Legos, God, and the Fallacy of Composition
Both critics and defenders of arguments for the existence of God as an Uncaused Cause often assume that such arguments are essentially concerned to explain the universe considered as a whole. That is true of some versions, but not all. For instance, it is not true of Aquinas’s arguments, at least as many Thomists understand them. For the Thomist, you don’t need to start with something grand like the universe in order to show that God exists. Any old thing will do—a stone, a jar of peanut butter, your left shoe, whatever. The existence of any one of these things even for an instant involves the actualization of potencies here and now, which in turn presupposes the activity of a purely actual actualizer here and now. It involves the conjoining of an essence to an act of existence here and now, which presupposes a sustaining cause whose essence and existence are identical. It involves a union of parts in something composite, which presupposes that which is absolutely simple or incomposite. And so forth. (For the details of this, see Aquinas, especially chapter 3.)
Criticisms of First Cause arguments that assume that what is in question is how to explain the universe as a whole are therefore irrelevant to Aquinas’s versions. Still, those versions which are concerned with explaining this are also important. One objection often raised against them is that they commit a fallacy of composition. In particular, it is claimed that they fallaciously infer from the premise that the various objects that make up the universe are contingent to the conclusion that the universe as a whole is contingent. What is true of the parts of a whole is not necessarily true of the whole itself: If each brick in a wall of Legos is an inch long, it doesn’t follow that the wall as a whole is an inch long. Similarly, even if each object in the universe is contingent, why suppose that the universe as a whole is?
There are two problems with this objection. First, not every inference from part to whole commits a fallacy of composition; whether an inference does so depends on the subject matter. If each brick in a wall of Legos is red, it does follow that the wall as a whole is red. So, is inferring from the contingency of the parts of the universe to that of the whole universe more like the inference to the length of the Lego wall, or more like the inference to its color? Surely it is more like the latter. If A and B are of the same length, putting them side by side is going to give us a whole with a length different from those of A and B themselves. That just follows from the nature of length. If A and B are of the same color, putting them side by side is not going to give us a whole with a color different from those of A and B themselves. That just follows from the nature of color. If A and B are both contingent, does putting them together give us something that is necessary? It is hard to see how; indeed, anyone willing to concede that Lego blocks, tables, chairs, rocks, trees, and the like are individually contingent is surely going to concede that any arbitrary group of these things is no less contingent. And why should the inference to the contingency of such collections stop when we get to the universe as a whole? It seems a natural extension of the reasoning, and the burden of proof is surely on the critic of such an argument to show that the universe as a whole is somehow non-contingent, given that the parts, and collections of parts smaller than the universe as a whole, are contingent.
So, that is one problem. Another problem is that it isn’t obvious that the sort of cosmological argument that takes as a premise the contingency of the universe needs to rely on such part-to-whole reasoning in the first place. When we judge that a book, an apple, or a typewriter is contingent, do we do so only after first judging that each page of the book, each seed in the apple, each key of the typewriter, and indeed each particle making up any of these things is contingent? Surely not; we can just consider the book, apple, or typewriter itself, directly and without reference to the contingency of its parts. So why should things be any different for the universe as a whole?
If anything, it is certain critics of the sort of argument in question who seem more plausibly accused of committing a fallacy of composition. Consider this famous passage from David Hume’s Dialogues:
"Did I show you the particular causes of each individual in a collection of twenty particles of matter, I should think it very unreasonable, should you afterwards ask me, what was the cause of the whole twenty. This is sufficiently explained in explaining the cause of the parts."
(Paul Edwards makes a similar objection—see the “five Eskimos” example in this famous article.)
The reasoning couldn’t be more plain: If you explain each part of a collection, you’ve explained the whole. Therefore (so this sort of objection to the kind of cosmological argument in question continues) if we can explain each individual thing or event in the universe as the effect of some previous thing or event in the universe, we’ve explained the whole collection of things or events, and needn’t appeal to anything outside the universe. And yet, to identify the immediate efficient cause of each thing in a collection simply is not necessarily to explain the collection as a whole. If a certain book exists because it was copied from an earlier book, the earlier book existed because it was copied from a yet earlier book, that book existed because it was copied from a still earlier book, and so on, we will hardly have provided a sufficient explanation of the series of books if we suppose that it either has extended backward into the past to infinity or that via time travel it forms a causal loop. So, hasn’t Hume himself committed a fallacy of composition?
A defender of Hume might reply as follows: It is only when each part of a collections has been sufficiently explained that the Humean claims it follows that the whole collection has been explained; and in the counterexamples in question (the book example and others of the sort explored in the previous post) each part clearly hasn’t been sufficiently explained but only partially explained (because, say, the origin of the information contained in the book still needs to be explained). So (the proposed reply continues) the Humean would not be committed to saying, falsely, that the whole collection has been explained in such cases.
This saves the Humean critique from committing the fallacy of composition, but only at the cost of making it question-begging. For a defender of the sort of cosmological argument we’ve been discussing could happily agree that if each part of a collection has been sufficiently explained, then the whole collection has been explained as well. He just thinks that to identify an immediate contingent cause for each contingent thing or event in the universe is not to give a sufficient explanation of it. If the Humean disagrees, then he needs to give some reason why identifying such a cause would be sufficient. Merely to assert that it would be sufficient—which is all Hume does, and which is all that is done by those who quote Hume as if he had made some devastating point—simply assumes what is at issue.
Note: Our goal is to cultivate serious and respectful dialogue. While it's OK to disagree—even encouraged!—any snarky, offensive, or off-topic comments will be deleted. Before commenting please read the Commenting Rules and Tips. If you're having trouble commenting, read the Commenting Instructions.