A late August 2026 article from the Institute for Creation Research argues that crediting natural selection for altering organisms is a form of idolatry. Behind that claim is a scientific question about whether an unconscious process can produce an organized outcome, and that question has an experimental answer.
A few years ago a team of biologists began building fences in the Sand Hills of Nebraska. Six enclosures, each fifty meters on a side. Three of them sat on the pale quartz sand of the dunes, three on the darker soil beyond the dune field. The fences went deep enough to keep ground predators out and mice in, but the tops were left open, which was the entire point, because hawks and owls needed to be able to get at whatever was inside.
Into those pens the researchers released 481 wild deer mice, stocked so that each enclosure held a mixture of light coated and dark coated animals. Then they waited and came back and counted who was still alive. On the pale sand the pale mice did better. On the dark soil the dark mice did better. Anyone could have guessed that much. What Rowan Barrett, Hopi Hoekstra, and their colleagues did next is more useful to us. They sequenced survivors and non-survivors, identified a specific deletion in a pigmentation gene called Agouti whose frequency had shifted in the direction that soil color predicted, and then carried that exact sequence variant into the laboratory to show how it alters the protein’s binding behavior and lightens the fur. The results appeared in Science in February 2019.
Nothing in that experiment chose anything. Sand has no preferences. The owls were not implementing a plan for the future of Peromyscus maniculatus. And yet months later the distribution of alleles inside those pens was measurably different from what it had been at the start, in a direction predicted in advance from the color of the dirt. That outcome is what biologists are giving a name to when they use the phrase natural selection.
I start with this research because on August 31, 2026, the Institute for Creation Research published an article by its president, Randy Guliuzza, titled “Modern Idolatry: Crediting Natural Selection as the Creator,” and the article argues that language of this kind attributes divine powers to nature.
Guliuzza’s War on Natural Selection
Guliuzza’s case runs almost entirely through language. He quotes the historian of science M. J. S. Hodge observing that Darwin liked the term natural selection because it could be used as a substantive governing a verb, and Guliuzza inserts his own bracketed comment on that phrase, explaining a substantive as a real agent that can act. From there he gathers a series of quotations from evolutionary biologists and philosophers. Francisco Ayala describes a process that is creative although not conscious. Douglas Futuyma writes that selection explains apparent design without a supernatural designer. I. M. Lerner describes selection building step by step. Daniel Dennett writes that selection executes what was traditionally God’s design work. Stuart Kauffman calls selection an invisible hand that crafts well-wrought forms. Stephen Talbott calls it an occult Power of the prescientific age. Two further writers extend selection style reasoning to the origin of the universe itself.
The conclusion Guliuzza draws is that these authors ascribe rational powers to unconscious environmental features, that this is structurally what idolatry has always been, and that the problem is not confined to atheists. By his account his charge of idolatry also reaches creationists who accept selection as a real causal process while rejecting common ancestry. It’s the critique of fellow creationists that is what makes the article worth replying to reply because it puts a large number of theologically conservative scientists inside his indictment.
I am setting the theological question aside here entirely. My question is narrower. Does the evidence Guliuzza presents establish the claim he draws from it? Does a scientist’s use of an active verb demonstrate that the scientist believes an unconscious thing is thinking? Am I assigning some mystical force to natural selection when I use the term in my classroom?
No doubt, scientists do personify. The habit is widespread and it can often mislead students, and sometimes it misleads working researchers into treating selection as a more comprehensive explanation than it is. Jason Lisle made a version of that same point in the Answers Research Journal in 2018. The extension of selection style reasoning to cosmology, which Guliuzza flags in his closing paragraphs, really is a different sort of claim from population biology, resting on speculation rather than on measurement, and treating the two as equally supported would be a mistake. And the objection raised by Jerry Fodor and Massimo Piattelli-Palmarini, which Guliuzza quotes, points at a real methodological difficulty that biologists have to solve. None of that is in dispute.
So where is the dispute?
What the Term Actually Names
Natural selection is a name for what happens whenever three conditions are met in a population. First, individuals differ in some measurable trait. Second, some part of that difference is heritable, meaning that offspring resemble their parents in the trait more than they resemble the population average. Third, individuals with different values of the trait leave different numbers of surviving offspring. Biologists call that last quantity fitness, and the word means nothing more exotic than the reproductive contribution to the next generation. In other words, a “fit” animal is not the strongest or healthiest one, just the one that leaves more offspring behind.
If all three conditions are met, then the frequency of the underlying alleles, the alternative versions of a gene, changes across generations. That is not a further hypothesis that has to be added. It follows arithmetically from the three conditions, in roughly the way that a change in the proportion of red and blue marbles follows from removing marbles at different rates by color. The word “selection” is a label for the outcome of that arithmetic under specified conditions.
Notice what is absent from those three conditions. There is no fourth clause requiring an agent, a preference, an intention, or a force added to the system from outside. Selection is not a thing that pushes on organisms. It is a description of a demographic result, and the causes producing that result are ordinary physical and biological interactions: a seed too hard for a small beak, a predator that spots a pale mouse on dark ground, an antibiotic molecule that binds one variant of a protein and not another.
This is why the personification argument does not reach the science. Erosion carves valleys. Temperature controls reaction rates. Diffusion redistributes molecules. Gravity and moving water sort sediment grains by size and density, a description that appears constantly in creationist’ Flood Geology literature published by these same organizations. Every one of those sentences makes a noun the subject of a transitive verb, and none of them attributes thought to anything. Making something the grammatical subject of a verb identifies it as causally relevant. It does not give it a mind.
Three different things need to be kept apart here: plain causal language, figurative description of purposive action, and literal claims about intention. Guliuzza’s article repeatedly moves from the first two to the third without providing evidence that any author intended for their words to be taken that way.
Read more about this study here: https://www.hhmi.org/news/a-grand-experiment-in-nebraskas-sand-hills-shows-evolution-in-the-wild Credit: Rowan Barrett
Measuring a Process That Has No Mind
The strongest response to the claim that selection is imaginary is not an argument about words. It is the observation that selection has been measured, repeatedly, with numbers, by people who then made predictions and checked them.
The best known case is still the one Peter Boag and Peter Grant reported in Science in 1981. A severe drought struck Daphne Major in the Galápagos in 1977. The medium ground finch population on that small island fell from roughly 1,400 birds to about 200. The birds that died had a mean beak depth of 9.42 millimeters. The birds that survived averaged 9.96 millimeters. The difference is about half a millimeter, which sounds trivial until you know that the seeds remaining late in the drought were the large hard fruits of Tribulus, and that cracking them requires a certain bite force, and that bite force scales with beak depth. The Grants had already measured the heritability of beak dimensions by comparing parents with their offspring, and it was high. So they predicted that the next generation would have deeper beaks than the generation before the drought. It did.
That is a complete example of selection, documented from trait variation through differential mortality to a heritable response in the following generation, in a wild population, by people who were counting individual banded birds.
David Reznick and colleagues did the parallel experiment by deliberately moving guppies between Trinidadian streams with different predator communities and reporting in Science in 1997 that life history traits shifted in the predicted direction within a few dozen generations. Richard Lenski’s long term experiment with Escherichia coli, now well past 75,000 generations, has tracked the same process at the level of individual mutations, with frozen ancestral populations available for direct comparison, something no historical science usually enjoys.
And the Sand Hills experiment went the extra step. Barrett and colleagues did not simply observe that light mice survived better on light sand. They identified the specific sequence variant, tested its effect on the protein, and tested its effect on fur color. Trait, gene, mechanism, and survival, connected end to end.
The Problem Fodor Raised, and the Paper That Answered It in 1983
Guliuzza quotes Fodor and Piattelli-Palmarini to the effect that Darwin never got mental causes out of the account of selection and merely hid them in the analogy with animal breeding. Their actual objection is more specific and more interesting than the quotation suggests. They argue that biologists cannot distinguish selection for a trait from selection of a trait that merely happens to correlate with it, because nature has no mind capable of picking out which property is the relevant one.
The difficulty is real, and it does not require a mind to solve. Imagine a mineral sample in which color happens to correlate with magnetic susceptibility. Pass a magnet over it and certain particles are pulled out. The magnet does not understand which property matters, and it does not need to, because the causal structure of the situation is a fact about the particles and the field rather than about anyone’s beliefs. Whether color or magnetism is doing the causal work can be settled experimentally by holding one constant and varying the other.
Biology solves the problem the same way, and had a standard quantitative method for it decades before the objection by Fodor and Piattelli-Palmarini was published in book form. Russell Lande and Stevan Arnold published the framework in Evolution in 1983, in a paper titled, appropriately, “The measurement of selection on correlated characters.”
The method uses multiple regression of relative fitness on several standardized traits at once, which separates direct selection on a trait from indirect selection acting through its correlates. Ned Block and Philip Kitcher laid out essentially this response to Fodor and Piattelli-Palmarini in the Boston Review in 2010. And the Sand Hills work is the experimental version of the same logic: manipulate the candidate causal variant, hold the rest constant, and see whether the outcome changes.
Guliuzza’s Quotations Do Not Say What He Thinks They Say
The most instructive example is the Ayala quotation, because the disproof is inside the quotation itself. Ayala describes Darwin’s discovery of a process that is, in his words, creative although not conscious. The clause denying consciousness is right there in the sentence that is offered as evidence that consciousness is being attributed. In the original essay, published in the Proceedings of the National Academy of Sciences in 2007, Ayala goes further and expressly distinguishes evolution from an engineer executing a preconceived plan, explaining the process in terms of differential survival and reproduction arising from ordinary biological and physicochemical interactions.
The verb creates has a perfectly ordinary non-mental sense, which is to cause something to come into existence. Erosion creates valleys. Crystallization creates ordered structures. Recombination creates new combinations of alleles. Ayala is making a substantive claim, that evolutionary processes account for biological organization without inserting a designer into the scientific account, and Guliuzza is entitled to dispute it. But an explanation that dispenses with a conscious designer has not thereby installed a second one.
I would add a qualification that cuts against can be a loose usage among biologists. Selection does not manufacture the variants it acts on. Its contribution to adaptive evolution is the differential propagation of variants that arise by mutation and recombination, operating over and over alongside the processes that generate variation. Describing the whole outcome as selection’s creation compresses that distinction, and the compression is sloppy. But it is not deification of the process.
The Dennett passage is a philosophical argument, and identifying it accurately matters. Dennett claims that evolution undercuts a traditional argument for God’s existence by explaining what that argument attributed to a designer. One can reject that conclusion, as I do, without pretending it says something else. The statement that a process explains what was formerly attributed to a designer is not the statement that the process possesses the designer’s consciousness. Lisle addressed this same quotation in his 2018 paper, recognized that Dennett was using selection as shorthand for a broader evolutionary account, and judged Guliuzza’s reading of it mistaken, all while disagreeing with Dennett’s conclusion.
The remaining examples raise questions about explanatory scope and about precision of language. None of them shows that ordinary biological selection requires rational powers.
The Same Verb, Inside ICR’s Own Framework
Here is where the argument runs into trouble that it cannot resolve by appealing to evolutionary biologists outside the movement, because the clearest counterexamples come from the author’s own writing.
In his 2018 Acts & Facts article on sensors, part of the continuous environmental tracking series, Guliuzza describes organisms whose systems and traits use internal logic to select targeted responses to changed conditions. The same article presents organisms as active, problem solving entities that detect changed conditions, solve challenges, and fill new niches. That is the framework ICR promotes as the alternative to selection.
I do not find this language objectionable in itself. Biochemical networks implement conditional responses all the time without deliberating about anything. A signaling cascade that produces one output above a threshold concentration and another below it can reasonably be described as selecting a response. But accepting the use of anthropomorphic terminology settles the question before us. The word “select” does not require the presence of a mind. It does no give agency to the molecules.
If Guliuzza’s own mechanism a form of idolatry? If the verb select entails a conscious selector, then continuous environmental tracking attributes consciousness to bacterial regulatory networks. If the verb does not entail a conscious selector, then the case against natural selection collapses, because that case rests on the verb.
The likely reply is that God designed the internal system to produce those responses. That reply addresses the proposed origin of the system. It says nothing about whether the system thinks. A thermostat is designed and remains unconscious, and its design history cannot make consciousness a necessary condition for selection in one context and unnecessary in another.
Jason Lisle made the parallel point with a chemical example in the Answers Research Journal, and it holds up well. Chemists speak of hydrophobic compounds. The word literally means water fearing. No chemist believes the molecules are frightened, and the metaphorical wording does not make the phenomenon unreal. A precise account would specify the actual thermodynamics rather than implying a repulsive force, but the remedy for imprecise language is precise language, not the conclusion that the phenomenon does not exist.
The Sand Hills experiment addressed both the internal (CET) versus external framing (natural selection). The Agouti variant is internal to the mouse. The soil color and the aerial predators are external to it. Neither alone accounts for the result. Saying that water currents sort grains does not deny that grain properties matter, and saying that an environment favors a phenotype does not deny the organism’s physiology. Demonstrating that an organism has an internal regulatory mechanism (CET) also does not remove differential reproduction. Consider a population of bacteria with inducible defenses. Environmental exposure switches the defenses on inside individual cells. Heritable differences make some cells’ defenses more effective or less costly than others. Those differences affect how many descendants each lineage leaves. The first observation is about physiological regulation. The second and third establish a role for selection. Both happen at once, and discovering the regulatory mechanism does not erase the demography.
I would not endorse every argument these creationist critics have made, either. Lisle’s 2019 follow up sometimes reduces selection to the observation that survivors survive, and that formulation is too weak to do the work. Survival by itself does not distinguish selection from chance mortality, and survival is only one component of reproductive success. The scientific question is always whether particular differences among organisms systematically affect their reproductive contributions, and then whether those differences are heritable enough to produce a response.
What Would Count as Evidence Against Selection
A claim that cannot fail any test is not very helpful, so it is fair to ask what would falsify a claim of selection in a particular case. The answer is quite clear in this case. If the trait differences turn out to have no heritable component, there is no selective response to be had. If manipulating the candidate trait leaves reproductive output unchanged, the causal hypothesis is dead. If the observed allele frequency changes fall within the range expected from chance alone in a population of that size, then genetic drift explains the data and selection is not needed to explain that traits variation.
That last test is seen in the experiment we have been talking about. Barrett and colleagues built it into the Sand Hills design, comparing changes at the Agouti locus against roughly 2,100 unlinked genomic regions used as a neutral demographic control. In their light soil enclosures the number of survivors was too low to reject the neutral model. They said so in the paper.
This is the ordinary condition of the field rather than an exception. Motoo Kimura’s neutral theory, developed through the 1960s and 1970s, holds that most molecular evolutionary change is not adaptive at all, and the null model derived from it is the tool by which claimed cases of selection are routinely rejected. A large body of scientific literature exist of adaptive stories that did not survive being tested with a properly specified neutral expectation. Selection is, in short, a hypothesis that biologists frequently test and frequently discard.
One could say that if natural selection were a deity that it’s certainly not one that is in control of all genetic chang unlike what Guliuzza seems to think biologists are suggesting.
The challenge to Guliuzza
The August ICR article does not offer a criterion for telling the two cases apart, and that is the challenge that Guliuzza needs to meet if we are being asked to dispense with natural selection. Organisms use internal logic to select targeted responses. Environments favor some phenotypes over others. Both sentences use active verbs of unconscious systems. Both describe processes that produce organized outcomes without anticipating them. Both have been defended by people who believe God made the world. If one of them is idolatry and the other is sound design science, then some principled difference must distinguish them, and it cannot be which organization published the sentence.
Until that criterion is established, the more useful place to look is the data itself! Six fenced pens in the Nebraska Sand Hills, 481 mice, one deletion in one gene, and three months. Nothing in that experiment had a mind. Something in it changed anyway, and the change was measured, replicated across enclosures, tested against a neutral control, and traced to a molecular mechanism. Whatever we decide to call that, it happened.
Blessings,
Joel Duff
References
Ayala, F. J. (2007). Darwin’s greatest discovery: Design without designer. Proceedings of the National Academy of Sciences, 104(Suppl. 1), 8567–8573. https://doi.org/10.1073/pnas.0701072104
Barrett, R. D. H., Laurent, S., Mallarino, R., Pfeifer, S. P., Xu, C. C. Y., Foll, M., Wakamatsu, K., Duke-Cohan, J. S., Jensen, J. D., & Hoekstra, H. E. (2019). Linking a mutation to survival in wild mice. Science, 363(6426), 499–504. https://doi.org/10.1126/science.aav3824
Block, N., & Kitcher, P. (2010, March 1). Misunderstanding Darwin: Natural selection’s secular critics get it wrong. Boston Review. https://www.bostonreview.net/articles/ned-block-philip-kitcher-misunderstanding-darwin-natural-selection/
Boag, P. T., & Grant, P. R. (1981). Intense natural selection in a population of Darwin’s finches (Geospizinae) in the Galápagos. Science, 214(4516), 82–85. https://doi.org/10.1126/science.214.4516.82
Fodor, J., & Piattelli-Palmarini, M. (2010). What Darwin got wrong. Farrar, Straus and Giroux.
Futuyma, D. J. (2013). Evolution (3rd ed.). Sinauer Associates.
Good, B. H., McDonald, M. J., Barrick, J. E., Lenski, R. E., & Desai, M. M. (2017). The dynamics of molecular evolution over 60,000 generations. Nature, 551(7678), 45–50. https://doi.org/10.1038/nature24287
Guliuzza, R. J. (2018). Engineered adaptability: Creatures’ adaptability begins with their sensors. Acts & Facts, 47(3), 17–19. https://www.icr.org/article/engineered-adaptability-creatures-adaptability
Guliuzza, R. J. (2026, August 31). Modern idolatry: Crediting natural selection as the Creator. Acts & Facts, 55(5), 20–21. https://www.icr.org/article/modern-idolatry-crediting-natural-selection
Jeanson, N. T. (2013). Does natural selection exist? A critique of Randy Guliuzza’s claims. Answers Research Journal, 6, 285–292. https://answersresearchjournal.org/does-natural-selection-exist/
Kimura, M. (1983). The neutral theory of molecular evolution. Cambridge University Press.
Lande, R., & Arnold, S. J. (1983). The measurement of selection on correlated characters. Evolution, 37(6), 1210–1226. https://doi.org/10.1111/j.1558-5646.1983.tb00236.x
Lightner, J. K. (2019). Dubious claims about natural selection. Answers Research Journal, 12, 41–43. https://answersresearchjournal.org/dubious-claims-natural-selection/
Lisle, J. (2018). Refuting dubious claims regarding natural selection. Answers Research Journal, 11, 301–340. https://answersresearchjournal.org/refuting-dubious-natural-selection/
Lisle, J. (2019). Responding to Lightner’s comments on natural selection: Points of agreement and disagreement. Answers Research Journal, 12, 44–51. https://answersresearchjournal.org/response-lightner-natural-selection/
Reznick, D. N., Shaw, F. H., Rodd, F. H., & Shaw, R. G. (1997). Evaluation of the rate of evolution in natural populations of guppies (Poecilia reticulata). Science, 275(5308), 1934–1937. https://doi.org/10.1126/science.275.5308.1934
Comments or Questions?