many other Christian theologians have said that God can work through natural causes, or outside them. But he achieves his design either way. Thus, within ID, you can be a Meyer (insisting on specific inputs of information by divine action) or you can be a Denton (believing that God could set up the evolutionary process so that such inputs would not be necessary). What is crucial for ID is the affirmation that design is a necessary causal factor of living things, just as an architect’s plan is a necessary causal factor for the existence of a building.

      Many EC writers do not acknowledge that design is a necessary causal factor; they think that design is a religious perception which is optional for explaining the phenomenon observed. Thus, they will say that Richard Dawkins and Ken Miller agree on the science, and disagree only over the voluntary religious interpretation they offer about God’s involvement or non-involvement in the evolutionary process. Whether or not design is a real cause is thus left entirely up to the religious world view of the scientist in question.

      But this is strange; we would not say that the belief that an architectural plan is necessary for the existence of building was a question to be left up to the religious world view of the observer in question. We wouldn’t say that it was OK if Ken Miller accepted the inference to an architectural plan but Dawkins denied that inference. We wouldn’t say that this was a question of “values, purpose, and meaning” which doesn’t belong in science, but in the realm of private faith. We would say that Miller was right and Dawkins was wrong, because the evidence for a plan was overwhelming.

      And that is what ID is arguing: that the evidence of nature is such that it’s reasonable to infer that a plan was necessary. Not that direct intervention was necessary, but that a plan was necessary.

      This is why I find it so irritating when people like Dennis Venema (who seems to be an honest and polite guy, so there is nothing personal in my remark) make out as if what they are objecting to is a claim by ID people that supernatural intervention is necessary to explain biology. In fact, Dennis’s steady position, sometimes stated, sometimes only implied, has been that, *even if it did not require miracles to implement, design would still be a superfluous hypothesis*, since as far as we can tell from our current science, everything in life could in principle have arisen if there was no design at all. So it’s more than inferences to supernatural action that he opposes; it’s inferences to design period.

      The logical consequence to Dennis’s position (and not just to his, but to Applegate’s, and Falk’s, and Giberson’s, etc.) is that God is the Creator of life and man, but didn’t *design* life or man. Our world does not need a designer, they tell us on their authority as scientists; but, they tell us as Protestant evangelicals, it does need a Creator. But what is the content of the notion “Creator” if that notion is severed from the notion of design?

  4. Edward Robinson Edward Robinson says:

    I enjoyed this column very much, Jon.

    Both the point regarding natural selection as science (how hard it is to pin down which of a million possible causes might have made a thing survive, and how hard it is to define “fitness”), and the point regarding contingency in history and biology (how hard it is to assimilate the study of historical events — human or natural — to the model of natural science which studies the repeatable), were well-made.

  5. Jon Garvey Jon Garvey says:

    Thanks Eddie

    A number of interesting issues arise in this theme – a third (after “definition” and “contingency”) is the teleological nature of “fitness”, when examined, that Jerry Fodor pointed out. Science deals, properly, with efficient causes:

    “Polar bears are white, therefore they reproduce well in snow, therefore they are defined as fit.”

    But then this idea of fitness for reproductive success becomes the focus, and one puts the cart before the horse:

    “Polar bears are fit for reproductive success because white, but unfortunately they actually don’t survive because the seals eat them.”

    The logical strangeness becomes apparent if one defines fitness in some other specific way, say (for example) “fitness as an anteater”. You then try and say that a particular armadillo is “fit” though it eats spaghetti, saying it would be a fantastic anteater if it could only eat ants.

    The only way to understand that is that fitness is for a purpose, but that circumstances can frustrate that purpose. The teleological supposed “analogy” has then lost sight of the scientific concentration on efficient causes: you must start from the result (reproductive success) and deduce the cause (fitness – whatever that actually is). As you suggested to Steve Schaffner on the BioLogos thread, science does not do the subjunctive!

  6. Robert Byers says:

    The conclusion can be that since biology is about what we see and so its origin about what we don’t THEN its only a line of reasoning that the present came from selection.
    Right or wrong its just lines of reasoning. its not testable. when they say they test it they only test the line of reasoning once more.
    Evolutionism has never been a intellectual investigation of biology. Its been lines of reasoning from some raw data.
    This is why Darwin, as Stephen Gould said and did himself, so relied on geology.
    Without the geology the evolution of biology not only is not evidenced but actually fails.
    New ideas on geology could destroy the evolution biology claims.
    This evolutionism is not a biological scientific subject.
    its something other.

Comments are closed.