Skip to content

Notes on novelty 6: Levels of description

Last updated on 21 Jun 2018

Notes on Novelty series:
1. Introduction
2. Historical considerations – before and after evolution
3: The meaning of evolutionary novelty
4: Examples – the beetle’s horns and the turtle’s shell
5: Evolutionary radiations and individuation
6: Levels of description
7: Surprise!
8: Conclusion – Post evo-devo

And having come to know that it is, we inquire what it is [Aristotle, Posterior Analytics, II.1.89b34–35, translated in Lennox 2006: 296]

Consider this diagram:

Beetle description

Stag beetle images from FCIT, royalty free.

Every description of the phenotype of the beetle has some grain of resolution, which is to say that some things are described while others are not. this may be because some grain of resolution is ignored due to it being irrelevant to the purpose of the description, or it may be because that grain is not yet known or understood. Traditionally, organisms are treated as what Sober once called the “benchmark entity” (Sober 1984: 280, 317), the standard grain of resolution. Sometimes, however, that grain is either too gross or too fine for the purposes of describing what is occurring. The notion of a “superorganism” (Hölldobler and Wilson 2009) is a case in point: the appropriate grain of description given what we know of insect colonies is the colony, not the organism.

However, sometimes (as is the case with Wilson and his colleagues), the need to describe at a certain grain is taken to imply some absolute “level” or “rank”; that is, to imply an ontological and metaphysical scale. This is a case of “Descartes before the horse”, taking semantic and conceptual properties to imply real world properties of the things being described. It is a common philosopher’s error, but equally a common biologist’s error, and in particular a common error of evolutionary systematists.

Let us generalise this a bit. Description of biological systems and facts is like this:

Levels of description

What counts as “observed phenomena” depends crucially on the instruments and assays used to observe. The naked eye when untrained has certain dispositions to observe, for example, organisms, but a trained eye can see traits, characters or even entire ecological objects; and having a grain of description at one level means the observer can decompose the observed phenomena into parts, or compose the parts (including organisms) into larger encompassing wholes, depending upon the needs of the describer.

If an explanation at a descriptor grain serves our purposes, then we can rest there. So, for example, if a representation of the spread of a gross trait can be cast in terms of organisms and their interactions with the rest of the world, then we do not need to go deeper (unless our purposes are reductive). Taking an organism grain resolution effectively makes the lifecycle of the organism type our focus, and so we have to decompose organisms into developmental parts and stages. But if we cannot explain what the causes of the parts are at that grain of resolution, we will then decompose the descriptions and accounts to finer grain descriptors (such as genes).

The mistake is to think there is a privileged grain or mode of description. If we understand that description is context, interest, and purpose relative, then a failure to explain something like an evolutionary sequence at one level of description is merely an invitation to change the grain, by composition or decomposition, until we find something promising as an avenue of explanation. In the next post I will discuss what make a grain satisfactory, and ask again: did Darwin give us the explanations of novelty?

Hölldobler, Bert, and Edward O. Wilson. 2009. The superorganism: the beauty, elegance, and strangeness of insect societies. New York, London: W.W. Norton.

Lennox, James G. 2006. Aristotle’s Biology and Aristotle’s Philosophy. In A Companion to Ancient Philosophy, edited by M. L. Gill and P. Pellegrin. Malden MA, Oxford UK, Carlton Australia: Wiley-Blackwell:292-314.

Sober, Elliott. 1984. The nature of selection: evolutionary theory in philosophical focus. Cambridge, Mass.: MIT Press.


  1. ckc (not kc) ckc (not kc)

    The mistake is to think there is a privileged grain or mode of description.

    …of course, while there may not be an inherently privileged grain or mode, there are those that have become “privileged” for reasons of historical accident, vagaries of fashion, or economic (commerce, human health) utility – and sometimes complicated combinations of these factors. “Changing the grain” can be a very political (in the broad sense) move.

    • John S. Wilkins John S. Wilkins

      Indeed, and it is the politics of such beliefs that inspired this series.

  2. William Morris William Morris

    Hi, William Morris here,
    I should say that what is novel is a subdivision of what is the same and what is different. Language with its flexible term boundaries enables us to argue that there is nothing new, since we can place anything into a category and assert that the category is universal and within it all is the same; all is the universe, or this creature is nothing new, it belongs in the insect category and so is the same as all insects.
    What is same and what is different and where and when it is so —-people differ.
    That each “grain” of description describes the same thing, is optional. Just as it is optional to say that water is the same as H20. Hydrogen and oxygen, I can argue, look nothing like water, so they are not the same. Even if shooting electricity through water produces hydrogen and oxygen, it is not necessary to conclude that therefore water is nothing but H2O, one may say simply that one thing is transformed into another, rather than that they are identical.
    SO too one may insist that parts ununited are not at all the same thing as parts united and rather the parts are transformed into something else entirely, and very unrelated, when the parts are united. A Ford truck is not a Ford truck if all there is is some kind of parts, nor is a beetle.
    The difference is patent and one can claim the difference is too great to consider them the same at all.
    Of course, one may reply that wholes come from parts and parts come from wholes. Seems to me like a chicken and egg stand-off.
    One may further argue that it is an odd thing to claim there is one object for all ” grains” of description the of the beetle or the Ford—since the differences are so great that you could not recognize the beetle reduced to individual molecules nor the Ford truck reduced to individual atoms.
    So why call them the same? One could argue that to call them identical is patently false. Instead one could say that they are related— which doesn’t say much since one could say all things are related as members of the universe.
    On the other hand, one might assert that z, y and x descriptons or “grains” of description of the beetle are really just one long description of one object. But if so, what is the object?
    If the “grains” are equally the object but the” grains” are separate things and not the same, yet each “grain” is supposed to refer to the same object—-then it seems there is no object to refer to—only the “grains” .
    It could be asserted that the visual of the beetle is the object and the other descriptions or “grains” or observations are just that visual modified or in a different context or some such .
    But without this move the object would seem to be just a grouping of separate descriptions or just the “grains” themselves.
    If we say that the beetle is x,y and z descriptions , including the visual aspect and they all describe an object, that is, they are all just aspects of one object—what object would that be? What could you point to that is not just one of its descriptions , its “grains”?
    One could argue there is nothing to point to at all . Seems entirely imaginary.
    The crux is that, in this view, some “grain” of the thing must be chosen as the object , and the rest of the “grains” as referring to the object, or there is no object at all—either of sense or of mind.
    So there must be some dominant ranking “grain” which is the object, there must be one “grain” at least that is more than equal to the other “grains” .
    And ultimately one could argue that where sense ends and mind begins is arbitrary and the distinction between mind and sense is not a necessity.
    If I say that there is an object that all the possible “grains” refer to and that object is not one of the “grains” , again, what the hell object
    are we talking about ? It seems none. All we have is the “grains” .
    But it could be replied that this is just like the forest and the trees. One person insists that there is no such thing as a forest, only single trees and forest doesn’t even qualify as an idea or concept —but is entirely empty—just a sound .
    But the rejoinder could be that there must be some kind of object there–either mental or sensorial, since one can focus on it as an object of attention .
    And the reply to this is may be that there is no such attention, and that there is such attention , is just an idea.
    And the reply to this is that “I have an intuition that there is such attention and you don’t” or some such.
    And there it will stand.
    T here is more than a little of the old nominalist versus realist position in this issue as well as the mind/sense, material/mental split.
    Words are like greased lightning, don’t you think?

    • An old joke has it that an F15 is not a collection of parts flying in formation, and that is true. But explaining why F15s fly is a matter of explaining the causal contribution of the parts to the functionality of the whole (likewise water is just the same as its constituent elements, although no actual water exists without something other than hydrogen and oxygen, usually salts and silicates). However I am not arguing for reductionism here, or rather not necessary reductionism. I am saying that the explanation finds the appropriate level, or as I call it, grain of resolution.

      The question that I am responding to is whether Darwinian evolution explains novel features, and some say that it cannot (and so needs to be supplemented with systems theory, for example). I think the difficulty is a matter of description; that is, one of our own making, and that is my point. Stay with me until the end; there are only two posts to go.

  3. For a long time I’ve been unhappy with the idea of “levels”, including levels of description. It seems to me that much talk of “levels” derives from a common conceptual confusion – a mix-up between the reduction of a thing to its parts and the reduction of one theory by another theory. Theories are descriptions, but they involve the adoption of explanatory “stances” – which might or might not refer to gaols or ends, and which might or might not refer to representations or symbol-like entities. In Daniel Dennett’s terminology, we might adopt the ‘physical’, ‘design’ or ‘intentional’ stance. Because there is no counterpart to a goal or end in the physical stance, and no counterpart to a representation or symbol in either of the other two stances, there are sharp discontinuities between them. Theories that employ a given explanatory stance are not reducible – in the sense of inter-theoretic reduction – to theories that employ a different explanatory stance.

    For example, the language we use when talking about organs is often “functional”, that is, it refers to the “purpose” organs serve in the organism. We adopt the design stance when we use this sort of description. Or again, the language we use when talking about genes is often “representational”, that is, it refers to what genes “code for” in the phenotype. We adopt a sort of representational stance when we use this sort of description.

    Now it seems to me that discontinuities between these types of description (which result from the impossibility of inter-theoretic reduction) do not coincide with differences in size of parts, or differences in grain of resolution. So I would say “scale” isn’t the real issue here. Functionally-described organs are smaller than organisms, and representationally-described genes are smaller than both, but there is no corresponding “scale” for the different explanatory stances and their corresponding descriptions.

    Even wholly within the physical stance (which does not mention goals or representations of any kind) there are discontinuities between different types of description resulting from less-than-ideally-smooth inter-theoretic reduction. For example, phenomenological thermodynamics is reduced by – but strictly speaking is inconsistent with – statistical mechanics. These types of description usually apply to bigger phenomena and their smaller parts, but the lack of smoothness is the result of a difference in understanding rather than differences in scale.

    In short, I am very sympathetic to your overall project of explaining novelty in terms of differences in types of description, but would stress the ubiquity of differences in explanatory stance and differences in understanding rather than differences in scale.

    • I do address the question of functionality in part 7, but let me say this for now.

      This is not about reduction, but about grain of explanation. This often means reduction questions come into play, but it need not, and I think that a lot of the claims made by holists have more to do with the explanatory project and stance they are undertaking than it does about whether there actually is some privileged grain of description. If your explanatory project is set at some grain, then of course that grain is privileged relative to that explanatory project or stance. In this way it becomes rather question begging.

      The target I have here is those who say that we cannot explain novel traits because they are without precursor. In every case adduced as an example of this, it turns out there are finer resolution homologs and explanations. It may be, however, that coarser grain descriptions are explanatory where finer grain ones are not – for example, the so called phylotypic stage of development, which I gather has a multiplicity of underlying causes and genetic regulations. However, as a brain-dead reductionist myself, I know which way my predilections run. I’m just not relying on it in this series of posts, that’s all.

Comments are closed.