*Responses are short posts engaging with a previous post - do you want to write one? Get in touch!*
Adrian Currie writes...
Earlier in the week, Derek challenged the emphasis placed on dinosaur paleontology by, I suppose, the public and, well, us (Extinct has certainly been a trifle dino-heavy…); arguing that other extinct critters—our friends the marine invertebrates in particular—provide better quality data. And this is surely prima facie true. Fossilization requires water and luck. Rare, lumbering terrestrial beasts are going to be far fewer in the fossil record than the abundant bounty of hard-shelled invertebrates. So, if I want a whole lot of data I should break for crabs and ammonites, not archaeopteryx and corythosaurus. Derek’s basic point is, I think, important and worth remembering: dinosaur paleontology is a pretty small part of vertebrate paleontology, which is itself in many ways dwarfed by invertebrate paleontology. And for good reason: you can do stuff with invertebrates which you can’t with vertebrates. The thought that paleo = dino is an impoverished one.
I think Derek’s argument deserves a bit of reflection. He argues that “Rarity… diminishes scientific or epistemic value.” The thought being that because big data sets are good, rarity leads to diminishing epistemic value. But can we connect rarity and epistemic value - small data sets - divorced from what you might call the epistemic context: what we’re using the evidence for, and how much we know about the case? I don't think so (by the way, David Bapst makes a similar point in the comments section of Derek’s original post, and very nicely: worth checking out!). I'm going to argue that evidential value is crucially context-sensitive: evidence is good evidence for something - and for some somethings, rare evidence is perfectly sufficient. Thus, Derek’s conclusion about rarity only goes through if you think the kinds of things the invert fossil record is good evidence for (questions requiring big, well-resolved data sets) are more important than the things the vert fossil record is good for.
Let’s look at a counterexample to Derek’s claim about rarity using one of my favourite examples (it forms the central case in my still-under-review book’s introduction). In 2013 Pian et al announced the discovery of a new species of platypus, Obdurodon tharalkooschild, which lived around 15 million years ago. She was a decidedly odd example of an already odd creature. As opposed to extant platypus, she retained her teeth into adulthood, perhaps ate small vertebrates like turtles and, like, was massive. A little over twice the size of your extant platypus. The media, as the media does, named her ‘platyzilla’.
Peter Schouten's wonderful reconstruction of O. tharalkooschild, with molar inset.
No one questioned the find: so far as I know, the paleo community was like “yep, new platypus!” But the funny thing is, the claim's material evidence looks vanishingly small—Pian et al only had a single tooth. If rarity leads to small samples and thus evidential crapness, then something’s gone desperately wrong in paleontology.
But things haven’t gone wrong. Why not?
We know heaps about mammal teeth (dear god the amount mammal-paleontologists know about teeth!): how they develop, their relationship to body-size, how they decay given different diets, how different dental morphologies are suited to different functions. These people can tell you a lot based on a tooth.
So, because we know so much about teeth, and because teeth retain a rich record of their owners, a single tooth can tell us a lot—and with high confidence! Statistics be damned. And in fact this is a common feature of vertebrate paleontology, going all the way back to Cuvier and his rather wonderful boast:
At the sight of a single bone, of a single piece of bone, I recognize and reconstruct the portion of the whole from which it would have been taken. The whole being to which this fragment belonged appears in my mind's eye. (Martin Rudwick’s Earth’s Deep History has a lovely discussion of the quote)
Cuvier, 19th century giant in vertebrate anatomy and paleontology. Given his work in establishing the reality of extinction, a suitable mascot for Extinct...
So, if you know a lot about something—if the epistemic environment is very rich—then diminutive data stands tall: it provides extremely good evidence. The lesson is that mere rarity or otherwise isn’t sufficient for making claims about epistemic value. Derek correctly points out that if you want to investigate macroevolutionary trends—understand the tempo & mode (as GG Simpson put it) of evolution, then you’re going to need a nice, rich, full data-set. But if you want to identify a new kind of platypus, a single tooth and a whole mess of background knowledge will do you fine.
One might respond by pointing out that the ability to track big numbers bequeathed from the invertebrate fossil record (in combination with new computational techniques and technology) allows palaeontologists to ask the big questions: that was what underlay the paleobiological revolution; it is in virtue of this that paleo gets a seat at the ‘high table’ of evolutionary theory. Sure: but here’s my question. Why, exactly, is testing evolutionary theory more valuable, epistemically speaking, than discovering a new species of platypus? I suppose you might appeal to the generality of one versus the localism of the other, or perhaps big questions are more fruitful, or you could gesture to some other epistemic virtue. But this needs lots of filling out, and I’m suspicious that two such different pieces of knowledge really have comparative value: show me your metric!
The discovery that, say, Cope’s Rule holds (that individuals in a lineage increase in size as the lineage ages) is a discovery that would need a large data set to establish (to say nothing of working out why). But the existence of a new lineage of platypus doesn’t. If I care more about the new platypus than I do about evolution’s Big Questions (not saying that I do, of course), what can the philosopher or the paleontologist say to prevent me?