Skip to content

Tautology 4: What is a tautology?

So, what is the problem, philosophically speaking, with something in science being a tautology?

The term has a particular meaning in the philosophy of science, and it is worth distinguishing the semantic from the logical here.

A semantic tautology is basically a definition. The Oxford English Dictionary defines it as

a. A repetition of the same statement. b. The repetition (esp. in the immediate context) of the same word or phrase, or of the same idea or statement in other words: usually as a fault of style.

The core of the criticism against natural selection is that it is a logical tautology, which amounts to it being an a priori truth (which most philosophers now deny exist, anyway).

In logic, a tautology is defined differently, and I quote:

A sentence ? is tautologous (or it is a tautology) if and only if it is assigned a truth-value T by every normal assignment of truth-values T and F to the sentences of L.

Further, a sentence ? is a tautological consequence  of a set of sentences ? if and only if ? is assigned the truth-value T by every normal assignment that assigns the truth-value T to all sentences of ?. [Benson Mates, Elementary Logic, second edition, page 89.]

Shorn of the logic-speak, it basically means that a sentence (not a phrase like “survival of the fittest” is a tautology if it is always true in some formal language L, and it is a tautologous consequence if it always gets assigned “true” in some set of sentences. The OED defines it more simply:

f. Mod. Logic. A compound proposition which is unconditionally true for all the truth-possibilities of its elementary propositions and by virtue of its logical form.

and assigns the first usage to Russell in 1919 (pp 203, 205).

For our purpose, this would mean that the Darwinian evolution by natural selection – call it NS – has “survival of the fittest defines fitness” as true no matter what else the theory of evolution says, under every interpretation, and, as Gould points out in his “Darwin’s Untimely Burial” essay, this simply isn’t the case. There are organisms with traits that are clearly not fit, and we can debate on empirical grounds whether traits are fit or not and why.

What is it about the logical tautology that is objectionable? Basically it is this: if NS is always true in the theory, then it is immune to falsification. This is why Popper took his original stance. For him, the very core of a scientific theory was that it could be falsified, so if anything cannot be, it is simply not science. Popper thought that NS was science, and so he had to find an accommodation, which he did with his “metaphysical research program”.

But nowadays we do not take Popper at his word. Science is regarded once again as a field in which testing statements and theories is both falsificationist, and verificationist, and neither, but merely a matter of establishing Bayesian likelihoods. So a principle can be something we gain confidence in as it “proves out” in experience without needing to go to the extreme lengths of the Positivists or the Popperians and find logical verification or falsification. Science is not the practice of classical logic.

As Maynard Smith once said, in a source I cannot now find, there’s nothing wrong with a bit of tautology in a mathematical system; every mathematical model must have them. So why should we now think that a tautology is problematic? The answer is, I think, dependent upon whether you think the tautology is useful or not. A “useless” tautology is one that is obvious (Brady 1979), whereas a “useful” tautology allows us to bring out the implications of the structure of a model and its implications. It is far from trivial, even if it is a truism and tautology, that natural selection can bring about a change in the overall makeup of a population. It is so far from being trivial that even though it had, in one form or another, been used to explain a lack of change for over 2500 years, selection-type explanations had not been used to explain change until Darwin suggested it. The equations that model natural and sexual selection have deep implications, which are very useful and often surprising.

Another problem with calling something a tautology rests on the notion of an a priori truth. Traditionally in logic and philosophy there were supposed to be some truths that were true by definition. One such is the statement “A is A“. A priori truths had a special place in reasoning, went the view, because they were accessible to all who reasoned, as the example of Plato “guiding” a slave boy to prove a theorem of geometry was supposed to show.

Since the middle of the twentieth century, however, analytic and other philosophers have come to conclude that the notion of “analyticity” or a priori knowledge is mistaken. If this is now the consensus, why should anything be regarded as a tautology in the sense of an a priori truth? Pragmatists have been saying something like this since Peirce, but given that pragmatism is roughly Darwinian theory in epistemology anyway, that might also be seen as question begging.


Brady, Ronald H. 1979. Natural Selection and the Criteria by which a Theory is Judged. Systematic Zoology 28 (4):600-621.

Mates, Benson. 1972. Elementary logic. 2nd ed. New York: Oxford University Press.

Russell, Bertrand. 1919. Introduction to mathematical philosophy. London; New York: G. Allen & Unwin; MacMillan & Co.


  1. b9n10nt b9n10nt

    “There are organisms with traits that are clearly not fit, and we can debate on empirical grounds whether traits are fit or not and why.”

    Can someone show me the empirical grounds of establishing the why?

    It’s all well to say that sharper beaks ? access to tougher seeds ?more food ? more mating opportunities ? more sharp beaks. But imagine running an enormous E. coli -type experiment involving thousands of generations of two populations exposed to the same environmental pressures. One can imagine finding that stronger necks that generate more thrust ? more mating opportunities ?…more finches with stronger necks.

    The statement “sharper beaks cause their own relative increase in environment X” is equally as true as “sharper beaks do not cause their own relative increase in environment X”. Thus, when a population of finches exhibit sharper beaks, can we still use “fitness” as a non-circular explanation?

    If causes are partial and contingent, are they causes at all?


    over my head but having fun,

    • John Wilkins John Wilkins

      A couple of examples that come to mind are the famous moths – we could see the birds preferentially taking the light morphs – and another being the lowered hybrid fitness of a lizard that is a crossbreed between those coloured to camouflage on rocks, and those coloured for sand. The hybrids stand out on both substrates, and again we see them being predated. In both cases there is a physical causal explanation.

      As to partial causes, I am not sure what that might be. A cause is a determination of the outcome. A class of causes might be partial, but an actual cause cannot.

      • inre lizards and moths: aren’t these simply examples of the transitional nature of NS? NS still holds true if, presumably, thousands of years down the line the hybrid lizards and light morphs will eventually have given way to their “fitter” cousins.

        However, if something cataclysmic occurred that by chance alone, rather than a bias based on genetic variation, wiped out the pure lizards leaving the crossbreeds to prosper, can we say that NS has been falsified?

  2. Just a very quick comment that doesn’t do justice to your 4 post series: last year in our uni discussion group, we read the paper Against Darwinism by Fodor (and a nicely-caustic reply by Dennett). Basically Fodor had the same kinds of objections, but he took it further to say that this means there cannot be an adaptationist explanation for anything in principle, that it’s all post hoc.

    I was just wondering if you’ve seen his line of argument? He’s apparently now working on a book length “philosophical demolition” of “Darwinism” and I’m already cringing…

    • John Wilkins John Wilkins

      I intend to address Fodor’s latest arguments in one of the next posts, along with a few others.

  3. Ah, great. I’ve found his one to be a bit like the ontological argument — it’s very wrong but it’s hard to point out the exact flaw.

  4. jeff jeff

    If you, as a novice god experimenting with universes, were given a “chemistry set” equivalent to our chemistry, could you deduce that what we call evolution and natural selection would occur under some special conditions, just based on knowledge of basic chemical interactions alone? You might be able to gain knowledge of more rudimentary things like gas laws or motion/force/acceleration in this way (like Newton) with minimal observation, but the only way to gain evolutionary knowledge is by observing it happening, and making post-hoc inferences and rationalizations about selection. It happened because… well, it happened. In this sense, existence itself is a tautology. Why does an electron or a universe exist? It exists because it exists. But perhaps the curious thing for many is why this tautology is manifest at such a relatively macroscopic level – where we are used to having a “why” for everything – rather than under extreme conditions where one is accustomed to unconventional thinking, such as the QM level, or origin-of-universe singularities.

    As for causality: If you reduce human understandable events to the molecular level and explain natural selection causality that way, then you are left with the problem of how to organize all those aggregate environmental and individual molecular patterns (occurring in both extended space and time) into higher-level meaningful constructs. Do you really think that your high-level grasp of causality is the only one? Do you really think that the human brain is the ultimate way of doing this mapping, and that higher interpretations are not possible – that the deepest of your abstract thoughts are not merely footnotes in a (higher being’s) larger apprehension? Even assuming the completeness of reductive physicalism, that would be the height of arrogance indeed, and would leave no room for further mental evolution.

    And even on the low end, to causally explain the behaviour of molecular interactions, you must keep reducing until you can’t anymore, and at the lowest possible level, you are still left with two or more concepts that may be “correlated” 100%, but with no real way to prove causation (if “causation” is even meaningful at that level). It is that way, because it is that way. As far as ultimate causation is concerned, you are back in the same boat you were at the macroscopic level: you observe object A collide with object B, perceive any correlated behaviour to the best of your human ability, and you make inferences about “causation”.

  5. Bill Morse Bill Morse

    I actually have no particular problem with the tautological claim. I also recall someone (perhaps with reference to Maynard Smith, but it wasn’t him – maybe Jared Diamond?) stating something to the effect that all good scientific theories are a recognition of a tautology.

    And Darwin’s main argument for evolution is to an extent true by definition: if you have excess reproduction and heritable difference you will get selection. That selection will be towards “fitness” if the environment and the inheritance is reasonably stable. The existence of convergent evolution tells us that in fact both are reasonably stable.

  6. jeff jeff

    Another thing that occurs to me: causality in general seems to require that the presentist view of time be correct. There has to be a “now” for something to cause something else. I can’t see how anything causes anything else in the block time model, where events are correlated based on relative distance from each other. In block time, “natural selection” is just a name for a reoccurring type of event pattern on the macroscopic scale.

Comments are closed.