Skip to content

The tautology problem

A long time ago I wrote a not particularly good piece on the tautology problem: that natural selection is merely circular definition. I was just out of being an undergraduate when it was published, so it was at best an undergraduate piece. I have been unsatisfied with it ever since. So I will from time to time post on this. To start, what’s the history?

You may have thought – I did – that this would be a criticism of Darwin from the start: that survival of the fittest is an empty tautology. But in fact it doesn’t seem to have started until the centenary of the publication of the theory of evolution in 1958, which is when a number of mythologies about Darwin seem to have originated. I wonder if this is because there really isn’t anything as revolutionary individually about Darwin’s ideas, when the totality is itself a revolution? The fallacy of division may be to blame; we want Darwin to be revolutionary in all his parts as well as overall. Anyway, that’s for another post.

At least one source suggested it began with the Bishop of Carlisle in the 1890s, but when I tracked him down, he in fact merely said that natural selection was not creative, but merely filtered out disadvantageous variants:

I am not going to make any objection to Mr. Darwin’s ingenious theories with regard to natural selection, the survival of the fittest, and selection in relation to sex; but I wish to impress upon you this point, namely, that there is in all these theories no real machinery (so to speak) for the improvement or change of species or races: supposing that you have got your improvement there is a reason assigned why it should become permanent, but no reason is assigned why it should in the first instance occur1.

1. See Mozley’s Essays, vol. ii, p. 396. “Natural selection is not an agent, but a result. . . . Natural selection only weeds, and does not plant; it is the drain of Nature carrying off the irregularities, the monstrosities, the abortions ; it comes in after and upon the active development of Nature to prune and thin them ; but it does not create a species; it does not possess one productive or generative function.”

. . . You will, of course, observe that in what has now been said I am by no means denying the possibility, or even the probability, of evolution of higher forms of life from lower: I have already agreed to grant all that may be asked in this direction: what I insist upon is that the survival and perpetuation of a higher form of life, when it has appeared, affords no explanation whatever of the fact that it did appear. [Goodwin 1880, p31, 33]

The logical nature of Darwin’s theory was in fact mentioned by none other than C. S. Peirce:

The Darwinian controversy is, in large part, a question of logic. Mr. Darwin proposed to apply the statistical method to biology. The same thing had been done in a widely different branch of science, the theory of gases. Though unable to say what the movements of any particular molecule of gas would be on a certain hypothesis regarding the constitution of this class of bodies, Clausius and Maxwell were yet able, by the application of the doctrine of probabilities, to predict that in the long run such and such a proportion of the molecules would, under given circumstances, acquire such and such velocities; that there would take place, every second, such and such a number of collisions, etc.; and from these propositions were able to deduce certain properties of gases, especially in regard to their heat-relations. In like manner, Darwin, while unable to say what the operation of variation and natural selection in any individual case will be, demonstrates that in the long run they will adapt animals to their circumstances. Whether or not existing animal forms are due to such action, or what position the theory ought to take, forms the subject of a discussion in which questions of fact and questions of logic are curiously interlaced. [Peirce 1877]

Clearly this is not a criticism. In fact, by making the analogy with the statistical theory of gases, Peirce is making a rather deep point, one I hope to return to later. The deep point is that a general, mathematical, explanation is one that the best science aspires to. Darwin’s theory of natural selection is a logical theory, yes, but it relies on questions of fact.

The first apparent statement of the tautology claim was in 1958, as I said, and it was not a criticism of natural selection:

These [struggle for existence, survival of the fittest] emotionally loaded phrases have been often misused for political propaganda purposes. A less spectacular but more accurate statement is that carriers of different genotypes transmit their genes to the succeeding generations at different rates… The “fittest” is nothing more remarkable than the producer of the greatest number of children and grandchildren. [Sinnott, Dunn and Dobzhansky 1958, pp. 100-101]

To head off the political misuse of natural selection, these leading authors point out that the principle of natural selection is merely a definition. “Fit” does not have any deeper meaning than “number of descendants”, so Nazi and other forms of political evolutionism are simply mistaken in their view of what matters. A nice side effect of this is that if the Nazis are right that the “subhumans” are outbreeding the “Aryans”, then the “subhumans” are more fit! Deal with it.

But, as Diane Paul went on to say, “Thus was born the famous “tautology problem”, which has bedeviled the field ever since” [Paul 1992, p114].

At first, this was not terribly concerning to anyone. The mathematics of evolution, developed as population genetics, simply posited a variable w for fitness. Individual organisms had a w, populations had an average w, and genes – alleles – had a w, which was basically the statistical number of copies after some number of generations (usually three, the F2 generation or second daughter generation).

Then, a philosopher, Michael Lerner, wrote this for the centennial of the Origin:

Thus, although Darwin may have erroneously accepted the view that survival is necessarily the major component of natural selection, he understood that the crux of the selection process lies not merely in the ability of an individual to survive longer than others, but in its greater capacity for production of living offspring; that is to say, not only in survival of individuals within a generation, but of groups over a period of generations.

This fact enables us to answer the third question, that regarding the interpretation which must be given the word fittest, or rather, the term fitness. The expressiveness of Spencer’s superlative seduced Darwin into accepting “survival of the fittest” as being equivalent to “natural selection.” As a result, many early evolutionists and a fair proportion of those naturalists of today whom the literature of population genetics has bypassed attached anthropomorphic value judgments to selection. Yet neither strength of character nor moral goodness, neither extreme size nor high intelligence, nor even long life per se causes an individual to produce more offspring, that is to say, make it fit in Darwin’s sense. Indeed, often organisms which are totally undistinguished by any physical standards, organisms exhibiting average dimensions for various properties, in fine, individuals which are mediocre for any traits obvious to the human eye, are the ones that are most successful in propagating themselves. They are therefore the fittest. If there is one thing upon which the most factious partisans of various currents of evolutionary thought agree, it is that fitness of an individual, in the context of the natural selection principle, can mean only the extent to which the organism is represented by descendants in succeeding generations. Fitness can be discussed in absolute terms or expressed relatively to the average of a group. Immediate or more remote generations of descent may be chosen as a point of reference. Enumeration of offspring or some other way of assessing the “amount” of progeny left may be resorted to. But in all instances, fitness must refer to the ability of an organism to leave surviving offspring.

A habitual fallacy, shared by both the opponents and the supporters of evolution by natural selection, is the idea that any part of evolution may be explained by saying that the fittest individuals have the most offspring. When fitness is considered with reference to evolutionary phenomena, such statement is logically circular and begs the question. If capacity of reproduction is the criterion of fitness, the only connecting proposition between reproduction and fitness which avoids tautology is that individuals having most offspring are the fittest ones. This is neither an assumption, nor a hypothesis to be proven, but merely a definition. [Lerner 1959]

Even here, this is about the term “fitness”, as a term in the theory. It’s just a variable that is used in the calculation. Again, Lerner is striving to avoid the moral value of the term “fit” (the term “fitness” itself appears to have arisen sometime in the 1930s, possibly by Wright.* Fisher in 1930 used the term “reproductive value” in analogy to an economic theory, to underpin his positive eugenics) that was used by the Nazis and other “social Darwinians” that those who strive hardest in some morally laudable manner are those whom evolution favours. But now, we have to deal with the apparent lack of explicability of the theory of natural selection. How can it explain when the key property – fitness – is merely defined? [It did not escape the attention of people that this is also true of Newton’s term g for gravity in classical mechanics.]And then, in a book that instituted the modern version (the third generation) of neo-Darwinism, George C. Williams wrote:

Like the theory of genic selection, the theory of group selection is logically a tautology, and there can be no sane doubt about the reality of the process. Rational criticism must center on the importance of the process and on its adequacy in explaining the phenomena attributed to it. [Williams 1966, p109]

Here it is most clearly – natural selection is a logical a priori truth. What we must now argue over is how important it is. As Shaw once said, we know what we are, now we are haggling over the price.In the next post I will discuss how this became a philosophical, and later a creationist, problem for evolution, leading to it being the single greatest problem in the philosophy of biology in the 70s and 80s. Then, we shall consider the various arguments that go by this name, and after that I will make a positive argument of my own.

* Wikipedia says it was J. B. S. Haldane’s 1924 paper, but while he has a term un that he determines the rate of change ?un. This he terms the proportion of alleles in a nth generation. Later: I found it in Fisher’s book after all, on page 37, which he terms as m, “the objective fact of representation in future generations”. It’s in his fundamental theorem. How did I miss that?

Later note: See the correction that follows.


Fisher, Ronald Aylmer. 1930. The genetical theory of natural selection. Oxford UK: Clarendon Press, (rev. ed. Dover, New York, 1958).

Harvey Goodwin, Lord Bishop of Carlisle. 1880. The origin of the world according to revelation and science: A lecture. London: The Christian Evidence Committee of the Society for Promoting Christian Knowledge.

Haldane, J. B. S. 1924. A mathematical theory of natural and artificial selection. Transactions of the Cambridge Philosophical Society 23:19-41.

Lerner, I. Michael. 1959. The Concept of Natural Selection: A Centennial View. Proceedings of the American Philosophical Society 103 (2):173-182.

Paul, Diane. 1992. “Fitness” in Keller, Evelyn Fox, and Elisabeth A Lloyd, eds. Keywords in evolutionary biology. Cambridge MA: Harvard University Press, pp. 112-14.

Peirce, Charles Sanders. 1877. The Fixation of Belief. Popular Science Monthly 12 (November):1-15.

Sinnott, Edmund Ware, L. C. Dunn, and Theodosius Dobzhansky. 1958. Principles of genetics. 5th ed, McGraw-Hill publications in the botanical sciences. New York: McGraw-Hill

Williams, George C. 1966. Adaptation and natural selection: A critique of some current evolutionary thought. Princeton NJ: Princeton University Press


  1. Veronica Abbass Veronica Abbass

    Is this post connected to the “Darwin on species and heredity” that is under review?

    • John Wilkins John Wilkins

      No, it’s a separate and more standard philosophical problem.

  2. b9n10nt b9n10nt

    Darwin’s theory is summarized as “survival of the fittest”. Although the statement is a logical necessity, one who thus summarizes Darwin’s theory is asking her audience to synthesize Darwin’s key observations about survival (a scarce commodity among members of a population) and fitness (an aggregate of traits that are inherited with variations).

    Perhaps it is only after a hundred years of taking these key insights about biological populations as a given that thinkers no longer see all that is implied in “survival of the fittest”.

  3. jeff jeff

    I suspect this problem is related, on a deeper level, to the concept of cause and effect, or the lack thereof. It is the difference between the how and why questions. It is naturally human to find a cause for everything that happens, but “cause” is a human attribution that can change dramatically as more facts become available about the universe. Are events ultimately “caused” from the bottom-up (micro to macro, upward determination), from the top-down, or neither? Perhaps the best objective science should assume neither, and attempt to describe what occurs, and how things appear to be consistently related or “correlated” (especially mathematically), without attributing any ultimate human-understandable causes or asking “why”.

  4. John,

    Have you been following Fodor’s recent thoughts on this subject? (He supposedly is doing a whole book on it, but I haven’t heard much about that lately). He doesn’t nail it all, but it’s provocative.

    One expression of the idea that natural selection is tautological is that it isn’t predictive. Given a population of genetically variant organisms, we know some will differentially pass on their genomes compared to others, but we can’t say which ones without knowing more contingent facts than are generally knowable. There is an element of novelty or creativity (metaphorically speaking) in acts of adaptation that cannot be anticipated. “Fitness” is always a matter of hindsight, because biology never stops to let all things be equal. (Simon Blackburn writes in response to Fodor that “camoflage helps across the board”; but only relatively few species are cryptic.)

    This is very different than the question of gas pressure and volume. Though we only know what will happen statistically in physics, it is nonetheless predictable at our (non-quantum) level of observation.

    I think Fodor’s point is not to assert that selection isn’t partially explanatory of evolution, but that it shouldn’t be exalted to the status of a law, since under such pressure it begins to lose meaning. His deeper point is that natural selection without contingent factors is tantamount to panadaptationism, which is more ideology than science.

    • John Wilkins John Wilkins

      I keep a weather eye on Fodor, as it were, but I’ll deal with him in more detail later; I haven’t had the chance to get deeply into his recent stuff. I think that in the end I won’t be far from him on this.

  5. bob koepp bob koepp

    I think the “problem” arises because people aren’t careful to distinguish the contingent effects of fitness (such as number of offspring) from fitness itself. This is a sort of operationalism that should be guarded against. If fitness is characterized independently of any measures of its effects, then the claim that the fittest organisms are more likely to survive and reproduce is not at all tautologous, but quite informative. Analysis of ecological relationships is an important source of characterizations of fitness that can stand independently of the role of the concept of ‘fitness’ in the theory of natural selection.

  6. The theory of natural selection is not an explanation of anything at all but rather a widely useful pattern for explanations. Which is why the form of the explanation can be tautological, even though its instantiations are not. To put the matter in a rather Aristotelian fashion, living thing does not evolve, particular populations of living things evolve. To the extent that natural selection rather than genetic drift or some other kind of explanation is involved in how this comes about, figuring out what’s going on always involves the empirical, decidely nontautological question of how a particular character improves or does not improve the odds of survival and reproduction. Where there is no causal connection between a character and its increasing or decreasing occurrence over time, natural selection simply has not occurred.

    • John Wilkins John Wilkins

      Jim, shut up! You;re stealing my thunder! 😉

    • jeff jeff

      Where there is no causal connection between a character and its increasing or decreasing occurrence over time, natural selection simply has not occurred.

      One could argue that genetic drift is quite “causally” implemented at the molecular level. Is that natural selection too?

      But in any case, I thought the whole idea of natural selection is not that a character (property?) causally increases or decreases it’s own occurrence over time, but rather it is the “environment” that causally favors it or restricts it. One problem is how to identify a such property (it could be almost any abstract concept).

  7. jeff jeff

    If fitness is characterized independently of any measures of its effects,

    And how is that to be done? First, fitness is never absolute, but always relative to the context or environment of the individual. Second, the number of potential variables making up a total fitness probability value are likely too complex to identify or compute in many cases, especially when viewed in the context of large dynamic population and environment. The only way to find out is to let the universe compute it, and analyze the results, attributing “selection” in hindsight to what you observe to be the most significant factors. If we actually had “fitness” nailed down precisely, we would probably predict how any species would evolve.

  8. bob koepp bob koepp

    Jim Harrison’s remarks are suggestive of Bob Brandon’s view, expressed in “Adaptation and Environment,” that the pinciple of natural selection is a “schematic law.” So along with John’s thunder, I hear echoes.

    And to respond to Jeff, I’ll gesture toward Brandon’s discussion in the same book of how ecological/engineering analyses give empirical content to hypotheses about relative fitness in an environment. As just one illustration of what I mean by ‘independence,’ it certainly is possible to assess the cryptic value of , say, wing color in a particular species of moth, independently of any consideration of whether crypsis has in fact played a causal role in the relevant history of selection. In fact, we need to do just this if we want to understand how crypsis affects the process of natural selection.

    • jeff jeff

      (I assume the crypsis is assessed in the context of all the moth’s natural predators, and not via lab instruments…)

      The problem, as I see it, it is identifying all the complex properties that are being selected for, as well as all the complex properties that are causally doing the selecting. Presumably, if you are a reductionist, the causal aspect is fairly straightforward. All these complex properties eventually map to sets of interacting molecules in both the environment and the individual (generally speaking, you probably don’t need to go lower – and there may be philosophical problems in mapping sets of molecules back to higher level human-understandable concepts and vice versa -called “predicate dualism”, but I won’t bother with that now). “Natural selection” would be when the environment’s molecules causally interact (or don’t interact) with the individual’s molecules in a complex way so as to favor its replication, as opposed to any causal interaction that compromises the integrity of the individual’s molecular pattern or restricts its replication.

      But thinking in this reductionist-causal way is not going to buy you much if you can’t identify all the appropriate subtle (possibly even overlapping) properties in both the environment and the individual, to begin with. What you have then, is an inscrutably complex, long-term molecular dance between individuals and the environment that you call “natural selection”, but don’t really understand.

  9. Gary Nelson Gary Nelson

    AR Wallace wrote in 1913 that “It is a mere truism that the fittest survive” (Social Environment and moral progress, p. 97). Earlier he had referred to Natural Selection as a “mere truism” (1873) and “self-evident truism” (1909), while arguing against that point of view, espoused by Darwin’s critics. But in 1913 he was of a different persuasion.

    • John Wilkins John Wilkins

      Ah, that’s brilliant, Gary. Many thanks. I should have looked at Fred…

  10. J.J.E. J.J.E.

    Is the tautological appearance of natural selection a result of language being insufficiently precise? The way I think about selection is in terms of an abstract stochastic simulation with parameters. If I were to define a life cycle, a hereditary unit, mutation rate, sampling scheme, and selection coefficients, I could easily demonstrate most of the key concepts of natural selection without any problem, and I’m pretty sure it would be merely descriptive and have no tautologies at all. And of course, it would be a relatively crude model, but not totally useless.

    Of course jeff points out that connecting the predictions that an abstraction like a genetic algorithm makes to an actual instantiation in nature can be very challenging. The field of population genetics is obsessed with evaluating the fit of competing models with data drawn from “natural” populations. Rarely, an instance of natural selection is so obvious that one can isolate the genetic cause and demonstrate the effect in a lab under conditions that might be similar to the environment the allele might have encountered when it first arose by mutation. However, such demonstrations are considered exciting and invariably get into Nature or Science or whatever. That just speaks to how rare such demonstrations actually are. If they were commonplace, there would be legions of such examples of high quality in field specific journals.

    Moreover, in population genetics, it is considered in principle possible to quantify the error surrounding hypotheses that particular hereditary units that have undergone frequency changes as a result of natural selection versus say demographic processes, even if the particular selective agent cannot be identified. Of course, this does presume that we know a rather large amount about not only the locus in question, but also about other loci as well as the general patterns of demography in the species, etc.

  11. RichardW RichardW

    There are two meanings of tautology:
    1. A needless repetition of an idea, statement, or word
    2. A proposition which is true by virtue of its logical form alone.

    “Survival of the fittest” is not a case of 1 because you can’t delete any part of it without changing its meaning. It’s not a case of 2 because it’s not a proposition. Moreover, it can form part of a non-tautologous proposition.

    The word “survival” refers to a property that is inherent in the word “fitness”, but that doesn’t make the expression a tautology. Consider a clearer analogue: “the blackness of black cats”. This is not a tautology either. It can play a useful part in a meaningful, non-tautologous proposition: “the blackness of black cats is what makes it hard to see them at night”. Similarly, we can say “the survival of the fittest has played a major part in evolution” or (untruthfully) “the survival of the fittest has played no part in evolution”. These propositions are not tautologous either. Neither is true by logical necessity. They are factual claims which could be true or untrue.

  12. RichardW RichardW

    P.S. “The fittest survive” is a tautology (provided you take it to mean “the fittest are most likely to survive”, not “the fittest actually do survive”). Similarly, “black cats are black” is a tautology. But it doesn’t follow that “the survival of the fittest” and “the blackness of black cats” are tautologies. By the way, I take “the fittest” to mean “those who–by virtue of their inherited characteristics–are most likely to survive”.

  13. RichardW RichardW

    Perhaps my last two posts were off the point, because the question is not whether the expressions “survival of the fittest” and “natural selection” are tautologous, but whether the proposition to which they refer is tautologous. I suppose the proposition to which they refer could be taken as something like this:

    If individuals possess (and pass on) inherited characteristics that affect survivability, then the next generation will tend to have a greater number of individuals with those characteristics. Those characteristics will tend to spread over successive generations as long as the factors affecting survivability remain sufficiently stable.

    I think you can state the conditions (inherited characteristics, stable environment, etc) in purely abstract terms, with no reference to the real world, and then deduce the conclusion. So you could say it’s a tautology in the sense that mathematical theorems are sometimes said to be tautologies: they follow deductively from certain axioms. But the claim that mathematical theorems are tautologies is controversial, which is probably why the claim remains controversial in the case of natural selection. Moreover, if you assert that the conditions have been met and this sort of evolution has occurred, then you’re certainly making a non-tautological claim about the real world.

    This is perhaps similar to what Jim Harrison said above.

  14. Richard W. wrote:

    There are two meanings of tautology:
    1. A needless repetition of an idea, statement, or word
    2. A proposition which is true by virtue of its logical form alone.

    Actually both meanings are the same. A logical tautology is a logical statement that can be shown to be nothing other than a restatement of one of the axioms of the logical system i.e. a ‘needless’ repetition. In any consistent logical system all ‘true’ statements are tautologies!

  15. Anon Anon

    Samuel Butler, in his Evolution Old and New (1879), was probably the first to make the tautological argument, describing natural selection as a “truism”, and remarking “’fittest’ has no force”.

    Interest article and comments.

  16. Eric Anderson Eric Anderson

    Dear John:

    I know this thread has long since passed, but I just stumbled across it today and was gratified to see you are still exploring the tautology issue. Many people on both sides don’t like to discuss it, because it is essentially a logical issue (and one that can turn on definitions and fine nuances), rather than a biological one, but the logical underpinnings of a theory are important and this is a key issue.

    As a younger man myself in those days you may recall that I wrote a critique of your other piece on Talk Origins, A Good Tautology is Hard to Find:

    Indeed, it was the first thing I wrote on the topic of evolution, and, like yours, contained some very initial thoughts on the matter. My humble website has long since gone extinct, but you will recall my piece was titled “A Good Tautology is Hard to Avoid.” (I think it was quoted essentially wholesale later in some google group, so probably still available via Google.)

    In any event, you sent me a very kind note afterwards, saying, in summary, that you essentially agreed with my criticism of the particular points in that piece, but that you still felt natural selection was not a tautology, “but for other reasons.”

    We’ve both written at least once more on this topic since then, so I was excited to stumble across this page today.

    Now to the question:

    From your article above, it looks like you have now decided that natural selection is indeed essentially a tautology, as noted by even several prominent evolutionists, but presumably you think it still has some value as a concept?

    I’d be curious to know your thoughts on this front. Assuming, for purposes of discussion, that natural selection is indeed a tautology, what value, if any, do you still see in the concept?

    And, on the other side, what are the risks, if any, in still using the concept in biological discussions? Is there a risk it can cloud the discussion or provide a false sense of explanation of this or that biological effect, where no substantive explanation has been provided?

    Again, I hope you may happen to check back to this old page, as I would be interested in your thoughts. I appreciated our brief prior correspondence and the professional and courteous reply you offered — a refreshing oasis in what is so often a very contentious debate.

    Best regards,

    Eric Anderson

Comments are closed.