The structuralist heresy

Larry Moran has heard the words of Michael Denton, and has come away with a creationist interpretation of structuralism. I have to explain to Larry that Denton, as you might expect of a creationist, is distorting the whole idea. Here’s the Denton/Intelligent Design creationism version of structuralist theory.

As Denton says, the basic idea is that the form (structure) of modern organisms is a property of the laws of physics and chemistry and not something that evolution discovered. He would argue that if you replay the tape of life you will always get species that look pretty much like the species we see today because the basic forms (Baupläne) are the inevitable consequences of the underlying physics.

Say what? Look, I’m a developmental biologist; I was baptized in the Stygian stream of structuralism by D’Arcy Wentworth Thompson, I reacted and diffused with Alan Turing, I danced disco by the light of the Belousov–Zhabotinsky reaction, and no, that is not the structuralism I have studied. There is a grain of truth to it, in that structuralism does imply that there are physical/chemical constraints on form, but only the extremists would suggest that that means life on Mars would evolve to look like life on earth. That overlooks the fact that structuralists are thoroughly familiar with the diversity of life on this one planet, and since those physical laws can generate both mushrooms and monkeys, it’s clear that there is some room for exploration of form.

I can see, though, why creationists would gravitate towards structuralism. Here’s Rudy Raff in The Shape of Life: Genes, Development, and the Evolution of Animal Form (a most excellent book on evolution and development, I recommend it highly), where he contrasts it with modern molecular genetic approaches.

One school of developmental biologists derides this gene-focused program as epiphenomenological. These “structuralists” envisage morphology as arising from the properties of physical rules that generate structure in developing systems. This is a long-standing perspective on development that arises from pre-Darwinian idealistic morphology and continued to be held by anti-Darwinists. J. Bell Pettigrew published a wonderfully illustrated three-volume treatise in 1908 entitled Design in Nature. Unlike modern structuralists, Pettigrew was arguing on behalf of “First Cause and Design.” Pettigrew based his argument for common rules of form in nature, such as the spirals present in nonliving and living structures. Thus, “the dendrites of minerals and metals, the frost pictures on window-panes and pavements in winter, the lightning flash, &c., are amazingly like the arborescent forms seen in the branches and leaves of plants and trees, the division and subdivision of blood-vessels and lymphatics, the branching of nerve cells…” and so on.

The core idea in structuralism is the importance of phenotypic interactions, rather than genetic control. Why do isolated cells tend to round up into spherical forms? It’s not because there are genes instructing them to do so, but because that’s a form that minimizes energy. Why do they flatten when they contact an adhesive surface? For the same reason. You don’t need to even mention genes. (Except, of course, that the adhesivity of certain surfaces is going to vary with the expression of specific cell surface molecules).

No modern structuralist would deny the importance of genetic regulation, but a lot of the old ones did. D’Arcy Wentworth Thompson pretty much denied this new-fangled genetics stuff any relevance at all — it’s all mathematical interactions all the way down — but let’s cut him some slack. He was schooled in biology before the rediscovery of Mendel, and published On Growth and Form in 1917, when biologists were in the middle of a brawl over how genetics worked at all, and he was deep into his alternative model for the inheritance of form.

There was, after Thompson, a long period where many were seduced by the elegance of structuralism. It offered an explanation for how simple processes could generate complex and diverse patterns. Look at snowflakes, for instance: simple rules, simple molecules, all assembling into complex symmetrical patterns. If mere water can do that, then maybe there’s something equally simple underlying us…a kind of seed crystal for a human being.

One example for how people were thinking: many animals are segmented, that is, they have this repeating pattern of body elements reiterated sequentially along the length of the animal. Maybe, just maybe, there’s a molecule, a morphogen, that is expressed in an oscillating pattern, a kind of standing wave of simple molecule X that would explain the whole morphology. Wouldn’t that be cool? Turing’s reaction-diffusion model showed one way you could spontaneously generate a repeating pattern of stripes with just two molecules, and that was temptingly elegant. I’ve got a small collection of papers from the 1950s that are just that, theoretical explanations for how the whole insect body plan could be set up with a source at one end, a sink at the other, and a couple of simple, unidentified molecules diffusing and interacting chemically to make the pattern, with minimal genetic interaction.

These models all got blown to flinders as the actual mechanisms emerged: it was all differential gene expression controlled by intricate, Rube-Goldbergian regulatory logic. I’d argue that there are still vestiges of physical/chemical phenotypic interactions in the importance of epistatic interactions between genes products and genes, but it’s impossible to deny the centrality of genetic control any more.

I have another complaint about Larry’s treatment of structuralism. This is just plain wrong.

One of the problems with structuralist explanations is that they are very animal-centric.

No, they’re not. The most persuasive examples focus on single cells, which clearly have structure and form, but you can’t fall back on patterns of differential gene regulation to explain them. One of the late Brian Goodwin’s favorite examples was Acetabularia. Here’s what it looks like:


The mind-blowing thing about it is that that slender mushroom-shaped thing is a single algal cell. How does that work? You’re not going to be able to address it by just looking at what genes are present in the nucleus, down in the rhizoid. It’s going to demand a structuralist approach to puzzle out what kinds of interactions are going on inside the cell — there has to be some kind of patterning information being propagated through the cytoplasm. Reaction-diffusion, maybe? Some kind of radial streaming to make the cap?

But of course there is some kind of genetic control. There are also different species of Acetabularia, with different morphologies. One cool experiment that has been done is to snip off the caps of two species with distinctive shapes, and swap them. Gradually, the transplanted cap will assume the species-specific morphology of the host nucleus. So some genetic factors are clearly being transported from the rhizoid to the cap, where they then dance together to make a characteristic parasol.

That’s the interesting part about structuralism. The old-school extremist version is pretty much dead, along with Thompson and Goodwin, but there are valid questions still waiting to be answered. They just aren’t the questions creationists are asking.

Now evo-devo, though, does have a distinct bias towards multi-cellular animals (and Arabidopsis). But evo-devo isn’t really a descendant of structuralist thought — it’s more a scion of comparative molecular genetics. It’s a bit more mainstream. Likewise, the people I’ve known who have structuralist leanings (hi, Jay!) don’t deny genetics or evolution — far from it — but like to point out that there are patterning mechanisms beyond genetic switches, and that there are physical constraints on possible forms.

Caricaturing modern structuralism as being all about constraints that force humans to evolve on every planet is even more ridiculous than caricaturing modern molecular biologists as people who claim that biochemistry is irrelevant. I suppose people like that exist, but no one takes them even as seriously as they do Michael Denton.


  1. kevinalexander says

    So, if physics constrains form to a nearly infinite variety then biology is constrained to a nearly infinite number of forms.
    That’s an important constraint, ergo Jesus!

  2. says

    Well, the constraints may run deeper than that. Maybe all multicellular life is constrained by history and complexity to be built of arrays of 2-dimensional sheets that then fold and interact to make 3-dimensional organisms.

    So it’s like origami. All forms must be made of square flat pieces that can be folded, but no scissors allowed. That’s a real constraint. Look how limited origami is!

  3. Menyambal says

    I once saw an old book that was speculating about color patterns on animals. The author had cut out metal in the shape of an animal’s skin, if skinned off and stretched out to dry. The shapes were mounted flat, and sprinkled with powder. The edges of the metal plates were then stroked with a violin bow to cause vibration. The powder danced around on the metal, and settled in patterns. Some of the patterns resembled animal color patterns.

    I forget what the author thought he was proving, but he seemed serious about the correlation. Me, I learned to watch out for first impressions and simple explanations.

  4. says

    The thing with the powder settling into shapes on a vibrating metal surface happens because you create a standing wave, and the powder collects in the places where the metal is moving the least. They’re called Chladni patterns, and they can be quite pretty as well as mathematically interesting, but they have nothing (that I know of) to do with how pigments on animals work.

  5. marcoli says

    My hypothesis is that if we replay the history of life we would still have plenty of individuals who do not understand the history of life. This will be understandable if life never evolves species with sentient intelligence, but it will still be true if intelligence were to evolve.

  6. marcoli says

    As for the patterns that emerge due to, say, reaction-diffusion systems versus having similar patterns organized by vibrating planes: It is fun to discover that patterns form in nature thru a variety of different (and unrelated) means, but they tend to fall into a small # of categories. Most patterns in nature, however they form, appear as stripes, spots, branches, or spirals.
    Why? I dunno.

  7. monad says

    I have a likely ignorant question I don’t know how to answer. You hear lots about Hox genes and other switches controlling differentiation in embryos. Before that, though, the first few cell divisions also follow a set arrangement like radial or spiral cleavage. Is that still similar or is that the sort of thing that gets a structuralist explanation?

  8. unclefrogy says

    seems to me that I do not understand enough about a lot of things but I have a very hard time with the difficulty some have with understanding the deeper understanding science reveals about the nature of reality.
    Take this idea as I understand it, the simple rules of physics and chemistry determine what forms life take OK that sounds obvious. Genes determine what forms life take that is demonstrably true.
    That does not mean that the forms life can take are unaffected by the “laws of physics and chemistry” a body can only get so big given the effects of earth gravity so no King Kongs or Godzilla, there are only so many ways to fly so no superman flying. That there are many variations with those constraints is observable. What is it that makes it so hard for some to see what is right there in front of them even when it is pointed out with evidence, it is the interaction of all of these things acting in time renders what we observe today.
    uncle frogy

  9. jmpur says

    I’ve seen references Thompson references in some Gould and Eldredge. Anybody have any ‘landmark’ papers or book on this topic to recommend? Or just some general favorites?

  10. chrislawson says

    Some things you expect to be highly constrained by physics, like wing shape, but even in this particular aspect of life that has evolved numerous times independently, there is a huge variety of structural variation — the wing of a dragonfly looks quite different to the wing of a blackbird or a bat, looks quite different to the “wing” of a flying fish or a sugar glider.

  11. John Harshman says

    D’Arcy Wentworth Thompson. On Growth and Form. That’s probably what you’re looking for.

  12. biogeo says

    Well, animal coloration patterns may be in part governed by reaction-diffusion equations (which I think PZ alluded to in the post), which have the mathematical form of wave equations, so it’s probably not entirely coincidental that the Chladni patterns you get by bowing on a sheet of metal have some resemblance to animal coloration. But neither is it particularly meaningful beyond noting that it’s awesome and beautiful that the same mathematical forms are repeated throughout nature; to actually understand the development of animal coloration patterns requires a heck of a lot more than just writing down a wave equation and solving it for the adult’s skin.

  13. Tethys says


    Likewise, the people I’ve known who have structuralist leanings don’t deny genetics or evolution — far from it — but like to point out that there are patterning mechanisms beyond genetic switches, and that there are physical constraints on possible forms.

    I simply don’t understand why pointing this out is a heresy. Aren’t geometry and chemistry considered physical constraints? The physical constraints for one celled life are different from the constraints for larger metazoa. There aren’t any square or rectangular mammals, but single celled life forms frequently exhibit such geometry.

    I also wonder what, if any, effect the basic mechanism of calcium sodium ion exchange and being carbon based has on structure.

    The environment, and the raw materials have an influence on what forms are possible in creating everything from machinery and engineered structures to art. It seems silly to say that the same factors are not constraints on life forms.

  14. throwaway, butcher of tongues, mauler of metaphor says

    unclefrogy @9

    What is it that makes it so hard for some to see what is right there in front of them even when it is pointed out with evidence, it is the interaction of all of these things acting in time renders what we observe today.

    I always used to attribute it to lack of understanding or wisdom. But now it seems that they’re simply nostalgic for a time when what they thought they knew enabled their a priori established beliefs. They choose not to.

  15. says

    As PZ has pointed out a mechanism for color patterns involving oscillations seems unlikely, and as biogeo@14 points out, these may have some similarity to reaction-diffusion systems. Those, however are more nonlinear than Chladni patterns. Color patterns also involve events in early cell divisions of the embryo and subsequent migrations — there is not a need for them to be continually reinforced by a standing wave.

    The 1941 edition of D’Arcy Thompson’s On Growth and Form can be read online at here. Wikipedia has a discussion of its strengths and limitations — refusal to think about the role of evolution being a major limitation. Still, it is a remarkable book, and quite well-written.

  16. dannorth says

    “if you replay the tape of life you will always get species that look pretty much like the species we see today because the basic forms (Baupläne) are the inevitable consequences of the underlying physics.”

    I think this idea appeals to creationists because it gives that veneer of scientific respectability to the idea that the world obeys a plan, a design and hence has a designer.

    They seem to have difficulty wrapping their heads around the concept of contingency.

  17. says

    Yeah, lovely batch of coconuts you found PZ.. lol

    This actually reminds me of “every” single old style “genetic algorythm” game/program there has ever been. Yes, the “physics” of the games/simulations did constrain the resulting morphologies, and, in principle, thus the evolution that was possible. Only… Those environments tended to have **existing constraints** already built into the models, so there was no way for them to apply unique physical solutions to said models, to break any of those constraints (or maybe one should say, bend them). The idea that the laws of our physical universe apply similar constraints… yeah, I have no problem, more or less, with this. Except for *two* things – 1) we have no way bloody way to define just what those constraints actually are, and how they *must* effect morphology, and 2) the.. shall we say, “granularity” of the system we exist in is such that the only real constraints, being basic things like soil composition and terrain, of which there is a vast, but not infinite amount of variations, is the environment of living things themselves. The result is not, “physics, therefor terrain, therefor basic designs, therefor humans”, its, “physics, therefor terrain, therefor ???, therefor interlinked, interdependent, constraints, based on the morphology of what ever step 3 produced, which **maybe** leads to something human like.” I would tend to, just from the requirements of being “able” to do physics (its actually not terribly plausible for non-terrestrial creatures without “some” sort of complex vision system and an effective way to manipulate objects, to develop technologies (but.. that may be just pure opinion on my part), so.. I would rate the necessity of something being at least “sort of” human-ish, more or less, being the “technological” species on any place they came up. The closest we get to something not human is, well, squid, and.. its hard to see a effective “development” of tech that would allow them to actually take advantage of say, volcanic vents, never mind develop a tech version, to develop metallurgy,, for example, or, at least not without undergoing morphological changes of their own, which would a) actually give them more durability to certain kinds of environmental conditions, and b) end up constraining the very mobility of their limbs, in the process, in ways that would lose them “most” of their adaptability for use to actually do such experimentation.

    I mean, there is, in principle, a pretty solid reason why you don’t actually see “land squid”, and the same limitations are bound to create issues trying to develop “underwater” tech, which goes beyond something like stone work, or sea farming… Though, again, just because I can’t see any plausible “intermediary” process to get from fishing.. err, fish, out of odd places, to landing on the moon, compared to a terrestrial, doesn’t mean there isn’t one. Just.. It should at least be possible to imagine one, if there was any plausibility to it.

    Bugs are about the same problem. The only time they grew large enough, in history, that hey might have, possibly, ended up becoming something more the oxygen levels we so high that one thunder storm would have lit one fire, and wiped out the first bug to invent the wheel. It seems, given “their” constraints that.. any environment capable of producing something of sufficient size to match a cat, never mind smarter, would be so dangerous it would never survived to get that smart. The environment itself would, almost certainly, change before it could happen, and rob them of the opportunity.

    But, yeah, what dannorth said – contingency. You can’t get there, a second time. You might get close, and it might even be argued, if we actually understood the actual requirements needed well enough, that you would *have to* end up with something close, but.. you still can’t get there again, the next try.