Skip to content

The false analogy between species and art

Last updated on 18 Sep 2017

Biological topics are used widely in philosophy to illustrate arcane and recondite philosophical topics,and one of the most widely used, and most abused, are species as examples of natural kinds. Kangaroos, swans, tigers, lions, cats, and of course humans are all brought in to assist our intuitions. As Umberto Eco wrote once,

The history of research into the philosophy of language is full of men (who are rational and mortal animals), bachelors (who are unmarried adult males), and tigers (though it is not clear whether we should define them as feline animals or big cats with a yellow coat and black stripes). [1999: 9]

Why this is, is obvious. We need to understand abstract concepts of science and philosophy in terms that we are easily disposed to appreciate. Humans are classifiers of their world and the things they most easily classify are other (large, moving) living things. It’s our Umwelt. But this leads to all kinds of problems. There is a thing known as folk taxonomy, in which the default cultural assumptions of a society are imposed upon the living world. One of the basic tasks of a science is to overcome the folk taxonomy and find out, or classify the world, so that the local biases and simple misunderstandings of the tribe are eliminated and the classifications denote actual things.

Still, philosophers use species as natural kind terms, when every biologist knows this fails in most if not all cases. It might not even be malign: if what you are doing is discussing the meaning of words, then English (or German, etc.) words that denote kinds of animals and plants can stand in for natural kind terms, so long as nobody is thereby misled into thinking that because there is a word, there is out there in the world a kind that answers directly to it. I wish I could say this was a truism all philosophers understood.

Actual species are messy items in the world. Philosophers of biology (that subset of philosophers who attend closely to the uses and ideas of biological sciences rather than resting content with “what everybody knows” about biology) argue at length about essentialism, in which species and other kind terms in biology are supposed to share properties that are uniquely theirs, versus individualism, a somewhat complex suite of ideas that rest upon the assumption that species and other groups in biology are historical objects that have their properties contingently. Philosophers of language try to resurrect essentialism, or treat species as the very exemplar of a natural kind, but generally this fails. The lessons are, slowly, being learned.

But there is a higher level error that can be made, and it has just been made by Christy Mag Uidhir and P.D. Magnus in a paper forthcoming in Metaphilosophy. The paper is titled “Art Concept Pluralism”, and is an argument that the very concept of art is pluralistic. An analogy is drawn, not with a species, but with the concept of species itself, a subject about which I have a few ideas of my own. Now Magnus himself drew attention to a mistake he made by a cursory reading of the literature. He “used” a “concept” of species that he called “the phenetic concept”, and says:

The PHENETIC SPECIES concept (also called morphological or typological) divides species based on organisms’ exhibited characteristics.

This is wrong. It is, indeed, as wrong as a wrong thing can be wrong. I won’t go into the details, but it is not all, or even most, of Mag Uidhir and Magnus’ doing. Types and morphology are distinct, and phenetic means something else, although not quite what Magnus says in the blog entry. He writes

… biologists and philosophers of biology use the word [phenetic] more narrowly to distinguish a specific movement: people who self-identified as employing a phenetic approach and who distinguished species just by doing statistical analyses of observable features

Umm, no. “Phenetic” means the use of multivariate character components in an analysis using clustering algorithms. Not just any statistical analysis will do, and a few characters would not make an analysis phenetic. More widely, a phenetic approach treats not species, but “Operational Taxonomic Units” which are rank-free objects right up to and including entire phyla, or down to single individuals. There is no “phenetic species concept”; that’s the entire point. And phenetic is not something one defines in relation to other approaches such as “pattern cladist” or “biological … ecological … and evolutionary approaches”, either. It is a well defined and coherent approach. If none of those other approaches existed, it would remain well defined.

I’m being picky, I know. The distinction between “character-based” and “process-based” that he makes has some purchase, although I would suggest that it is the distinction between empirical and theoretical classifications. But despite his honesty and self-correction, there is an interesting, and crucial, point in their paper that he has not corrected.

First, let me note that he is using a folk taxonomy of scientific concepts. Scientists do use concepts and words in ways that are relatively reflexive, but that doesn’t mean we should take them at their word. Like any subject of investigation, they are as likely to employ culturally biased and sociologically determined meanings as anyone else. They are a tribe. As philosophers we have to look at them closely and determine if they are clear on their ideas or not. The essentialist story I have so often railed against is just such a social construction. But that is not Magnus’ problem, that is mine.

However, he invents a concept and then crucially uses it in the argument. Here is the way he does so with Mag Uidhir. ART (concepts are capitalised) is like SPECIES, and what is true of SPECIES may be, mutatis mutandis, true of ART. We know species concepts are pluralistic, and so too may concepts of art be. Multiple concepts of SPECIES are used by biologists profitably, so we can presume this for ART. While one might dispute untrammelled pluralism works in biology, either to delimn the natural world or to aid communication, thus far an argument has been made. Then this:

Some of the concepts involve an arbitrary fineness of grain. Using the PHENETIC SPECIES concept, biologists may make species larger or smaller depending on the refinement of their observations and their need to distinguish populations from subpopulations. The PHYLOGENETIC SPECIES concept is similarly plastic. For a monist who thinks that there is a single correct partition of species, this open parameter in a SPECIES concept is a terrible embarrassment. Provided specific biological projects sufficiently constrain the scope of a SPECIES concept, the pluralist may simply accept this result.

What he seems to be saying is that we can enlarge or reduce the scope of a species kind term arbitrarily. But the arbitrariness in phenetics is the threshold at which we include or not organisms in the OTU. It depends, rather centrally, on what we know about the organisms. For bacterial species, we may use a 70%, or a 95% or even a 99% threshold of clustering and similarity. We cannot just arbitrarily say that we will pick one of these in order to redescribe or reclassify bacteria. You do that based on what works at identifying relevant strains or ecotypes and so forth. In fact bacteriologists use a “polyphasic” approach, utilising multiple lines of evidence (genetic, molecular, phenotypic, ecological, cf. Colwell 1970 and Vandamme, et al. 1996). What is arbitrary is what the natural world makes work, not the choices of the taxonomists. Iterative refinement of the thresholds depends on what nature does.

But is there an “open parameter” here anyway? Well not if you think that we can identify species in the absence of prior theoretical commitments, as I do. In other words, we may have no rank for species, as Ereshefsky (1999, 2000) argued, but that hardly licenses the view that species is entirely arbitrary, to be identified as it suits the taxonomist. We named species in the 16th century, before any definitions or theory existed, that we still think are species. That needs explanation. My answer is that we observe species, often. Species are phenomena.

The same point can be made for the so-called “phylogenetic species concept” cluster of ideas. If there are unique traits that all and only members of a species have, which we have identified, then perhaps the PSC will enable us to identify species. But most species have structure well below the species level, such as haplotype groups. We do not identify them as species (although taxonomic inflation is a serious problem in conservation, mostly to do with the over-reliance upon molecular genetic criteria). But pretty well any specialist in, say, fishes or beetles or even eucalypts will be able to say where the species are, nearly all the time.

How does this affect the argument given? ART is a human concept, based upon what humans do and think. SPECIES is not. If there are many kinds of SPECIES, it may be because the world is complex. I have previously argued that the modality of a species is an evolved property, like a trait. Such properties are historically contingent. One cannot be an unrestrained pluralist. Can one be unrestrained about ART? I cannot say. It seems to me that what counts as art has more to do with social structures, economics and functional roles than it does any shared meaning, but that is perhaps my cynicism in play. In any case, the parallel with SPECIES is fraught with problems.

Colwell, Rita R. 1970. Polyphasic taxonomy of bacteria. Culture collections of microorganisms:421–436.

Eco, Umberto. 1999. Kant and the Platypus: Essays on language and cognition. London: Vintage/Random House.

Ereshefsky, Marc. 1999. Species and the Linnean hierarchy. In Species, New interdisciplinary essays, edited by R. A. Wilson. Cambridge, MA: Bradford/MIT Press:285-305.

Ereshefsky, Marc. 2000. The poverty of Linnaean hierarchy: a philosophical study of biological taxonomy. Cambridge, UK; New York: Cambridge University Press.

Vandamme, P., B. Pot, M. Gillis, P. de Vos, K. Kersters, and J. Swings. 1996. Polyphasic taxonomy, a consensus approach to bacterial systematics. Microbiol Rev. 60 (2):407–438.


  1. So, what problem about ART are Uidhir and Magnus trying to solve? Is it, for example, that some people think Picasso’s paintings are art while others think they are trash? Or that some people think preliterate people don’t do ART while Denis Dutton said, “why of course they do!”?

  2. Interesting. Metaphors, sometimes, are bastards.

    I would like to reiterate your point about words implying the existence of kinds: more people would benefit from awareness of Whitehead’s fallacy of misplaced concreteness.

    On pluralism – and I’m far from expert in species, so not sure if there’s a relevance here – I think pluralism is useful when, in abstracting something, we by necessity pare away some detail and generalise, because generalising is useful (is my pragmatism showing?).

    As a result, there might be multiple ways of abstracting, and each may be useful in different ways. Each approach might introduce a *distortion* – it shaves off something to arrive at a useful abstraction. But what you don’t want to do is introduce *error*, such as by misrepresenting some feature of whatever you’re abstracting.

    Now, to the (bastard?) metaphor: cartography. I’m a pluralist about map projections (as you may have read before on Ockham’s Beard). Each projection introduces distortions, but you tolerate them to find a projection that is useful for your purpose. Yet they all link back to the 3D globe. And errors, such as placing Australia next to Belgium, are identifiable non-subjectively, and are not tolerated.

    As for art, I don’t know that there’s a 3D globe to attach the plurality of concepts to. Sure, there are interesting facts about neuroscience that suggest aesthetics is at least somewhat non-subjective, but *concepts* of art? Seems to me relativism is the way, as there’s no non-contentious way to arbitrate between the various concepts – unless you tie art to neuroscience, but I can imagine someone plausibly calling that link contentious.

    So yeah, what you said. I think.

    • John S. Wilkins John S. Wilkins

      I have a pragmatist slip that shows from time to time, myself. But I think that you may have misunderstood why I am a limited pluralist – it is not that there are many intensional definitions of SPECIES, all of which have some underlying theoretical justification and interpretation that maps onto various kinds of things in the biological world. There are no intensional definitions needed at all. SPECIES is a term that talks about phenomenal clustering of organisms that reproduce like each other. The intensional definitions apply because there is a formal explanation of that fact. And since there are many ways organisms can reproductively cluster, including asexual ways, ecologically maintained ways, and so forth, there are many intensional definitions of a single term.

      In my badly written and incoherent 2003 paper I argued that since the physical traits that causally maintain lineages as distinct from their closest relatives are themselves evolved, so too is the modality of being a species an evolved trait, and we ought not expect that they will all share some universal property (apart from clustering).

      How that applies to ART, I cannot say. It seems to me that ART is entirely functionally circumscribed (see my reply to PD below), and nothing intrinsic to the ART species maintains it as a category, whereas the SPECIES species have such evolved properties.

  3. John Wilkins John Wilkins

    I’ve done that myself, but that doesn’t stop me from taking advantage of your mistake here.

  4. Ack! That phantom comment from John Wilkins was me cutting and pasting into the wrong window, trying to add the comment he was forced to e-mail me to my own blog. Sometimes I don’t think I should be allowed on the internet.

  5. OK, let’s see if I can muster up a comment that has real content in it.

    The central thrust of your reply, it seems to me, is that you are a realist about species (you think they are phenomena which we observe) but not about art (because it depends on what humans do and think). So you think that multiple SPECIES concepts could reflect complexity of the phenomena, but that multiple ART concepts could not.

    I see the cases as more analogous. With art, just as with species, there are many cases about which all practitioners can agree. Despite the points of agreement, there are other cases which are less clear. How these boundary cases are resolved depends on how we specify the category. Both the biological world and the history of art are complicated, and different specifications will be useful for different purposes.

    Bill: Just to be clear, my coauthor’s last name is Mag Uidhir. And it’s more the latter kind of case we have in mind.

    • John S. Wilkins John S. Wilkins

      The real difference between the concepts is that overall, art is constructed by social role, while species are forced upon us by empirical facts about the world (in my view – I am indeed a species realist, although not a rank realist). Now for a philosopher of language, sure, the social world and the natural world are similar; for a philosopher of biology they couldn’t be much less similar. Species exist if not language users exist – I do not think the same can be said about art.

      In Wittgensteinian terms, the reference for ART is specified by how the language community uses the terms. The reference for SPECIES is not. But there’s a complex and probably unconvincing argument behind that assertion.

      • If we used language differently, then the word ‘art’ would mean something else. The same is true of every word. This includes ‘species’, even though species are real. (We agree about realism for species but not for the rank.)
        If the biological parts of the world were different, then species would be different than they are. The same is true of every category for the relevant parts of the world mutatis mutandis. This includes art, which would be different if human practices were different.
        Of course, there is this disanalogy: Before people came on the scene, there were already species. There was no art before people came on the scene. Yet it does not follow from this that art is whatever we say it is.

  6. I’d like to note that in the phenetic context as John uses (above), a phenetic OTU is often just a single specimen, or multiple specimens, and can be used to test the robusticity of a series of single specimens versus an acknowledged species, or formally named organism. In this case, I can take a sample among modern dogs by using specific specimens of (individual) animals and compare them across dog breeds for which a generalized form, and against a validated array of species (by taxonomic agreement, as recognized), and thus use three different forms of OTU as equivalent phenetic subjects in a phenetic analysis.

  7. Jeb Jeb

    Ive been looking at Emanuel Fremiet’s plaster statue “Gorilla Troglodytes (sav) From Gabon”, 1887 and other related work.

    I thought P.D Magnus may enjoy it, its an interesting blend of Art, Folklore and taxonomy based on folktale but stimulated by 19th century discoveries in taxonomy and biology.

    Fermiets theme is drawn from folktale and folk taxonomy a point the paper does not really explore in full depth.

      • sbej sbej

        Good luck taking the research forward. Its an interesting subject despite the difficulties.

  8. John Harshman John Harshman

    I would like to know just what you think species are. Above, you give this definition: “SPECIES is a term that talks about phenomenal clustering of organisms that reproduce like each other.” Now, the only part of that that immediately means anything to me is “clustering of organisms”. Can you explain the rest in a way that infuses “phenomenal” and “that reproduce like each other” with meaning?

    Now I would say that, if there really is such a thing, the universal meaning of SPECIES, uniting all concepts, would be something like “a minimal cluster of organisms separated by some noticeable gap from other such clusters”. This has many of the same problems as other such universal definitions. What do we do about structure within species? Perhaps there is some minimum threshold of gap extent. What about hybrid zones and such that render gaps tenuous? What about ecotypes, sexual dimorphism, and the like? Like I said, many of the same problems. What I would merely like to emphasize is the apparent importance of the gap in defining the cluster. Species are clearly distinct from others in some recognizable way.

    • Josh Harsham writes:

      What I would merely like to emphasize is the apparent importance of the gap in defining the cluster.

      I think this is the problem. There is no “gap;” organisms during descent don’t stop for a breather for some perspective to subjectify their existence, and it seems equally problematic to remove the form of continuous descent as a contradiction even to “lineage segments,” one of the species definitions most in keeping with the definition befitting clustering as “kinds.”

      Types and kinds are good on a philosophical level, but lend no weight to what is a continuosuly undifferentiated (at high resolution) progression of slight successive changes in form.

      John, your initial premise is correct, I think: Species are not art … but taxonomy, and the attempts to validate it, is art. Taxonomy relies on the innate existence of the “species,” “type” and/or “kind” to be useful, but then it is no different from labeling paintings in the Louvre, splitting rooms by subject matter or medium or artist, etc.. Species only become art when species and taxonomy are conflated, then the distinction is blurred or (more likely) ignored — as I am increasingly aware is a validation of the nomenclature from which a person may “leave his mark” on the world, or an expression of his ego of perception (validating his view of species.

      • John Harshman John Harshman

        And that’s why the biological species concept works just fine (usually) when we talk about the species at one time and place, but less well when we try to extend it over evolutionary time (especially) and continental space.

        Fortunately, though species are connected continuously through history, continued divergence erases those connections, leaving, in the present, those important gaps that let us form clusters. The gaps really are there.

        No. Taxonomy is not art. It’s science, despite there being some places where the edges are fuzzy.

    • John S. Wilkins John S. Wilkins

      John, you have asked me this before and I recall I gave an answer. I think “species” is a term given to clusters of organisms that can be regarded as sharing reproductive traits and ecological traits due to some similarity of biological properties. This is broad and vague because, as I have argued in print, species are not all of a kind (like vehicles or works of art, just to undercut my argument above).

      The “gaps” here are often honored in the breach than the observance. Hybridisation is common between species, that are species on many other criteria, so reproductive, or morphological or genetic, gaps are not necessary to make species species. I think still that we decide what counts in a particular case by noting what makes things species in that particular group of organism, and testing this case against that. Denisova might not be a new species where Neandertals are because we know fairly well from our primatology what counts as a primate species; it doesn’t therefore follow that this will work in ducks or roses.

      When you do have unambiguous gaps, and those gaps are exemplifications of being that group’s kind of species, all is well. When you have problem cases, and the local criteria do not help, then you cannot say. When you have problem cases and the local criteria do help you, then you no longer have a problem, and that is all one can say about this. But gaps exist within species often enough, without undercutting the subspecific nature of the variants, and so a simple gap is not enough. It needs to be the right sort of gap for that kind of organism.

  9. Jim Thomerson Jim Thomerson

    An anecdote about species. The females of the killifish genus Rachovia are plain little fishes which all look about the same. We made a mixed collection of R. hummelinki and R. brevis to use in constructing a DNA phylogeny. The DNA guy, who knows fishes pretty well, was watching me sort them out of the aquarium and doubting my success in doing so. Now, I could not tell you what my sorting criteria were, but after doing the DNA, my friend congratulated me on getting them all identified correctly. whew!
    It is a strong compliment to say a taxonomist has a good eye.
    The business of knowing things are different, and not being able to say why, is not uncommon among taxonomists.

    • John S. Wilkins John S. Wilkins

      The single most sophisticated classifier system that exists does so between the ears of a suitably informed and experienced taxonomist. That system cannot be beaten, IMO. It’s a complex kind of neo-Hebbian feed forward neural network, with inbuilt dispositions.

Comments are closed.