Skip to content

Does teleology hang on in Venice?

Last updated on 22 Jun 2018

Here’s an interesting paper, which I haven’t had time to digest, but which I thought I’d better mention before it enters the fog my brain contains these days…

It’s by David Depew, one of my favourite philosophical writers on evolution (in no small part because he takes a historical approach to the topic):

Is Evolutionary Biology Infected With Invalid Teleological Reasoning? A Review Of Not By Design: Retiring Darwin’s Watchmaker By John O. Reiss, University Of California Press, 2009, in Philosophy and Theory in Biology

It’s open access right now.

Depew argues against Reiss’ claim that teleological thinking infects evolutionary biology. As I tend to agree with that contention, I have to do some considered reflection after I mark all the exams and assignments. Soon. Sometime.

For now let me say this: there is nothing wrong with using teleological language; but only when the subject is a teleological system. I can talk about the purpose of something when a purposive system has intended it, but natural selection is not purposive; it only appears to be to observers (who are). Natural selection is simply a process of thus and so a kind of mechanical causation (retentive, etc.). To assert that it is a property of the things being selected that they are “for” something, rather than a property of the investigator’s conceptual scheme, is what I call the ontological fallacy, basically to project the properties of ideas and words onto the world. Maynard Smith used to say to students “Is this [discussion] about words or the world? If it is about words, I will go, but if it is about the world, I will stay”.

Elliot Sober famously wrote of a distinction between “selection of” and “selection for”, to cover the known issue of hitchhiking (pleitropy) in the selection of genes and traits:

“Selection of” pertains to the effects of a selection process, whereas “selection for” describes its causes. To say that there is selection for a given property means that having that property causes success in survival and reproduction. But to say that a given sort of object was selected is merely to say that the result of the selection process was to increase the representation of that kind of object. … I offer the following slogan to summarize this logical point: “selection of” does not imply “selection for”. … “Selection for” is the causal concept par excellence. Selection for properties causes differences in survival and reproductive success … [The Nature of Selection (1984), p. 100.]

I quote these because I have been debating Douglas Theobald in email about whether the selection-for distinction is even necessary; it looks like Reiss also thinks it is not. I am going to have to read his book. For my money (before I get back to examining) here is the way it should be conceived.

There is no objective, real, observer-independent distinction between selection of and for. Selection occurs through causal differences, and every causal difference makes a difference. If there is a physical difference that is a trait of the organism, and it ends up spreading through the population, then it is selected. What the causal story is, can only be discerned, if at all, post hoc. Any actual causal difference may play some role, and probably does, in the selection process. Physical differences make a causal difference, and we cannot ignore them.

For example, consider Gould and Lewontin’s famous attack on the “Spandrels of San Marco“. They argue that the spandrels (pendentives, actually, but that’s verbalistic pedantry) in the spaces between the columns and the dome, which show pictures of the Gospel writers, are not there to show those pictures, but are post hoc decoration. It would be a mistake to think they are the purpose of those spaces. Here’s the image in their paper:


Taken from here.

It’s a compelling point. Or it was, until I went there, and saw this:

san marco.jpg

From here.

The entire church is there solely to display decoration. It is as obvious a display of wealth and power as it is possible to imagine. The dome, the columns, and yes, the pendentives, are all there to display these gold mosaics. What level of description you choose to give of the cathedral determines the purposiveness of the structure and its parts. Gould and Lewontin might be accused of inappropriate atomisation of the church.

Now I am not saying that some features of an organismic type are not more causally important than others. Obviously some must be. Instead I am saying that we cannot ever characterise what is, and what is not, important unless we have complete information, which we almost never do (the Grants’ study of Darwin’s finches, and the long-term study of the Soay sheep may count as sufficient evidence to make claims at least at the observable phenotypic level). And yet that caveat is almost never taken seriously. We talk unconstrainedly about “selection for” this or that, as if we had some basis for our claims apart from intuition. The reason why I object to selection for talk is that it usually if not always turns out to be talk about us and our dispositions rather than those of the organisms under study. And as the size of the potential data set increases, our intuitions are less reliable. Even with the use of statistical analyses of molecular data, we have limited warrant for our inferences if we do not already know (or intuit) what “the players” are in each selective sweep. That something has been the effect of a selective sweep is relatively (!) easy to identify. That it is the trait selected for is not. We use it out of all proportion to our knowledge.

I am something of a panadaptationist, in this sense: I think that all traits at all times are subjected to some degree of selection, and almost all traits are maintained at a high level of fitness. Noise, in the form of drift, neutrality and so on, all apply, also, to all traits at all times. At best we can say that the rates of change of a trait or gene in a population are predominantly noisy or selective at a given time. We cannot, though, say with any real warrant that some particular trait is selected for, unless we have done a major study after the event and during.

The reason this is uppermost in my mind right now, apart from my debate with Doug, is that I think this is where whatever philosophical purchase the recent Fodor and Piatelli-Palmerini book lies. Like their “Darwinian” opponents, FAPP, as I call them, consider that for a selective explanation to work you must identify the traits selected for (although I think they do real harm to that distinction in the book); only, they think you have to do this ahead of time for the explanation to be scientific. Because selection is not that sort of a theory (unlike, say, a physics theory), it is not, they argue, a scientific theory; the special pleading is obvious. But their confusion lies in thinking that we have to have a prior idea of what gets selected and what is causally significant; I think selection is a schematic explanation that is filled out in each case.

Lose the of/for distinction, and we can say well enough that there is selection and that the spread of an allele or trait in a population is selected (or noisy). Anything more is about us and our way of describing, selecting for investigation, and understanding. Now, having made that blunt claim, I must appeal to the philosopher who wrote, “there’s the bit where you say it, and the bit where you take it back”.* Of course we will continue to say that this or that trait is the main, or even the sole, cause of a selective sweep. We almost cannot help it. And it is probably harmless in most cases. This is not about what scientists do, but about what philosophical implications one can draw from it. Philosophers like to privilege aspects of organisms and ideas as being somehow “the” point; in this lies the inherent teleology. Fine, but if we then conclude that selection, which is an entirely post hoc process, gives us intentionality, meaning, and purpose, well then I agree with Reiss and not, say, Millikan or Sober. In selection, stuff happens, and afterwards we admire the results.

There’s a lot more to say about this, but I really must finish my marking before the Dean has a go at me…

This is me in Venice, to prove I am not merely conducting a thought experiment about San Marco, which is behind me on the left:

Venice 006.jpg

Photo by Jenny Webster.

  • J. L. Austin in Sense and Sensibilia, page 2.


  1. John — you write:

    Fitness … is not a physical property; it is merely a metric we apply to physical properties. A virus, a fungus allele, and a bacterium that reproduces asexually can all have exactly the same fitness and nothing else uniquely in common.

    What, in your opinion, is a physical property? A virus, fungus allele, and a bacterium can all have the same temperature, or entropy, or free energy, in principle. Or they can have the same momentum, or velocity, or mass, etc. All these are classic physical properties. So what do you mean? How is fitness different from these, aside from having a different name?

    • John S. Wilkins John S. Wilkins

      Basically, an observable (i.e., measurable) property of an object. Temperature, etc., are all measurable with some apparatus that is able to measure a physical magnitude. When you come up with a commoditomometer that measures fitness, I shall retract. 🙂

      More seriously, fitness is not a physical magnitude because it depends entirely upon the model being used for its measurement. But we can measure a physical magnitude like temperature or mass without a model of what those things are. There may not be a clear demarcation between theoretical metrics like fitness and observable metrics like temperature, but we can tell them apart most of the time, and here is one place I would say there is no ambiguity (or amphiboly).

      • we can measure a physical magnitude like temperature or mass without a model of what those things are.

        I’m unconvinced this is correct. How, for instance, can you measure negative temperature without a model for what that is?

        And mass — you can measure something’s weight, I suppose, without a model. People did for millennia. But mass? How do you estimate the mass of an object without reference to a scientific model of mass? The usual method is via the equation m=F/a, which is very theory laden. Or do you mean something different by “model”?

      • How, for instance, can you measure negative temperature without a model for what that is?

        In physics there are no negative temperatures.

  2. George Sullivan George Sullivan

    My god , all this talk of intention. Define intention as you wish and I’ll define it as I wish. In my definition the universe does intend to create all that it has created. Its nature is to create such and I insist that that nature entails the intention to create as it has. I think this makes more sense than saying the universe was created by accident and likewise the planets and man and so on. It seems to me that intention is grounded in the nature of the creature acting and that wanting and intending or desiring are about the same thing
    If something was done–there was fundamentally a desire, a wanting to do that. I don’t think one can intend and not want.
    Presuming the universe is the act of creating all this stuff–then it is its nature to do so. And SO it intends to do so. To me it is irrelevant whether or not there is some cosmic thought that says, as a person would, “I’m going to create such and such…” And I don’t think any agent–like a God— need be invoked at all.
    And I don’t agree that intention must apply to people or animals only. The moon intends to pull the oceans. The universe intends to create what it has created.

  3. The link to the “”recent Fodor and Piatelli-Palmerini book” does not work properly.

Comments are closed.