The charge is familiar and, in its popular form, often persuasive. Cosmological and contingency arguments are said to reason as follows: every thing in the universe has a cause or explanation; therefore the universe as a whole has a cause or explanation. But this looks like the fallacy of composition. The fact that every tile in a mosaic is small does not mean the mosaic is small. The fact that every human being has a mother does not mean the human race has a mother. Likewise, even if every event or object within the universe has some cause or explanation, it does not follow that the totality of those events or objects has a further cause or explanation.
This objection has a serious pedigree. David Hume pressed versions of it in his criticisms of cosmological reasoning. Bertrand Russell made the point sharply in his 1948 BBC debate with Frederick Copleston. Copleston defended a Leibnizian contingency argument: the existence of contingent things calls for an explanation, and that explanation cannot ultimately be another contingent thing. Russell resisted the move from explaining the members of the series to explaining the series as a whole. Russell's point was not merely that a large object can lack a cause. He denied that one may move from explanations of particular things to an explanatory demand for the totality unless a further principle, such as the principle of sufficient reason, has been defended. On his view, "the universe" need not name an additional object over and above all the things and facts it includes, and total physical reality may be where explanation stops.
The argument steelmanned
The strongest version of the objection is not the crude claim that wholes never share properties with their parts. Sometimes they do. If every brick in a wall is made of clay, the wall is made of clay. If every member of a group is under six feet tall, the group may still not be under six feet tall, because groups do not have height in the same way individuals do. The issue is not whether part-to-whole reasoning is always invalid. The issue is whether it is valid in this particular case.
The atheist or naturalist critic is making three connected points.
- First, "has a cause" or "has an explanation" may be a property that applies to events or objects within the universe, not necessarily to the universe as a totality. Many causal concepts are drawn from intra-world relations: one event earlier in time produces another event later in time; one physical state gives rise to another. It is not obvious that the same concept applies to the whole spacetime order.
- Second, even if each member of a collection has an explanation, it does not follow that the collection has an explanation over and above those member-level explanations. If every person in a room is present because they were invited by another person, we may still ask why there is a gathering at all; but in other cases, once the individual explanations are given, no further question remains. The proponent of the cosmological argument must show which kind of case the universe is.
- Third, the theist may be smuggling in a controversial principle of sufficient reason: that every contingent fact, or every totality of contingent facts, must have an explanation. But this is precisely what many atheists deny. J. L. Mackie, Graham Oppy, and other critics have argued that the demand for an explanation of the whole cosmos is not self-evident. The universe, or the total physical reality, may be ultimate.
Put this way, the objection is not a cheap debating trick. It asks the defender of cosmological reasoning to justify the crucial step. It is not enough to say that every item inside the universe has a cause and then infer that the universe has one too. The whole may not be the same sort of thing as its parts. The demand for an explanation of the whole requires an argument, not an assumption.
Where the objection succeeds
The objection succeeds against many popular presentations of cosmological arguments. If someone argues, "Everything has a cause; therefore the universe has a cause; therefore God caused the universe", the argument is vulnerable on several fronts. It raises the familiar question, "What caused God?" It seems to treat the universe as just one more object inside a larger container. It also appears to infer from the causal dependence of ordinary things within the universe to the causal dependence of the universe itself.
Serious defenders of cosmological arguments usually know this. Classical and contemporary versions are typically more careful. They do not usually say "everything has a cause". The kalam cosmological argument, associated in contemporary philosophy with William Lane Craig, says that whatever begins to exist has a cause, and then argues that the universe began to exist. Leibnizian contingency arguments, defended in different forms by philosophers such as Alexander Pruss, Robert Koons, Joshua Rasmussen, and by Richard Gale and Alexander Pruss in their modal cosmological argument, tend to say that contingent reality requires an explanation. Thomistic or Aristotelian arguments focus on present dependence, change, actuality and potentiality, or hierarchical causal series, not simply on a count of caused things.
That distinction matters. The fallacy-of-composition objection is strongest against arguments that really do move from "each part has a cause" to "the whole has a cause" without further support. It is weaker against arguments that include an independent premise about beginnings, contingency, dependence, or the intelligibility of contingent facts. Whether those premises are true is another matter. But they are not simply the fallacy of composition.
The strongest reply
The strongest theistic reply is to deny that the serious cosmological argument depends on illicit composition. The defender can say: "I am not arguing that because every part of the universe has a cause, the universe must have a cause. I am arguing that the totality of contingent reality is itself contingent, or that the existence of any contingent reality calls for explanation. The issue is not a property being transferred from parts to whole. The issue is whether contingent existence, as such, can be ultimate."
Consider the Leibnizian line. A tree exists, but it could have failed to exist. Its existence is explained by seeds, soil, sunlight, biological history, and so on. Those explanations invoke further contingent realities. The question then becomes whether the whole network of contingent realities can be explained merely by citing other members of the network. If every explanation remains inside the set of contingent things, then, the theist argues, we still have not explained why there is any such set at all.
This is not obviously fallacious. Suppose every book in a library was copied from another book in the same library. That might explain why each book has the text it has, but it would not explain how the information entered the library system in the first place. The analogy is imperfect, as all analogies are, but it illustrates the point: explaining each member by reference to another member may fail to explain the existence or character of the whole system.
The defender can also challenge Russell's claim that once every particular thing is explained, no further explanation is needed. That depends on what the explanations are like. If each thing is explained by another contingent thing, and the entire explanatory chain never leaves the domain of contingent things, then the critic has not obviously provided a complete explanation. They have described an infinite or circular pattern of dependence. Whether such a pattern is explanatorily satisfying is precisely the point under dispute.
On this reading, the cosmological argument is not: "Every member has property P; therefore the whole has property P." It is closer to: "If contingent reality exists, and if contingent facts cannot be ultimate brute facts, then explanation must terminate in something non-contingent." That conditional form makes the real burden visible. The defender still has to argue for the sufficient-reason or anti-brutality premise; the conclusion does not follow from composition alone. But neither is it the same as saying that because every sheep in a flock has one mother, the flock has one mother.
Does contingency transfer to the whole?
A central question is whether a totality of contingent things is itself contingent. Many theists answer yes, but the argument cannot simply be: each member is contingent, so the whole is contingent. That would invite the very objection under discussion. The stronger claim is modal and explanatory: if no member of the totality exists by necessity, and if the totality contains nothing whose nature guarantees the existence of concrete reality, then the totality appears not to explain its own existence. From there, the defender must argue for a principle connecting contingent reality with the need for explanation.
The atheist or naturalist critic has replies. One can deny that the totality is an additional object with modal properties over and above its members. One can say that "the universe" is a way of referring to all concrete reality, not a thing alongside the things it includes. Or one can press a modal distinction: from "for each particular physical thing, that thing could have failed to exist" it does not follow that "all physical reality could have failed to exist." Perhaps necessarily there is some physical reality or other, even if every particular physical state, object, or law-like arrangement could have been different.
This last move is important. It shows that the composition objection is not the whole debate. A naturalist might accept that individual objects are contingent while maintaining that the total physical order exists necessarily, or that there is no coherent contrast between the existence and non-existence of all reality. Graham Oppy, for example, has defended naturalistic views on which the natural world is the ultimate explanatory terminus. Such views may have costs, but they are not refuted merely by pointing out that trees, planets, and particles are contingent.
The defender of the contingency argument, in turn, may press for clarification. If the physical order is necessary, in what sense is it necessary: logical, metaphysical, or merely nomological? If the answer is not necessity but brute ultimacy, then the debate shifts. The naturalist is not showing that the demand for explanation is incoherent; they are denying that it applies without exception at the level of ultimate reality. That is a possible position, but it is a substantive metaphysical commitment rather than a straightforward result of science.
Causation and explanation
The objection is especially forceful against causal cosmological arguments when "cause" is left at the level of ordinary temporal production. Causes, as ordinarily understood, occur within time. If time is part of the universe, then asking for a cause of the universe can sound like asking for an earlier event before time existed. That worry is not trivial. Any argument for a cause of the universe must clarify whether it means temporal causation, atemporal dependence, grounding, or some other relation.
However, not all cosmological arguments require temporal causation. Some speak of explanation, grounding, dependence, or sustaining causes. A mathematical truth can explain a result without occurring earlier than it. A structural feature can explain why a system behaves as it does without being temporally prior to every part of the system. Whether such notions can be extended to the existence of the universe is disputed, but the extension is not automatically meaningless.
Nor should either side pretend that contemporary cosmology settles the issue. Big Bang cosmology does not by itself prove creation from nothing, and speculative models without a first moment do not by themselves show that no ultimate explanation is needed. Physics investigates the structure and history of the observable universe, and perhaps deeper physical reality. The metaphysical question of why there is any contingent reality, if there is, is not answered simply by naming an earlier physical state or a more fundamental physical law.
What the objection establishes
The fallacy-of-composition objection establishes an important negative point: a cosmological argument cannot simply infer from the caused or explained status of things within the universe to the caused or explained status of the universe as a whole. The theist owes an argument that the relevant explanatory principle applies to total reality, not only to its internal members. Russell's challenge to Copleston remains powerful at exactly this point.
But the objection does not establish that all cosmological or contingency arguments fail. The best versions are aware of the problem and try to avoid it. They argue from contingency, dependence, beginnings, or the principle of sufficient reason, not from a naive transfer of properties from parts to whole. To defeat those arguments, one must engage those premises directly: Is the principle of sufficient reason true? Can there be brute contingent facts? Can an infinite regress of explanations be complete? Is the totality of contingent things itself contingent? Is "necessary physical reality" a coherent stopping point?
The honest conclusion is modest. The composition objection is a decisive criticism of simplistic cosmological arguments. It is also a useful warning against treating "the universe" as if it were just another object inside the universe. But it is not, by itself, a refutation of serious contingency arguments. It forces the debate onto deeper ground: whether reality as a whole must be intelligible through some ultimate explanation, or whether explanation can legitimately end with the natural order itself. That is where the real disagreement lies.