Skip to content

Species, phenomena and data

Just lately I have been trying to support my belief that species are not units of biological theory, but phenomena that call for explanation. Several things have followed from this:

  • Species turn out on this view not to be causal entities, but rather epiphenomena of causal processes at the individual and populational levels. Species don’t cause anything (apart from thoughts in the minds of taxonomists)
  • If species are phenomenal objects, what is a phenomenal object? Is it an ordinary object that
    • Is refined into a theoretical entity under something like entity realism?
    • Is divided into things that are theoretical entities once we get the right theories, leading to theoretical pluralism (or explanatory pluralism, depending how you parse that)?
    • Is deflated into a non-entity that is replaced with theoretical entities?
  • And just what is a phenomenon anyway? How is it different from data (a distinction that Bogen and Woodward made, and which is implicit I think in van Fraassen’s new book)? Is a phenomenon just an aggregate of data or is data one end of a continuum of which phenomena are the other end?
  • Can phenomena be untheoretical?

Some of this stuff will be familiar to those who are recently educated in the philosophy of science, but a lot of it has passed me by since I did philosophy of science in the late 80s. So I would appreciate any help anyone can give me.


  1. I like Sober and Shapiro’s “Epiphenomenalism – the Do’s and the Don’ts” for some very similar issues in the causal structure of natural selection. In particular, they defend an account of epiphenomenalism and the relationship between macro- and micro-causation that I find really useful in understanding biological issues.

  2. Perplexed in Peoria Perplexed in Peoria

    I am just an amateur, but it seems to me that the source of your problems is a commitment to the false dichotomy that “species” must be either a unit of biological theory (and hence a causal entity) or else phenomena that call for explanation. Why can’t they be both? Or rather, one or the other, depending on which biological theory you are talking about.

    I see terms like ‘causal entity’ and ‘phenomenon’, not as elements of some fundamental metaphysical true ontological system, but rather as the names of roles in an account of the nature of theory.

    You seem to pay lip service to theory pluralism, but to pay homage to the (as yet unknown) “one theory to rule them all (and in the brightness reduce them)”. A true pluralist, while admitting the existence of a “reduces to” relation between theories, remains agnostic regarding the existence of a unified fundamental theory.

    I second Pence’s endorsement of the Sober and Shapiro paper. Very nice.

  3. It seems strange (to me, at least), to say that species are phenomena.

    As a mathematician, I would probably think of a species of an equivalence class (all of the thing that are equivalent with respect to some characteristic).

    If species are phenomenal objects, what is a phenomenal object? Is it an ordinary object that …

    For that matter, what’s an ordinary object? It seems to me that a rock or a table are very different kinds of things from a rabbit or a cat. The rock and the table are made of atoms. But for the rabbit or cat, the atoms that are there today will be gone in a few months, replaced by different atoms.

    And just what is a phenomenon anyway?

    A good question. Whatever it is, it surely is not the same as data.

    I sometimes think about Conway’s Game of Life as an illustration. In some sense, what really exists are the pixels that are turning on and off. The gliders and glider guns are appearances, and presumably can be called phenomena. But, I suppose with that way of thinking, the rock and table are objects while the cat and the rabbit are phenomena.

    I think we sometimes get into trouble when we over-analyze and try to be too formalistic. This may be one of those times.

  4. I like the equivalence class parallel. It explains why there is no single species concept (because there is no single equivalence relation).

    You can go a bit further and say that all taxonomic units are equivalence classes. Species are the ones we like to label with binomials.

    • It follows from their being classes that taxa are equivalence classes, although that doesn’t mean what philosophers of biology have understood it to mean. It simply means that one may, salva veritate, substitute one member of a statement involving the class for another. This, of course, depends crucially on the grain of resolution. I can replace statements involving John Wilkins with a statement involving Mike Keesey and get the same truth conditions with respect to typical criteria for hominidity, but not with respect to the law or marriage status (your accountant and wife may object!). Interchangeability here is crucially dependent upon context. You may play a similar anatomical, ecological and genomic role to me without thereby sharing all “essential” properties.

      I am not being a rampant pluralist here – we know what sorts of criteria serve to identify members of a mammalian species already, so that when I ask if Obama and Wilkins are members of it, I am not able to choose just any arbitrary criteria. Neither does it imply that all members of a species are “identical” in any sense other than, “for the purposes of identifying species”, so the equivalence class point doesn’t resolve anything.

  5. Gary Nelson Gary Nelson

    Your shift in thought about species prompts me to ask if you see species as different from taxa generally.

    • Yes and no. Species are the basis for taxonomy/systematics in general, IMO, but they are not formed from shared homologies necessarily, which I gather higher taxa are. For instance, all higher taxa are monophyletic but species need not be (respeciation and polyphyletic origins mandate that claim).

      But inferentially, a species is just like a higher taxon in that it will comprise members that have a consensus of properties that indicates the members should be grouped together and which licenses inferences about as-yet unexamined members. So if I have examined a number of Mus musculus and found that, developmental abnormalities aside, all members have a given trait, it is very likely a novel member will too, just as inferences about higher taxa can be inductive based on knowledge of some members of the clade, as you noted many years ago.

      Or is there something else you were asking that I missed?

  6. John S. Wilkins:

    Interchangeability here is crucially dependent upon context.

    Of course; that’s how I took it. Any taxonomic unit (species included) is dependent on context.

  7. Gary Nelson Gary Nelson

    Species and taxa generally. My problem is that what is called a species today might be found tomorrow to be a group of species. And it might survive as a group (e.g., genus or subgenus). And the evidence indicating that it a “something” might not have changed in the process. Similarly, what is called a group of species today might be found tomorrow to be just a single one.

    • Well of course both have occurred. This doesn’t affect their being phenomena, but the salience of the phenomenal criteria may have shifted (I’m thinking of Murray Littlejohn’s frogs).

      The theory-dependence claim is that we need to gather the data to have a phenomenon, and that relies upon theory. This is always true in one rather weak sense (or else we’d need science to see the table in front of us), but it doesn’t follow that the theory you need to account for a phenomenon (say a theory of microbes) is the theory you use to see the phenomenon (which might be a theory of optics, microscopy, or chemistry, but not the theory of the things being observed).

      I think we always have assays, such as our visual perception or our cloud chambers. How theoretical the assays are depends a lot on us, but I don’t think we need much theory to see many species. Littlejohn’s frogs initially needed sonography to identify, but later we could see things like mating behaviours to back up the splitting of the leopard frog (IIRC you told me about this example).

      Higher taxa are also phenomena. The real question is how theory-dependent they are. Can you only see a eukaryotic kingdom, for example, because you have Techniques? For many such taxa, the answer is ambiguous. Some you can place because they are salient phenomena to any trained observer. Others you need things like gram-negative staining or the ability to identify biochemical structures of cells. So my question is when something is a phenomenon and when it is the result of our assays and techniques, and when these are the same and different.

  8. Martin Brazeau Martin Brazeau

    I consider the concept of a species (and also characters) to be a primitive, which is perhaps why the equivalence class example seems to work. That’s just my two cents.

  9. DiscoveredJoys DiscoveredJoys

    I’d like to propose the Abandoned World Test. If human beings all vanished would ‘species’ still exist?

    Personally I think the answer is no. The remaining animals might divide the world with emotions which tag ‘stuff I can mate with or compete for mates with’ and ‘stuff I can eat’ and ‘stuff that is dangerous’, but I suspect that any finer categorisation would be unavailable.

    In which case ‘species’ are a phenomenon – a phenomenon of patterns in our brains, not anywhere else. How we define ‘species’ is therefore a matter of how we choose to mentally represent the stuff in the world to ourselves and others.

  10. AK AK

    This may well already be obvious to everybody, but a class of organisms that can be grouped together by eye (or other senses) and distinguished from other classes will be a phenomenon involved in the behavior of animals living in the same (ecological) community. This will, in turn, apply selective pressure of some sort to the population and its members.

    Leaving aside (as obvious) such things as mimicry and sexual selection, consider the targeting of members of a herd that are “different” in some respect by predators such as wolves that engage in long pursuit. This aids the predator in continuing pursuit of one target until it gets tired, but also produces selective pressure for members to look similar.

    Human taxonomy of “species” is in many respects just following the process(es) used by mammals and their ancestors probably going back to the earliest vertebrates, if not before.

  11. If only we had better access to non-human taxonomies….

  12. Gary Nelson Gary Nelson

    Species and taxa generally. An example of relevant discovery is the paper entitled “Deep-sea mystery solved: astonishing larval transformations and extreme sexual dimorphism unite three fish families. Biology Letters (23 April 2009, 5(2):235-239), by G David Johnson et al (6 coauthors), wherein one reads:
    “…we show that fishes currently assigned to three families with greatly differing morphologies, Mirapinnidae (tapetails), Megalomycteridae (bignose fishes) and Cetomimidae (whalefishes), are larvae, males and females, repectively, of a single family Cetomimidae).

  13. John Harshman John Harshman

    Papers like this are why I love systematics. This example will now take the place of Eclectus parrots in my mental files. Can’t wait to bring it up to a paleontologist.

  14. rob wilson rob wilson

    are individual organisms phenomena, in the same sense?

  15. John S. Wilkins John S. Wilkins

    Yes, that had occurred to me also, and possibly. But organisms of some kinds (oops) seem to be required by some theory – due I think to Weismannian sequestration for genetics and by ecological agency (your purview!) . In short, organisms are causal objects. Species aren’t, and yet we seem to have to resort to them persistently.

  16. John Harshman John Harshman

    For the record, I don’t at all understand what you’re saying. For example, what is a unit of biological theory? How does it differ from a non-unit-of-biological-theory?

  17. Just a quick technical point, I assume Neil means “equivalence class” in the mathematical sense in which case John’s comment “it follows from their being classes that taxa are equivalence classes” is false. If you add the claim that each organism that has ever lived is in exactly one species, then yes. But no need for this and actually, it is pretty hard to get anything like this that is reasonable. For attempts, see some papers by Kornet e.g. Kornet DJ, Metz JAJ, Schellinx HAJM. 1995. Internodons as equivalence classes in genealogical networks: building- blocks for a rigorous species concept — or Samadi S, Barberousse A. 2006. The tree, the network, and the species. Biological Journal of the Linnean Society 89: 509–521. [[but see my criticism: ]]

    • So far as I know, an equivalence class is a class in which any member is equivalent to any other in some respect. In virtue of drawing a class, you are saying there is some such property or character.

      Kornet’s argument is, IMO, not a matter of drawing classes. Nodes identify groups, not classes, and I’m not keen on her approach. But we can discuss this sometime.

  18. Your definition could be equivalent to the formal one with some massaging. The standard understanding of an equivalence class is that it is one member of a partition of the space. So take all the organisms, divide them into exhaustive and exclusive groups by whatever criteria you feel like and each will be an equivalence class. The natural way to do it is via an equivalence relation (i.e. one that is reflexive, symmetric, and transitive). Dividing up shoe-wearing humans by “wears the same shoe size as” will yield equivalence classes, taking each person and forming a class of people “within 1 cm of height as” will not yield equivalence classes (since it is not transitive). Transitivity (especially over time) is the biggest worry with species. For example, x is conspecific to y iff x is part of a population that can interbreed with y’s population is not an equivalence relation unless you say much more and warp the definition beyond recognition (the population bit is there so that it has a remote chance of success). And I agree Kornet’s approach is problematic. But as far as I know, it is the only thing remotely hopeful that actually has a chance of conspecificity being an equivalence relation.

  19. I might as well point out that my own species definitions given in and in take contemporaneous organisms under current consideration and divide them up into equivalence classes (one based on organism level reproduction, the other based on genetic history). Neither would work well at all if we included all current organisms (like hybrids) and both collapse to triviality if we include ancestral organisms.

Comments are closed.