Hume's argument from design: The argument from experience

It will help if we make a little clearer the structure of Cleanthes' argument. The argument aims to conclude that the universe is an artifact, that is, something made by an intelligent designer.

First he introduces the idea of an argument from experience.

When two species of objects have always been observed to be conjoined together, I can infer, by custom, the existence of one whenever I see the exis­tence of the other; and I call this an argument from experience.

Thus, suppose I regularly find a strong cheese in the kitchen when I smell a particular odor in the dining room. The odors are one “species” (i.e., sort) of thing; the cheeses are another. Provided that this is so, when I experience that odor again, I may infer, by way of the argument from experience, that there is cheese in the kitchen. This is reasonable even if, occasionally, the cheese has been eaten and only the odor remains. And it is reasonable even if, sometimes, the cheese is present but the odor is not. So the real principle is more like this:

AE: If, usually, when you have experienced an A in the past, you experience a B in association with it, then, if you experi­ence another A, you may infer that there's likely to be a B in association with it.

When two things are related in this way, so that when you have experienced an A in the past, you have usually experienced a B in association with it, we can say that there is a strong empirical cor­relation between A's and B's.

If the principle AE is right, then Hume's statement of the argu­ment from experience is too strong: he shouldn't have said “always” (“very often” would have done), and he shouldn't have required them to be conjoined, since, as we saw, it's the fact that A's are asso­ciated with B's (odors with cheese), not that B's are associated with A's (cheese with odors) that matters.

You will notice that the principle AE looks awfully like a state­ment of the validity of enumerative induction, which I defined in 4.9 as the process of arguing from many cases of As that are B's to the conclusion that all A's are B's. (Since Hume was interested, as we know, in induction, this isn't too surprising, of course!) But AE can be used more widely than enumerative induction, because it doesn't require that there are no counterexamples, that is, no A's that are not B's. In this sense, AE is a stronger principle than enumerative induction. Given the difficulties with enumerative induction that we have already discussed, that is some grounds for concern about rely­ing on AE. Nevertheless, as I say, AE looks reasonable enough. Someone who said that they thought there was a strong cheese in the kitchen because the odor they could smell in the dining room was just like the odor that had been associated with strong cheeses in the past would not normally be thought to be unreasonable!

What Cleanthes does—this is Hume's second step—is to argue that there is a strong empirical correlation between exhibiting a mutual adaptation of parts and being an artifact. The argument goes:

1.      The world contains many things that exhibit a mutual adaptation of parts.

2.      Some of these things—machines, for example—we know a posteriori to be made by intelligent designers. Call these the “known artifacts.”

3.      Others of them—eyes, for example—we do not know not to be made by intelligent designers. Call these the “possible artifacts.”

So:

4.      There is a strong empirical correlation between exhibiting a mutual adaptation of parts and being an artifact.

It is now easy, putting the results of the first two steps of the argu­ment together, for Cleanthes to produce an argument from experi­ence whose conclusion is that the universe is an artifact.

The universe exhibits a mutual adaptation of parts,

which is, of course, just a version of the harmony of nature.

In the Dialogues Concerning Natural Religion, there is, along with Cleanthes, who proposes the argument from design, and Demea, the character who defends the ontological argument, a third character, named Philo, who, though he is a religious believer, is also a philosophical skeptic. Philo objects to Cleanthes' argument from experience on the grounds that the evidence for the harmony of nature is not very good. This is not because he thinks that nature is not harmonious; Philo is a skeptic, so he is more inclined to insist on what we don't know than on what we do. But he thinks we haven't really got enough evidence about the universe as a whole to suppose that it exhibits a mutual adaptation of parts. As Philo says: “A very small part of this great system, during a very short time, is very imperfectly discovered to us.”

I have already insisted that it is far from clear what the content is of the claim that the universe displays a mutual adaptation of parts. Nevertheless, we have been offered examples of things that do: watches and eyes among them. Since it is not clear how to apply this idea to other cases, we do not really know how many others of the things in the universe are also harmonious. Cleanthes' talk of a uni­verse “subdivided into an infinite number of lesser machines” sug­gests that he thinks that most things are. But now we can follow Philo's hint about the smallness of our sample and catch Cleanthes on the horns of a dilemma.

Suppose most things we know do indeed display mutual adapta­tion. Still, only a very tiny, almost infinitesimal proportion of the things in the world are known artifacts. The vast majority of them are just possible artifacts. And the argument from experience seems very far from compelling when you have established that A's are B's in only a tiny sample of the available cases of A's. (Imagine a world containing a myriad of swans and someone who claims that they are all white on the basis of examination of a very few.)

So suppose it's false that most things display mutual adaptation. Then there seems no reason to grant that the universe as a whole displays mutual adaptation of parts.

Either way, the argument fails.

8.12