In which I talk myself up

Mon 19 Nov 2012 11:25 AM

I was at the Philosophy of Science Association meeting last week. Good times. Edification. All that stuff. I won't try to estimate the schmoozing to learning ratio, but the tallies in both columns were considerable. I have now been in the profession long enough that the conference lets me catch up with friends who I haven't seen in years.

Even though my paper was about something else, I talked a bunch about natural kinds. My book is sufficiently new that the publisher did not have a copy of it yet for the book exhibit, so it came up naturally in conversations about what I've been working on. And I spoke with other people whose papers did touch on the topic.

Palgrave's webpage for my book has a link to download a "sample chapter", but that really just means the three-page Introduction. I've pasted in the contents of the Introduction below the fold, in case you'd rather read it here. The download also contains the complete table of contents and index, if you want a more telegraphic but complete summary.


from Scientific Enquiry and Natural Kinds

In Plato's Phaedrus, Socrates is concerned with how things are grouped together and how they are divided. He makes an analogy with carving up an animal. Just as cuts of meat should be carved at the joints rather than broken across bones, our account of the world should carve nature at its joints. So we inherit this grisly metaphor for what scientific enquiry does when it aims to discover the real divisions in nature: When science finds the natural kinds, its concepts are the chops and steaks of the world.

The metaphor is both grisly and unfortunate. A butcher is only interested in the animal for one project: butchering. Scientific enquiry is not so unified. The pig which is butchered just one way may be described by scientists as consisting of molecules, cells, or metabolic systems, and none of these descriptions is more natural than the others. The problem is magnified when we enlarge our view from the pig and think about the world altogether. The problem is not that there are no rewarding places to cut, but that there are too many. There are so many joints in the world that we could not possibly carve it up along all of them. This might be a cause for nominalism or despair, but it need not be. The divisions we find might be real features of the world, even if they are more numerous than those split by the butcher. It is an important fact about the pig's biology that it is made up of cells, and an important fact of its biochemistry that it is made up of molecules. No single enquiry can make all the right cuts. Rather, different enquiries require cutting along different joints.

Let's leave the ancient butchershop and consider the night sky. Even Socrates knew about planets like Mars and Jupiter. There are others which he did not know about: Uranus, Neptune. And there is a dim point in the sky - Pluto - which I was taught about as a planet but which astronomers no longer consider to be one. In 2006, when astronomers regimented the use of the world `planet' in a way that excluded Pluto, were they carving the solar system at a joint? Is the planet category a natural kind?

Questions like this can be raised about any of the categories that are employed in scientific accounts of the world. As a philosopher of science, I want the resources to think about such questions.

Of course, natural kind talk brings with it a train of philosophical baggage. In recent decades, the issue has mostly been treated as a matter of natural kind terms, the language that we use to pick out categories. This makes it an issue about nomenclature and about the way that language works. Applying such an approach to the case of Pluto, the issue of whether there is a natural kind structure to the solar system becomes entangled with the semantics for the word `planet' as it was used by astronomers at various times. I suggest we start somewhere else - asking instead about taxonomy, the categories we posit, and how they fit the world. This alternative starting place allows us to identify natural kinds without pretending to have a complete account of semantics or deep insight into fundamental metaphysics. The resulting account of natural kinds is modest and pragmatic, but nonetheless realist.

To put it loosely, the account of natural kinds which I defend maintains that a category of things or phenomena is a natural kind for a domain if it is indispensable for successful science of that domain. Scientific success involves making sense of the things or phenomena - both accurately predicting what they will do and explaining their features. The account conflicts with the tradition that associates natural kinds with fundamental and precise essences.

I claim no great originality in associating natural kinds with conditions for inductive and explanatory success. The connection to successful science echoes many voices in the pragmatist tradition. Although pragmatism is sometimes taken to be antirealist or even idealist - even by some pragmatists themselves - I argue that the natural kinds that I describe are features of the world in a straightforward way. The appropriate metaphysical picture is one of pragmatic naturalism and modest realism.

Ultimately, an account of natural kinds must pay its rent by helping us understand actual science and its relation to the world. So I discuss both a large number and wide range of examples. Some of these, like chemical kinds and species, are well-explored terrain. Some of the specific cases, like gold, water, and tigers, are much interrogated denizens of that terrain. Yet I try to avoid the uninformed casual gesture to tigers or lemons. My arguments will range over mosquitoes, ducks, and deepsea fish; the details turn out to matter. I also consider cases, like the planets, which have received little philosophical attention.

In Chapter 1, I make a survey of things that people have thought about natural kinds. I argue for the centrality of three constraints: that natural kinds should support induction, that natural kinds should figure in successful science, and that natural kinds are relative to domains of enquiry. In Chapter 2, I offer my account as a promising way of satisfying these three constraints. In Chapter 3, I apply the account to several specific examples: planets, species, and distributed cognition. In Chapter 4, I articulate pragmatic naturalism and clarify the sense in which my account is realist. In Chapter 5, I entertain the objection that my account is too liberal and includes too many categories as natural kinds. In answering the objection, I consider examples of culinary taxonomy and animal signaling. In Chapter 6, I consider the relationship between natural kinds and homeostatic property clusters.


from: Greg

Tue 20 Nov 2012 05:44 AM

Just for clarification, is the following your view? Natural kind terms pick out real features of the world, but whether or not a particular kind term is a NATURAL kind term is indexed to/dependent on whether that term is indispensably used in a successful scientific theory at time t.

If that's not your view, where do you get off the boat?

from: P.D.

Tue 20 Nov 2012 08:28 AM

Greg: There are two ways in which I'd dissent, although these might be a matter of emphasis.

First, I do not pose my view as being about terms at all. My focus is on categorization rather than language. So I claim that categories are (or are not) natural kinds, and I am not concerned to say whether terms are (or are not) natural kind terms.

Second, I don't think there needs to be as much indexing to times as your formulation suggests. Very often, something which might be described as indispensable at a time (but not later) really only ever seemed indispensable. For example, phlogiston was never a natural kind.

On my view, a kind is only a natural kind relative to a domain of enquiry. I try to distinguish domains mostly as parts of the world, but it would be possible instead to build more of the scientific community and its capabilities into the domain specification. Then there would be different natural kinds at different times, but because crucial facts about the community (and so the domain) had changed.