4/14/2008

Are there empty natural kind terms? (The 2nd in a series)

There are empty names, like 'Santa' and '(Planet) Vulcan'; there is a fairly large literature dealing with them in both the philosophy of language and in logic (called 'free logic'). But there has not been any discussion of empty natural kind terms -- which prompts the question: are there any such terms? I ask because it seems to me that 'phlogiston,' 'caloric,' and other central terms of now-discarded scientific theories may qualify as empty natural kind terms.

This question is complicated by the fact that there is not widespread agreement on what natural kind terms are. The two candidates are (I) predicates and (II) names. (Predicates are the leading contender, so you can skip the final paragraphs if your patience for this kind of thing is limited.)

(I - Predicates) This takes us back to the subject of the previous post: are there any empty natural kind predicates? As noted in the last post on this subject, if 'empty' just means 'has the null set as its extension, and the whole domain of discourse as its anti-extension,' then the answer is obviously yes. But then it's uninteresting -- empty names are interesting because (both for direct reference theorists and Frege) sentences containing them will lack truth-value, unless the direct reference theorist proposes an ad hoc fix (cf. David Braun). If empty predicates behave classically/ nicely, they won't generate truth-value gaps.

So here's an argument that natural kind predicates like 'phlogiston' are empty in the same way that 'Vulcan' is, i.e. they can fail to express semantic content sufficient to determine truth-value. How? On a Kripkean account, natural kind terms express properties that objects have essentially. To determine what property is expressed by a natural kind term, we take samples of the stuff that that term refers to in the actual world, and determine what is essential to them, i.e. what property (or combination of properties) those samples must have in order to be that kind of stuff. So the natural kind term ‘water’ expresses the property of being H2O, because in the actual world, having that inner constitution suffices for something to be water. But what is the essential, inner constitution of all the stuff we call ‘phlogiston’ in the actual world? Nothing —- since there is no such thing as phlogiston, there is no essential property or inner constitution to discover. (Note: James Woodbridge helped me a lot with this argument; if you like it, you should attribute it to him, not me.)

So if phlogiston has no inner constitution, then 'phlogiston' lacks semantic content. And thus (atomic) sentences containing the term will be semantically defective, and thus (presumably) will lack a truth-value. But is this argument any good?

(II. - names) One might think natural kind terms are names, because they appear in subject position:
'Water is wet'
But if they are names, what are they names of? Scott Soames (Beyond Rigidity) gives two 'obvious candidates':
(i) the merelogical sum of all the water everywhere, or
(ii) an "abstract type" that is instantiated by the stuff that comes out of our faucets etc.

If it's (i), then there are clearly empty natural kind terms, and 'phlogiston' and 'caloric' are examples. However, Soames gives two arguments against (i): first, if (i) were correct, then 'Water weighs more than 1 million pounds' should be true and felicitous, but it seems clearly not so. Second, if (i) were right, then 'water' would not be (anywhere close to) a rigid designator, and there is a widespread intuition (or Kripkean dogma?) that it is.

If it's (ii), it's not clear to me that there are empty natural kind terms; I don't know how one shows a type does not exist (or, for that matter, how one shows a type does exist). As James Woodbridge and Seyed (in the comments on the first installment of this series) pointed out to me, it seems reasonable to say that an abstract type that is somehow contradictory does not exist, but we'll have nothing like the set-theoretic or semantic paradoxes when it comes to natural kind terms. But I am not all that worried about (ii), because it's not clear (again following Soames in Beyond Rigidity) that natural kind terms are names at all -- they seem to be predicates first and foremost. As Soames points out, 'Whales are mammals' is naturally understood as 'Anything that is a whale is a mammal.' So natural kind terms are not names.

7 comments:

Richard Zach said...

Doesn't Kripke discuss empty natural kind terms in Naming and Necessity (specifically, "unicorn", the first item in the addenda)?

Greg Frost-Arnold said...

Hi Richard --

Thanks for the reference -- I just looked it up. Kripke does talk about 'unicorn' in the first lecture. However, he mentions it as a counterintuitive consequence of his views, and says that he won't have time to give the argument for his unicorn claim.

The post refers to Soames' Beyond Rigidity (whose subtitle is The Unfinished Semantic Agenda of Naming and Necessity) because he attempts to spell out in some detail a semantics for natural kind terms that balances following Naming and Necessity as closely as possible while being defensible.

Kenny said...

A common idea these days in the philosophy of biology, as I understand it, is that a species is an individual, which is roughly the mereological fusion of all members of the species. This fusion has a sort of natural unity, because to be members of the same species, individuals must share recent common ancestors (for humans, the last common ancestor is about 30,000 years ago, even though Homo sapiens has been a distinct species for about 200,000 years) and also form a reproductively connected community. I don't think it's ever been considered an objection to this view that it seems false to say "Dogs weigh over a million pounds" and nonsensical to say "Dog weighs over a million pounds". Probably, this is because the response is just to say that none of our terms actually refers to the species. However, that seems awkward for "The species Canis familiaris weighs over a million pounds". Perhaps some of these claims just sound false because we're not used to what a species actually is - after all, creationism leads us to think that there is an essence to each species, which suggests that even completely isolated populations can be of the same species over millions of years.

Anyway, I guess all this is to say that some of this talk sounds awkward for species, which I would have thought of as paradigmatic natural kinds. But perhaps the term "unicorn" and others like it are bound up with a false theory of species, and therefore fail to refer (or whatever it is kind terms normally do).

Anonymous said...

Hi,
Under number 1, when you discuss the ‘predicate’ option, I am puzzled by ‘phlogiston’ talk, shouldn’t it be ‘is phlogiston’ or something like this? Here are two points about the first argument, though I don’t think they bring about any serious problem for the argument. Let us call the method of taking some stuff and determining the inner constitution of it, ‘method A’, then the first argument might be represented as follows:
(i) We use method A to determine the property expressed by a predicate.
(ii) We cannot use method A to determine the property expressed by ‘is phlogiston’.
So
(iii) There is no property expressed by ‘is phlogiston’.
The first point is this. The above argument is plausible, if method A is the *only* way to determine the property expressed by a predicate. Moreover, and this is the second point, it seems that we are shifting from ‘we cannot determine the property expressed by ‘is phlogiston’, in the sense of ‘determination’ used in method A’ to ‘there is no property expressed by ‘is phlogiston’’ in our conclusion. Do these points raise serious issues for the argument? I would like to say 'no'!
Best
Seyed

Greg Frost-Arnold said...

Hi Kenny --

I should've thought of that example of species-as-individuals! I've always wondered whether anything turned on that debate... I don't know a lot about that debate, but my understanding is that the people who say that species are individuals (often) take themselves to thereby be denying that species are natural kinds. But I could be wrong, since I don't really know the topography of the debate.

Seyed --

You make a very good point -- one that I hope I can dodge, as follows: Stalnaker draws a distinction between semantics and metasemantics. Semantics studies what words mean; metasemantics studies how words come to have the meanings they in fact have.

If this distinction is legitimate, then I can say "You are making a point about the metasemantics of 'phlogiston,' whereas I am making a point about its semantics." (And then I cross my fingers and hope the problem disappears.)

Unknown said...

Kripke's Locke Lectures "Reference and necessity" do have a more substantial discussion of cases of empty naturam kind terms such as "unicorn". Unfortunately, the lectures are not published. Let me know if you are interested in reading them, I can send you the copy of them I have.

Greg Frost-Arnold said...

Hi Carlotta -

Thank you very much for the generous offer. I would greatly appreciate a copy of "Reference and Necessity." I would put up a link to my email address here, but I'm afraid the spambots would find it. If, on the blog's main page, you scroll down to "other links", you should see a link "my homepage". If you could send me an email, that would be fantastic.

Thanks again,
Greg