The strange inversion of natural classification 6 Aug 2010 Around the time Charles Darwin headed off on his world tour (Rio! Sydney! Capetown!), taxonomists exercised themselves greatly over what was a “natural classification” in natural history, roughly in biology and geology. The shared view was that, as the system of Linnaeus was artificial, relying as it did solely on the sexual organs of plants, or a few characters in animals, the classification that resulted was at best only roughly telling us what the nature of things were. Competing systems included that of Antoine Laurent de Jussieu and Michel Adanson, who used many characters to draw up their tables of classifications. But the classifications themselves were regarded as being statements about the natural order. The “method” used was independent of that fact. Classifications were attempts to work out the order of nature. But there was no theory involved; these classifications were made on the basis of empirical investigation. “Affinities” were identified on the basis of the overall characters of organisms. There simply was no theory. Just as species had been identified long before even the rules of heredity were understood, so too classifications were done before the science had explained them. Indeed, the classifications themselves triggered the development of theory, most notably that of Charles Darwin’s theory of common descent.* Sometime around the end of the nineteenth century, things had changed. In his 1906 book La théorie physique: son objet et sa structure (in English published as The Aim and Structure of Physical Theory), Pierre Duhem, the physicist, wrote that natural classification was “those ideal connections established by [the zoologist’s] reason among abstract conceptions [that] correspond to real relations among the associated creatures brought together and embodied in his abstractions” [p25]. Later [p297], he says that physical theory approaches a “natural classification” by arranging experimental laws: “The ideal form of a physical theory is a natural classification of experimental laws”. A natural classification is a theory. With Quine, and later David Lewis, classifications are simply derivations from theory. In the famous paper “On what there is” (1948) Quine said that “Our acceptance of an ontology is, I think, similar in principle to our acceptance of a scientific theory, say a system of physics: we adopt, at least insofar as we are reasonable, the simplest conceptual scheme into which the disordered fragments of raw experience can be fitted and arranged.” A classification, although Quine does not use that word, is a conceptual scheme, and, famously, “To be assumed as an entity is, purely and simply, to be reckoned as the value of a variable [of a formal statement].” The later development of this led Lewis to use the “Ramsayfication” of theories as a way to determine what things existed under the theories, and how they related to each other. Over the course of a century or so, a natural classification went from being what we observed without theory to what we derived from theory. Things weren’t helped much by the insistence by Popper that theories were not inductively developed from experience and data, and that all observations were bound to theory. So far from being the case that we developed theories by observing and classifying, as the early nineteenth century taxonomists had presumed in a naively Baconian manner, we instead could not even observe things without a prior theory. Despite Popper’s view of science being largely abandoned by the mid-1970s among professional philosophers of science, it was still influential upon the new breed of taxonomists known initially as phylogenetic systematists, and later as “cladists”. A classification was a constructed theory, a hypothesis to be tested and falsified if possible. How is it, that the classifications that early naturalists developed, and which remain largely relevant, if not still valid, today, was done without theory, if the modern view is correct? The answer, it seems to me, is that the modern view is not entirely correct. It does not exhaust the possibilities of the dynamic of scientific investigation. Induction itself has had a revival among the “new experimentalist” philosophical movement led by Hacking, but classification has had no such revival. It is time it did. How that is to be undertaken is the subject of my next post in this series. * Darwin called it “descent with modification”. Epistemology Metaphysics Natural Classification Philosophy Science Systematics Philosophy
Epistemology Are smarter people irreligious or just nonconformists? 14 Aug 201314 Aug 2013 There’s been a lot of discussion around the traps that studies show repeatedly that those who are atheists or otherwise irreligious are on average a bit smarter than those who aren’t. The usual ballyhoo has followed, with atheists claiming that religion makes you stupid or only stupid people follow religions,… Read More
Administrative Vote now! 3 Jun 2010 I have a couple of posts nominated (one by me – it’s time to stop being humble) at the 3quarksdaily Science competition here. I won’t win, because that torrent of good science writing, Ed Yong, is nominated, not to mention Carl Zimmer and sundry others, and the Pharynguloids will of… Read More
Philosophy A question to my readers 10 Jan 2011 Early January is by tradition a time of soul searching, reflection and dieting. I am also about to move from Brisbane to Sydney, and take up a new direction, and I was wondering about the use of this blog. I’ve been blogging now since February of 2007, first at Blogspot,… Read More
The idea of a natural order comes from a creationist (or intelligent design) way of thinking. A more naturalistic view ought to be that there is no reason to suppose that there should be a natural order. And therefore as pragmatists, we will need to construct our own orderings (classification systems) in order to make it possible for us to be able to express facts about our world. It seems that almost everybody, including many evolutionists, base their epistemology on creationist thinking (finding the natural order). I have been following the more naturalistic path, but I am feeling quite lonely out here.
I think that if the world has a definite structure, there is a natural order. There’s nothing creationist about it.
I don’t find much evidence of a natural structure. If there were such a natural structure, we would not have seen Kuhnian paradigm shifts.
Thorny, care to do a post on this? Or perhaps you already have. I’ve read internet scribes using Kuhn paradigm shifts as a weapon to attack science, hoping to reduce it to another narrative, no less faith-beholden than religion. I’ve not read Kuhn, as there’s so many things to read, and I’m not sure I’d get a good view of the battle-field anyway. Sigh.
Maybe I’m too steeped in modern philosophy of science, but the idea that classification prior to Darwin was done w/out theory seems odd to me. It isn’t as if classifications were done along arbitrary lines, picking just any old character trait, right? Didn’t the classification of living things under species reflect species essentialism that would have been prevalent? Otherwise, how were the character trait differences that indicated a new category identified? ‘Overall affinity’ doesn’t seem like it would provide very consistent or objective inclusion criteria for classes of things.
Species essentialism is a myth, as I argue in my book on the history of species. This is what leads me to this point.
That certainly is an interesting shift. The way you’ve laid it out here at least suggests that the change may be due partly to positivism — that’s certainly the origin of Duhem’s view. On the positivist view it’s not so much that classifications are theories as that theories are simply classifications; in Duhem’s version, for instance, theoretical progress in physics consists of refinement of artificial classifications of physical phenomena in such a way as increasingly to approximate the natural classification. One can imagine a course of events in which people with a non-positivist view of theory nonetheless came to accept the positivist idea that theory should classify. But I don’t know if that fits the actual course of influence and modification.
Two things. First, I’m not sure about your interpretation of Quine (1948). You make it sound like Quine is saying that the act of classifying comes after commitment to a theory. But I read him as saying that (a) adopting a theory is equivalent to categorizing the world in a certain way, and (b) one should adopt a theory only if the theory is properly related to the phenomena (where proper relation is spelled out in some detail). Second, Quine in his later work argues for observation categoricals, which work like little theories that are pretty directly accountable to evidence. Why not think that classification schemes like those of the 19th century couldn’t be handled by moving up from observation categoricals?
Quine’s views are a symptom rather than a cause. It’s my view that the shift occurs with the acceptance by the Cambridge empiricists of Mill’s view of natural kinds. Quine is used by some as a way to justify the theory-first view, but his “qualitative spaces” suggests that he thought these were mediated to us by evolution, so he might have agreed with you. Lewis, however, wouldn’t have. I toyed with putting Carnap’s Aufbau in the mix as well, but I don’t know the literature well enough. However, few analytic philosophers even talk about classification – Woodger is perhaps the only one in the period leading up to the cladistics revolution. I leave the Foucauldian school out of this lineage for now.
Sorry … that last sentence has one too many negations! Should be: Why think that classification schemes like those of the 19th century couldn’t be handled by observation categoricals?
I’m not sure whether it counts as a theory, but doesn’t the proper taxonomy of kinds depend on the actual genealogy of individuals? When I encounter a spider in the bathtub, shouldn’t I recognize her as literally my zillionth cousin, umpteen times removed, and base my understanding of the relationship between her clan and mine accordingly?
Is classification a kind of theory, a prerequisite for theory, a component of a theory, or a consequence of theory? I would think that it could be any of these, depending on what “theory” means and upon what kind of theory you are talking about. There is no one right answer. (As long as you use a naive Bayes classifier for the actual work, that is.)
I will discuss this in one of the next posts. You are, I think, absolutely correct that there is no correct answer. I have a metaphor for that…