Comments Elsewhere

Here’s another stab and replying through the comment box; let’s see if it works this time.

The best place for you to go for a reply is Chapter 4, section 2 of Thompson’s essay (pp. 33ff), in which he discusses the irreducibility of natural-historical categoricals to other more familiar kinds of categoricals. I’ve tried to give some glosses of the reasons, and I’ll try to give some more, but of course he has more space and more talent than I do, and it was his idea to begin with. But here’s a few of the reasons he suggests to think that your analysis of natural-history categoricals won’t actually pan out.

1. Determining what counts as “intervention.” First, the attempt to finesse problems by attaching a qualification of “without (serious) human interference” (either inside or outside the scope of the quantifier — I’m not sure which you meant, but I’m not sure it matters, either). You’re right that I doubt whether “interference” can be cashed out in terms that are independent of the sort of natural life-cycle categoricals that you’re trying to use the notion of “interference” to explain. Thus Thompson: “[T]he question what counts as [‘intervention’] is surely to be answered, in any given case by appeal to the system of natural historical judgments with the relevant kind as subject. And so we cannot simply take such a category for granted and then employ it in an account of our form of thinking. —If the mother bobcat leaves her young alone, they will wither and rot; if she nurses them they will develop thus and so. In whichcase, though, do we find ‘intervention’, and in which rather ‘what happens, ceteris paribus’? No one will insist that the mother’s nursing be viewed as the intervention of something alien, from without, into an otherwise inviolate cub-system set to evolve in its own direction. But to deny it is just a more stilted way of expressing the thought that bobcats are not to be compared with caterpillars—they do not strike out alone and set themselves straightaway to munching. No, ‘the mother nurses them for several weeks’; I heard about it on a nature documentary.”

You might object that my quotation here isn’t responsive to your condition, which depends on human intervention specifically (so bobcat intervention might not cause problems). But there are certainly cases in which serious human intervention (if “intervention” means anything other than the sense that transparently depends on the aristotelian categoricals themselves) is precisely what makes aristotelian categoricals true. For example, there are aristotelian categoricals that are true of humans: for example, humans master language at an early age, but this would hardly be so if not for massive efforts on the part of other human beings towards babies and young children (as both the case of feral children, and also the fact that the overwhelming majority of babies would simply die if abandoned, demonstrate). You could try to change it to an “alien species” criterion instead of referring to humans particularly, but there are also aristotelian categoricals that are true of domesticated plants and animals; fig trees reproduce by grafting (which is somethign we do), wheat grows in such-and-such a way and ripens in the fall (thanks to tilled soil), domestic cats like to be touched by humans (but feral cats don’t), etc. (And the same could be said of any symbiotic pair of species you cared to pick; for what it’s worth, I think you’re right that “rats have fleas” is true as an aristotelian categorical — that’s part of the reason rats are such loathsome creatures! But that’s because fleas are a permanent and salient feature of the rat’s distinctive form of life, not because some percentage of rats do or don’t have fleas.)

I have some other worries about this criterion, but they’re mostly more trifling, so let’s move on to…

2. Semantic problems with the statistical quantifier. One of the first problems with an attempt at a statistical reading of the quantifier, even with the sort of qualifications you’ve suggested, is that any particular value you pick for the statistical level needed to satisfy the quantified formula is likely to be wrong in some cases. Examples (1)-(5) might plausibly be thought to imply that most, or the overwhelming majority of domestic cats, coyotes, humans, male and female emperor penguins, etc. etc. But that’s not true of all true aristotelian categoricals. For example, the mayfly breeds shortly before dying. I know this is true; I saw it in a nature documentary. However, as a matter of statistical fact, most mayflies die long before breeding at all. (Similarly leatherback sea turtles, to take a nearer cousin of ours.) You could try to patch this up by adding the qualification, “most S’s that reach the appropriate stage of their life exhibit trait T,” but of course this obviously does nothing more than relocate the problem to “appropriate stage of their life.” (And of course you cannot define it as “the stage in their life at which the overwhelming majority of them exhibit trait T,” since that is as tightly circular an analysis as you could hope for.)

3. Logical problems with the statistical quantifier. There are also some straightforward, if technical, logical reasons not to read “The cat has four legs” as “Most cats,” “The overwhelming majority of cats have four legs (… when such-and-such defeaters aren’t operative …).” As Thompson points out, aristotelian categoricals support inferences that statistical generalizations don’t; in this respect they are more like universal than statistical generalizations. For example, just as “All Greeks are European” and “All Greeks are mortal” jointy entail “All Greeks are both European and mortal,” so also “The domestic cat has four legs” and “The domestic cat has a tail” jointly entail “The domestic cat has both four legs and a tail.” But if you tried to do the same thing with a statistical generalization you would be committing a fallacy; “most Americans are white” and “most Americans are female” don’t jointly entail “most Americans are both white and female.” (The same is true for “overwhelming majority;” due to the size of the majorities that are usually required to be “overwhelming,” concrete examples might take several conjunctions before they fail to preserve overwhelmingness, but the important thing is that they can fail to preserve it eventually. Whereas with universal generalizations and aristotelian categoricals you can preserve the same level of generality no matter how high you stack the conjunctions.)