Arguments based on observation - Hume, evolution and chance
Learn Hume's challenges to design and cosmological reasoning, distinguish evolution from random assembly, and evaluate what observational arguments can still support.
Hume's challenge to the design analogy
Hume's Dialogues Concerning Natural Religion presents arguments through speakers, rather than a single uninterrupted declaration of his own conclusion. Cleanthes defends a design inference; Philo challenges it; Demea offers a different theological approach. Attributing every speaker's statement directly to Hume obscures the debate.
In Part II, Philo questions the analogy between human artefacts and the universe. We recognise a house as designed because we have experience of houses and their production. We have no comparable experience of universes being made. A universe differs greatly from a machine, so the inference from one to the other is weaker than an ordinary inference between similar artefacts.
Even granting design, the evidence does not establish every divine attribute. A finite effect supports only the power needed for that effect, rather than infinite power. Human products often have several makers, imperfect makers or makers learning from earlier attempts. The analogy therefore leaves alternatives to one perfect God. Nature can also resemble an organism that grows more than a manufactured machine.
This is an underdetermination objection: the evidence could fit different causes. A defender replies that organised function is the relevant similarity and may still make intelligence probable. The challenge then becomes comparative: why prefer a single unlimited intelligence to the other possibilities? Design's immediate conclusion and classical theism must be distinguished.
Hume wrote before Darwin's account of natural selection. His criticisms do not depend on that later mechanism, although evolution supplies a substantive natural alternative to the kind of inference he challenges.
Hume's challenge to causal explanation
Hume challenges the claim that reason alone establishes a necessary cause for every existence. In Part IX of the Dialogues, Cleanthes argues that what we can conceive as existing can also be conceived as not existing; existence is not demonstrated by logical necessity in the proposed way. This challenges the inference to a necessarily existing being, not simply the claim that events often have causes.
Hume's wider analysis of causation distinguishes observing one event regularly followed by another from perceiving a necessary connection. Our expectation of the effect grows through experience. We cannot simply transfer that expectation beyond all possible experience and claim certain knowledge of the cause of the entire universe.
He also challenges the need for a cause of a whole collection once each member is explained. In Part IX, explaining the causes of individual members is offered as an alternative to demanding an extra cause of the collection. This supports a composition objection to some cosmological arguments. It does not automatically answer Aquinas' account of present derivative causation: the defender must say why dependence of the order creates a further explanatory need.
Aquinas can respond that his argument is metaphysical, rather than an induction from many observed beginnings to an unobserved beginning. Yet that response still requires justification of actualisation, causal dependence and a non-dependent source. Hume presses whether those concepts establish the proposed reality or extend our explanatory habits beyond their warrant.
The result is a dispute over the limits of explanation. A believer may find an ultimate ground more intelligible than a brute fact; a critic may judge that we lack grounds to specify such a being. Preference alone is insufficient on either side: the argument must explain why its stopping point is justified.
Evolution is not random assembly
Evolutionary explanation challenges the inference that complex biological adaptation must be the direct arrangement of an intelligent designer. Natural selection works on heritable variation: some variants survive and reproduce more successfully in their environment, making the relevant traits more common over generations. The process has no foresight of a finished organism.
Variation is not directed towards what an organism needs; in that sense mutations can be described as random with respect to adaptive advantage. Natural selection, however, is not a purely random choosing process. Differences in reproductive success depend on interactions between traits and environment. Cumulative retention can build adaptations through intermediate stages; it does not require the chance assembly of a complete eye in one event.
This directly challenges Paley's inference from adapted parts to a maker. Watches do not form reproducing populations with inherited variation in the way organisms do. Paley's self-reproducing watch thought experiment addresses reproduction, but reproduction alone is not the whole evolutionary account: variation, differential success and accumulation change the explanation.
Richard Dawkins' blind-watchmaker description stresses this contrast between organised outcomes and a process without foresight. It does not mean that evolution is random assembly or that one biological mechanism settles every question about the origin of laws or existence. Hume's earlier argument and the later evolutionary challenge are historically distinct.
A theist can accept evolution and propose that God creates through natural processes. This shows logical compatibility, but compatibility is different from evidence: if evolution explains adaptation without specifying a designer, adaptation by itself may supply less independent support for design. A broader argument about why there is a law-governed world requires separate premises.
Weighing evidence and conceptual certainty
Observational and conceptual arguments have different strengths and vulnerabilities. A posteriori reasoning starts from publicly accessible features of the world, but its explanations can be challenged by new evidence and alternative causes. A priori reasoning aims at necessity through concepts and inference, but certainty depends on the truth of its premises and its right to move from a concept to reality.
For design, a serious defence against chance must engage natural selection rather than a random-assembly caricature. It may shift to the regularity of laws or the ultimate conditions of life. That can be a legitimate development, but it changes what is being explained and does not preserve every step of Paley's original analogy untouched.
For cosmological reasoning, Hume's objection to observed causation is strongest against an unqualified extrapolation beyond experience. A dependence argument can reply that it addresses the metaphysical status of causes. The remaining disagreement concerns whether that status genuinely requires a transcendent ground or is an assumption built into the explanation.
An evaluation should therefore separate three outcomes. An argument can fail as a demonstration, remain a reasonable probabilistic consideration, or provide little independent evidence once alternatives are considered. Those are different judgements. A priori reasoning is not automatically superior because it aims at certainty; a posteriori reasoning is not automatically superior because its starting point is visible.
One defensible conclusion is that evolutionary explanation substantially weakens design from biological adaptation, while ultimate dependence remains a live philosophical question. That conclusion neither proves God nor proves that reality has no ultimate explanation. The subsequent ontological lessons examine whether reasoning from the concept of God can achieve what observation has not conclusively established.