The burden of proof and the presumption of atheism

The argument is familiar, but it is often stated too quickly. In its strongest form, it is not the crude claim that "science has not found God, therefore God does not exist." Nor is it the stronger claim, sometimes heard in online debate, that only laboratory-testable propositions can be known. That stronger view collapses into a kind of verificationism or scientism, and many serious atheist philosophers reject it. Logic, mathematics, moral reasoning, historical inference, memory, testimony, and the philosophy of science itself are not established by controlled experiment in the ordinary sense.

The more defensible argument is evidentialist and procedural. Belief should be proportioned to evidence. The person who asserts that God exists is making a substantive existential claim, and so must provide adequate reasons for accepting it. The person who withholds belief is not required to disprove God, just as one is not required to disprove Bertrand Russell's famous celestial teapot orbiting somewhere between Earth and Mars. Antony Flew gave this idea a philosophical label in his essay "The Presumption of Atheism": he proposed a presumption of atheism where "atheism" means, not the positive thesis that God does not exist, but the absence of theistic belief until reasons are supplied. That usage should be distinguished from another common academic convention, where "atheism" means the belief that there is no God and "agnosticism" means suspension of judgment. W.K. Clifford's severe evidentialism is often invoked in the background: it is wrong, he argued, to believe on insufficient evidence.

Stated this way, the argument has real force. It rightly rejects a common reversal of burden: "You cannot prove God does not exist, therefore belief is justified." That is bad reasoning. The inability to refute a claim does not make the claim credible. Countless incompatible possibilities cannot be disproven. A rational inquirer needs positive grounds, not merely an absence of disproof.

What the argument establishes

The argument establishes at least three important points.

  • First, disbelief need not be dogmatic. A person can be an atheist in the modest sense of lacking belief in God without claiming to possess a demonstration that God does not exist.
  • Second, theists who want public assent owe public reasons. If someone asks others to accept a claim about reality, it is not unfair to ask what evidence supports it.
  • Third, mystery is not evidence. Gaps in present knowledge do not automatically support theism. An unexplained phenomenon is just that: unexplained.

Professional atheist philosophers often press the issue in a more careful way than popular polemicists. Graham Oppy, for example, typically frames the debate as a comparison between worldviews: which theory best explains the total evidence with the least theoretical cost? Paul Draper has argued that the biological role and distribution of pain and pleasure are more expected on a "hypothesis of indifference" than on theism. William Rowe's evidential argument from evil similarly treats apparently pointless suffering as evidence against an all-powerful and perfectly good God, without claiming a deductive disproof. J.L. Schellenberg's argument from divine hiddenness does not merely say "I see no evidence"; it argues that a perfectly loving God would make relationship with God available to all nonresistant persons, and that the existence of nonresistant nonbelief is therefore evidence against such a God. These are not slogans. They are serious arguments.

Where the slogan overreaches

The slogan "the burden of proof lies with the theist" becomes less clear when it is treated as a complete philosophy of rational belief. Burdens of proof are not fixed by grammar alone. They are partly contextual. In a courtroom, the prosecution carries a special burden because of the moral risks involved. In a debate, the person trying to persuade others carries a burden. In inquiry, any person who advances a substantive conclusion carries some burden, including the atheist who says not merely "I do not believe" but "there is no God" or "theism is irrational."

This distinction matters. Minimal atheism, understood as withholding assent, requires less defense than positive atheism, understood as the claim that God does not exist. But minimal atheism also establishes less. It does not show that God probably does not exist. It only says that, from the subject's perspective, belief has not yet been warranted.

There is also an ambiguity in the phrase "positive evidence." If it means "repeatable scientific evidence," then the standard is too narrow. Many rational beliefs are not formed by repeatable experiment. We know many historical claims by documents, testimony, background knowledge, and inference to the best explanation. We accept mathematical truths by proof, not by microscopes. We rely on memory and perception before we have scientific evidence for their reliability. Even science depends on non-scientific assumptions: logic, induction, measurement norms, and the trustworthiness of cognitive faculties under suitable conditions.

So the serious evidentialist should not say that only scientific evidence counts. The better claim is that the evidence for God must be accessible to critical assessment and strong enough to justify the conclusion. That evidence may include philosophical argument, historical reasoning, experience, moral facts, consciousness, cosmology, or the structure of the universe. Whether any of these succeed is a further question. They cannot be dismissed merely because they are not laboratory results.

The teapot analogy and its limits

Russell's teapot is powerful because it exposes a bad move: invent an undetectable object, then demand disproof. But analogies can illuminate without being exact.

A celestial teapot is a contingent physical object within the universe. It has no explanatory role. It is introduced ad hoc. It is antecedently improbable because it fits poorly with what we know about teapots, rockets, and the history of human space activity. By contrast, classical theism is not normally proposed as one more object inside the universe. It is proposed as an account of why there is a universe at all, why contingent things exist, why the universe is intelligible, why consciousness and moral normativity exist, or why there is something rather than nothing.

This does not mean theism is true. It means the teapot comparison can become misleading if it treats all unseen claims as epistemically identical. Electrons, numbers, moral duties, other minds, possible worlds, and teapots are not assessed by one uniform evidential test. A claim's prior plausibility depends partly on its content, explanatory role, coherence, and relation to background knowledge.

The atheist can reply that God is even more metaphysically costly than the teapot, not less. That reply has weight. Even if classical theists deny that God is complex in the way physical objects are complex, a perfect, immaterial, necessary creator remains a major ontological commitment. But then the discussion has moved beyond burden-of-proof rhetoric into comparative metaphysics: which worldview has the lower theoretical cost and greater explanatory power?

Absence of evidence and evidence of absence

Another point needs care. "Absence of evidence is not evidence of absence" is often repeated by theists, but it is only sometimes true. If a claim predicts that we should find certain evidence, and we do not find it after looking carefully, the absence is evidence against the claim. If someone says there is an elephant in the room, not seeing, hearing, or smelling it is evidence that there is no elephant in the room.

So the atheist may reasonably argue: if a personal God exists, especially one who desires relationship with human beings, we should expect clearer public evidence or at least less sincere nonbelief. This is the force of divine hiddenness arguments. Likewise, if God is perfectly good and powerful, the scale, intensity, and distribution of suffering may be evidence against theism. These arguments are far stronger than the bare claim "God has not been scientifically detected." They should not be dismissed by saying only that human beings lack God's perspective, or that some suffering may produce goods. Those points may form part of a theistic reply, but by themselves they do not erase the evidential pressure created by apparently gratuitous suffering and nonresistant nonbelief.

But the evidential force depends on what theism is claimed to predict. Some forms of theism expect miracles, answered prayers, or unmistakable public signs. Those versions expose themselves to more empirical pressure. More classical or philosophical forms of theism may make fewer specific empirical predictions, and so are not refuted simply by the lack of a laboratory test. The cost, however, is that they may also become less easy to confirm by public evidence.

The strongest atheist reply

The strongest reply from the evidentialist atheist is this: even if theism is not a teapot-like claim, it is still a claim. If its defenders appeal to cosmological arguments, moral arguments, consciousness, fine-tuning, or religious experience, those arguments must be examined. And when examined, they remain deeply contested. Cosmological arguments face disputes about causation, contingency, necessity, and explanation. Fine-tuning arguments face questions about probability measures, selection effects, multiverse hypotheses, and whether "design" explains more than it assumes. Moral arguments face robust secular accounts of moral realism or anti-realism. Religious experience is weakened, at least publicly, by religious diversity and the possibility of psychological explanation. Meanwhile, evil and hiddenness appear to count significantly against theism.

This reply succeeds in an important respect. The atheist is not obliged to accept theism merely because theism is meaningful, philosophically sophisticated, or emotionally satisfying. If, after serious assessment, the evidence seems insufficient, suspension of theistic belief is rational; and if one also judges evil, hiddenness, incoherence objections, or explanatory comparisons to make theism unlikely, then positive atheism can be rational too. The theist need not agree with that verdict, but should concede that it is a serious philosophical position rather than a mere failure to understand theistic arguments.

But the reply does not establish everything that the popular slogan often suggests. It does not show that the only rational starting point is a naturalistic worldview. It does not show that all non-scientific arguments are worthless. It does not show that theism is equivalent to fairies or teapots. And it does not show that absence of decisive public proof is itself a decisive proof of absence, except where the relevant kind of theism would lead us to expect such proof.

A more precise conclusion

The burden-of-proof argument is best understood as a rule of intellectual responsibility, not as a disproof of God. It says: do not believe merely because a claim has not been refuted; require adequate reasons before assent; and assign the burden to the person asking others to accept a substantive claim.

That principle is sound. Its overextensions are not. If atheism means withholding belief until the case for God is strong enough, it can be a rational position. If atheism means the positive claim that no God exists, it too requires argument. If the demand for evidence means "give publicly assessable reasons," it is fair. If it means "give only scientifically testable evidence," it is an indefensible narrowing of knowledge.

So the critique of the slogan is limited but important. The theist does carry a burden when presenting theism as true. The atheist who merely withholds assent does not carry an equal burden to disprove God. Yet the slogan by itself does not settle the question. It clears away bad defenses of theism, but it does not eliminate the need to compare the actual arguments on both sides.