Going Complex

Adrian Currie writes...

How we think about scientific progress reflects our ideals about good science; as such, I think how paleontology progresses—successfully but (perhaps) unusually—can challenge some of those ideas.

I recently argued we should be wary of simplicity in science. Often the world is a complex, messy place which demands complex, messy explanations, so we should be cautious of the intuitive appeal of the simple, the neat, and the tidy. Today I’m going to make a related point about the nature of progress. In a nut-shell, here’s the idea:

‘Progress’ is a pretty dirty word in some philosophical quarters: it carries the whiggish whiff of the bad old days when philosophers saw science as a naïve march from false stuff to true stuff (if such bad old days ever existed...). And indeed science is a much trickier beast than that! Regardless, what I have to say about progress won’t depend on whether you think of it in ‘truthy’ terms or not.

How about this: science progresses insofar as it achieves some of its goals. These goals are extremely diverse – from grasping true propositions, to developing new technologies, to asking new questions, to positing new hypotheses, to developing new treatments, to (if you must) getting new grants, publishing new papers, and so forth. In this extremely weak sense, we can all agree, science progresses (let’s save realism for future posts…).

So, today’s question is: how does science progress? That is, are there patterns in scientific progress? At a glance, I’d say philosophers tend towards two kinds of answer to this question. First, go small—that is, progress via reduction. Second, get together—that is, progress via unification. Let’s glance at both.

A fairly recent example of going small is from Carl Craver’s 2007 book Explaining the Brain. There, he uses case studies from cognitive neuroscience to analyse mechanistic explanation. Roughly speaking, a mechanistic explanation of some target (say, a clock) breaks it down into components (cogs, springs, and so forth), and then explains the target’s behaviour in terms of the causal relationships between those components. For Craver, scientific investigation progresses by extending these decompositions and thereby providing a deeper and more complete understanding of our target. We progress from explaining how mechanisms could work to how in fact they do work via further decompositions piled on decompositions, arriving at last at the ‘complete mechanism’—a mapping between the components in our target system, and those in our mechanistic explanation. Progress is achieved by zooming in; going small.

Now, to unification. Here, sciences progress when we produce theories, models or explanations which do a lot of work, that is, they explain a whole bunch of things. The poster-philosopher for such views is Philip Kitcher:

Science advances our understanding of nature by showing us how to derive descriptions of many phenomena, using the same patterns of derivation again and again, and, in demonstrating this, it teaches us how to reduce the number of types of facts we have to accept as ultimate (or brute) (Kitcher, 1989, p. 432).

On this picture, good science provides ‘argument schema’, re-usable models which can be applied to a whole bunch of different phenomena. I’ve previously discussed how theories of island biogeography can explain a wide range of strange critters on islands. In virtue of knowing a few things about islands (they are isolated, who reaches them is a bit chancy, etc…) I can explain cases of island gigantism and dwarfism, flightlessness, and a range of other things. Basically, I plug the relevant facts into the biogeographical framework, and hey presto, I get an explanation (to be fair to Kitcher, his account is of scientific explanation as opposed to scientific progress – but it’ll do for my purposes). Progress is via unification; getting together.

Now, of course, science often proceeds as Craver and Kitcher paint it, and indeed going small or getting together (or both!) are great ways of going forwards, but as we’ll see, that’s not how it goes in paleontology. Instead, I see a characteristic shift from a bunch of simple explanations, to increasingly complex, integrated ones. We don’t succeed by going small, or by getting together. No: we succeed by going complex.

To see this in action, let’s consider those enormous sauropods again.

As I’ve discussed before, explaining how sauropod dinosaurs got so big is tricky. Here are some hypotheses.

Low nutrition = Big Bodies. Maybe Mesozoic plants were, compared to our fabulous modern plants, low in nutrients. This means that if Mesozoic critters want the same amount of nutrients, they should eat more. That means a lot more digestion, and hence a lot more space. Midgeley et al discussed this in 2002.

The oxygen hypothesis. Atmospheric oxygen fluctuates over evolutionary time, and high oxygen appears to correlate with important biological events: the emergence of oxygen-producing microorganisms and vascular land plants, the Cambrian explosion, both arthropod and vertebrate land invasions; plus increases in animal size: arthropods in the Carboniferous, proto-reptiles in the Permian and mammals in the Tertiary. Perhaps the same could be to blame for sauropod size?

The vacuum hypothesis. Ruxton and Wilkinson have used mathematical modelling (and analogies with vacuum cleaners!) to argue that sauropod gigantism was enabled by their long necks. Having a long neck lets you take in more food without having to move, thus conserving energy. Mammals can’t manage similarly long necks (because of our fancy teeth; but that’s a story for another day…), so this explains how they got so big and we didn’t.

Gigantothermy. Here’s some super-basic thermodynamics for you. If you’re really big, you retain heat. Geomorphologists and climate scientists use this principle, ‘thermal inertia’, to understand temperature in massive bodies like planets. But it also seems to apply to massive organisms, and that could be a real advantage if you’re cold-blooded—as sauropods *might* have been (I’ll save the debate on sauropod thermoregulation for another post!). Stored external heat would last longer for sauropods than their diminutive cousins: they wouldn’t have to waste so much time sitting on rocks like your basic lizard. Gillooly, Allen et al did some lovely modelling of this in 2006.

Notice that these four hypotheses have some of the features which Kitcher or Craver might like. Low nutrition = Big Bodies and Gigantothermy understand sauropod physiology in relatively reductive terms (well okay, It’d be a stretch to call them ‘mechanistic’ exactly…); and they have unifying features: Gigantothermy links thermoregulation in large animals to principles from geomorphology and physics, the oxygen hypothesis links sauropod gigantism to gigantism in other animals, and so forth. The hypotheses are also all simple: they identify particular causal relationships as the change-makers of sauropod evolution.

Although these hypotheses have similarities, they've met with different fates...

Midgeley et al’s appeal to low-nutrient plants in the Mesozoic fell down when Hummel et al experimentally examined plants thought to be similar to those sauropods munched on. They turned out to have similar nutrient content to modern flora. The oxygen hypothesis has a pretty major flaw: as Sander and Clauss pointed out back in 2008, it doesn’t look like fluctuations in atmospheric oxygen tracks increases in sauropod size. Oh well.

Gigantothermy and the vacuum hypothesis haven’t been rejected, per se, but seem to be in the process of being incorporated into bigger, more complex explanations. It seems plausible that the long necks were part of the story, but surely they can’t be the whole story. If sauropods are cold-blooded (and that’s a big if!), gigantothermy likely plays a role in explaining some of the advantages of being massive (besides being unpredatable of course!).

A pattern emerges.

First, a series of apparently independent – and simple – hypotheses are put forth. Sauropod gigantism was required to mitigate the low-nutrient flora of the Mesozoic; sauropod gigantism was enabled by long necks which enabled maximal intake at minimal outlay; and so on.

Second, these simple hypotheses are put to empirical tests. Some are found wanting and abandoned – the Mesozoic flora was not so impoverished! So far, this looks a lot like Carol Cleland’s "smoking gun" view of method in historical science – but there’s an extra step:

Third, the survivors are integrated into complex, multi-leveled explanations. The simple hypotheses become ingredients in a larger recipe.


This integration stage looks like progress to me – epistemic success, not failure (as might be implied by views that emphasize unity or reduction). It’s a success because in this circumstance paleontologists are trying to explain a highly contingent, complex phenomena which suits just this kind of narrative explanation.

It seems as if – at least sometimes – paleontology progresses by shifting from the simple to the complex. Instead of “getting together” – of providing general, unified explanation schemas -  palaeontologists often give us highly localized, specific and contingent explanations: they emphasize how fragile and unlikely the event in question is. Instead of “going small” – of decomposing our targets and providing bottom-up explanations – palaeontologists provide integrated explanations which span a wide variety of hierarchical levels and temporal grains.

[Joyce questions whether integration is really separate from unity. I'm inclined to think they're relevantly different: (1) integration is local and piecemeal; (2) integration is multiscalar; (3) integration is complex; (4) integration is incomplete. By my lights, what matters about unity is that it is (more-or-less) general and systematic, focused on a single scale, simplifying and complete. Rhetorically, detail and autonomy are de-emphasized by unity but emphasized by integration.  I'd be really interested to hear what people think in the comments...]

Why does this matter? It matters because, among other things, views like Craver’s and Kitcher’s are expressions of what we should expect successful, legitimate science to look like. We are told to expect science to get neater, more unified, and more concerned with smaller grains as it goes along. But science doesn’t need to be like that. Often science drives towards the complex, the partially integrated, and the holistic: one way of progressing is to go complex.

So, paleontologists, if you’re feeling the need to compare yourself to other sciences, remember:

Or, as Skeletor puts it:

Further Reading

The ideas of this post are drawn in part from:

Currie, A.(2014). Narratives, mechanisms and progress in historical science. Synthese, 191(6), 1163-1183.

For a view of science which emphasizes integration, Sandy Mitchell is a good place to look.

See here for more Skeletor affirmations

Here are the philosophy references:

Cleland, C. E. (2002). Methodological and epistemic differences between historical science and experimental science*. Philosophy of Science, 69(3), 447-451.

Craver, C. F. (2007). Explaining the brain. Oxford: Oxford University Press.

Kitcher, P(1989). Explanatory Unification and the Causal Structure of the World. In: Kitcher, P.& Salmon, W. (Eds.), Scientific Explanation (pp. 410-505). Minneapolis: University of Minnesota Press.

And here’s the science:

Gillooly JF, Allen AP, Charnov EL (2006). Dinosaur fossils predict body temperatures. PLoS Biology. 4:1467–1469

HummelJ., C. T. Gee, K.-H. Südekum, P. M. Sander, G. Nogge, and M.  Clauss. (2008). In vitro digestibility of fern and gymnosperm foliage: implications for sauropod feeding ecology and diet selection. Proceedings of the Royal Society B 275:1015-1021.

Midgley JJ, Midgley G, Bond WJ.(2002) Why were dinosaurs so large? A food quality hypothesis. Evolutionary Ecology Research. 4:1093–1095.

Ruxton, G. Wilkinson, D. (2011). The energetics of low browsing in sauropods. Biol. Lett. 7(5) 779-781.

Sander, P. M., and M.  Clauss.(2008). Sauropod gigantism. Science 322:200-201.