Burden of proof

The burden of proof is a principle governing rational discourse that specifies which party in a disagreement bears the obligation to provide evidence or argument for their position. In the context of debates about God's existence, the question of who bears the burden of proof has significant consequences: if the burden falls on the theist, then atheism (understood as the absence of theistic belief) is the rational default position, and God's existence must be positively demonstrated before belief is warranted. If, on the other hand, belief in God is rationally permissible without argument — as defenders of reformed epistemology contend — then the burden-of-proof demand may rest on a contestable epistemological framework. The question is not merely procedural; it shapes the entire landscape of natural theology and determines what counts as a satisfactory conclusion to philosophical inquiry about the divine.^{1, 4}

The presumption of atheism

The most influential modern formulation of the burden-of-proof argument in philosophy of religion comes from Antony Flew's 1972 paper "The Presumption of Atheism." Flew distinguished between "positive atheism" — the affirmative claim that God does not exist — and "negative atheism," which he defined simply as the absence of theistic belief. His thesis was that rational debate should begin from negative atheism as its default position, just as legal proceedings begin from a presumption of innocence. The theist, like the prosecution, bears the burden of making a positive case; if no adequate case is made, the rational verdict is non-belief, not agnosticism or continued suspension of judgment. Flew argued that this presumption was not an arbitrary procedural rule but a requirement of rational methodology: one should proportion one's beliefs to the available evidence, and in the absence of evidence for a proposition, the rational stance is to withhold belief in it.¹

Flew's argument drew on a long tradition in epistemology. The principle that the party asserting a positive existential claim bears the burden of proof has roots in both legal reasoning and scientific methodology. In science, a hypothesis is not accepted until evidence supports it; the null hypothesis — the default assumption that no effect or entity exists until demonstrated — plays a structurally analogous role. Flew contended that the same logic applies to theism: because God is an entity whose existence is in question, the claim "God exists" requires support, while "I see no reason to believe God exists" requires none beyond pointing to the absence of persuasive evidence.^{1, 11}

The presumption of atheism has been challenged on several grounds. Stephen T. Davis argued that Flew's framing was tendentious because it redefined "atheism" to mean mere non-belief, thus making the presumption appear more modest than it actually is. Davis contended that in ordinary philosophical usage, atheism involves a substantive position — that there is no God — and that this claim carries its own burden of proof. On this view, the truly neutral starting position is agnosticism rather than negative atheism, and both theists and atheists bear symmetrical obligations to support their respective claims.¹⁷ Richard Swinburne, while accepting that theism bears a burden of proof, argued that this burden can be discharged through cumulative probabilistic reasoning: no single argument need be decisive, but the total evidence — cosmological, teleological, moral, and experiential — can render God's existence more probable than not.⁹

Extraordinary claims and Russell's teapot

A closely related principle, often summarized as "extraordinary claims require extraordinary evidence," reinforces the burden-of-proof argument. The idea is that the more antecedently improbable a claim, the stronger the evidence must be to make it credible. This principle has roots in David Hume's treatment of miracles in An Enquiry Concerning Human Understanding (1748), where Hume argued that testimony for a miraculous event must be weighed against the overwhelming regularity of natural law, and that it is almost always more rational to believe the testimony is mistaken than to believe the miracle occurred. Marcello Truzzi, the sociologist of science, later coined the exact phrase "extraordinary claims require extraordinary evidence," and Carl Sagan popularized it further.^{7, 9}

The claim that an omnipotent, omniscient, perfectly good, non-physical being created and sustains the universe is, by any measure, an extraordinary claim. It posits an entity unlike anything in ordinary experience, with attributes that raise well-known philosophical puzzles — the omnipotence paradox, the tension between omniscience and free will, the problem of evil. Critics argue that the evidence adduced for theism — cosmological arguments, design arguments, personal religious experiences — is not extraordinary enough to meet this threshold, particularly when each argument is subject to well-known objections and when naturalistic explanations can account for the relevant phenomena.^{7, 8}

Bertrand Russell's celestial teapot, introduced in a 1952 essay, provides what has become the most famous illustration of the burden-of-proof principle applied to religious claims. Russell asked his readers to suppose that he claimed a china teapot orbits the Sun between Earth and Mars, too small to be detected by any telescope. Nobody could disprove this claim, yet no reasonable person would regard inability to disprove it as a reason to believe it. Russell's point was that the burden of proof falls on the person asserting the teapot's existence, not on the person who disbelieves it, and that unfalsifiable claims about undetectable entities do not deserve credence merely because they cannot be refuted. The analogy was explicitly directed at theism: Russell argued that the claim "God exists" is in the same epistemic position as the claim about the teapot unless the theist can supply positive evidence that the atheist lacks.³

Theistic responses to the teapot analogy typically argue that the comparison is misleading. Swinburne and others have contended that God is not an arbitrary posit like a teapot in space but a being invoked to explain features of the universe — its existence, its order, its moral structure, the phenomenon of consciousness — that would otherwise lack adequate explanation. On this view, theism is an explanatory hypothesis, and the burden of proof is met (or at least substantially reduced) by the explanatory work that God's existence performs. The teapot, by contrast, explains nothing and is posited gratuitously. Whether this response succeeds depends on whether one judges the theistic explanations to be genuinely illuminating or merely to push the explanatory question back a step.^{9, 8}

Clifford's ethics of belief

The burden-of-proof argument receives moral as well as epistemic support from W. K. Clifford's famous 1877 essay "The Ethics of Belief." Clifford's central principle is stark: "It is wrong always, everywhere, and for every one, to believe anything upon insufficient evidence." His argument was not merely that evidenceless belief is irrational but that it is morally blameworthy, because beliefs are not private mental states with no consequences — they inform actions, shape character, and influence others. A shipowner who suppresses his doubts about his vessel's seaworthiness and sends it to sea "believing" it to be safe is guilty of the deaths of those aboard, even if the ship happens to arrive safely, because his belief was formed irresponsibly.²

Clifford's principle, applied to theism, implies that believing in God without sufficient evidence is not a matter of personal preference or pious humility but a violation of one's epistemic duty. If the evidence for God's existence is inadequate, then theistic belief is not merely unwarranted but wrong — ethically as well as epistemically. This is a stronger claim than Flew's presumption of atheism, which merely establishes the default position. Clifford's principle converts the absence of evidence into a positive obligation to disbelieve or at minimum to suspend judgment.^{2, 15}

William James's "The Will to Believe" (1896) is the classic response to Clifford. James argued that Clifford's principle was too restrictive: in cases where a question is "living, forced, and momentous" — where the options are genuine, the choice unavoidable, and the stakes significant — it can be rational to believe in advance of conclusive evidence. The question of God's existence, James contended, meets all three criteria. Waiting for sufficient evidence may itself carry costs if God exists, because one would have missed the goods — hope, moral motivation, community — that come with religious commitment. James was not arguing that one should believe whatever one wants; rather, he held that when evidence is genuinely indeterminate and the consequences of withholding belief are significant, the passional nature of the believer legitimately tips the balance. James's argument has structural affinities with Pascal's wager, though it differs in grounding the permission to believe in psychological and existential considerations rather than in a calculated gamble on expected utility.⁶

Evidentialism versus fideism

The burden-of-proof debate maps onto the broader epistemological divide between evidentialism and fideism. Evidentialism, as formulated by Richard Feldman and Earl Conee, holds that a belief is justified if and only if it is supported by the person's evidence. On a strict evidentialist account, all beliefs — including religious ones — must be proportioned to the available evidence, and a belief held without adequate evidential support is ipso facto unjustified. The burden of proof falls squarely on the theist to produce evidence or arguments sufficient to satisfy evidentialist standards.¹⁵

Fideism, by contrast, holds that religious belief is properly grounded in faith rather than in evidence or argument. In its strongest forms, fideism maintains that the demand for evidence is itself misguided when applied to religious commitments — that faith, by its very nature, transcends or precedes rational justification. Søren Kierkegaard's notion of the "leap of faith" and Karl Barth's insistence on the radical otherness of God's self-revelation are often cited as fideistic positions, though neither thinker fits the label neatly. Fideism sidesteps the burden-of-proof question by rejecting the evidentialist framework within which the question arises: if religious belief is not the kind of thing that requires evidential justification, then the demand for evidence reflects a category mistake rather than a legitimate epistemic standard.^{8, 16}

Most contemporary analytic philosophers of religion occupy a middle ground between strict evidentialism and thoroughgoing fideism. Swinburne, for example, is an evidentialist about theism who believes the cumulative evidence renders God's existence probable — he accepts the burden of proof and attempts to discharge it through a Bayesian assessment of the total evidence.⁹ Others, like Plantinga, reject the evidentialist framework not in favor of fideism but in favor of an alternative epistemology that denies evidentialism's starting assumptions altogether. The debate thus turns not only on whether the evidence for theism is sufficient but on whether the demand for evidence is itself the correct standard by which to evaluate religious belief.^{4, 5}

The reformed epistemology challenge

The most philosophically sophisticated challenge to the evidentialist burden of proof comes from Alvin Plantinga's reformed epistemology. Beginning with "Reason and Belief in God" (1983) and culminating in the warrant trilogy — Warrant: The Current Debate (1993), Warrant and Proper Function (1993), and Warranted Christian Belief (2000) — Plantinga argued that the demand for evidential justification of theism rests on a flawed epistemological assumption: classical foundationalism. Classical foundationalism holds that a belief is properly basic (justified without needing support from other beliefs) only if it is self-evident, incorrigible, or evident to the senses. Since belief in God meets none of these criteria, classical foundationalism entails that it can be rational only if supported by argument from basic beliefs. Plantinga argued that classical foundationalism is self-defeating — it fails its own test, since the principle itself is neither self-evident, incorrigible, nor evident to the senses — and that a more adequate epistemology should allow a broader range of properly basic beliefs, including belief in God.^{4, 10}

On Plantinga's account, belief in God can be properly basic for a person in the way that belief in other minds, belief in the reliability of memory, or belief in the external world is properly basic for most people. These beliefs are not held on the basis of arguments; they arise naturally from the functioning of our cognitive faculties in appropriate circumstances. Plantinga proposed, following John Calvin, that human beings possess a sensus divinitatis — a natural cognitive faculty that produces belief in God when triggered by certain experiences (the beauty of nature, the sense of moral obligation, awareness of one's own finitude). If this faculty exists and functions properly, the beliefs it produces are warranted without any need for inferential support, and the evidentialist demand for arguments is simply misdirected.^{5, 12}

This move effectively dissolves the burden-of-proof argument rather than meeting it on its own terms. If theistic belief is properly basic, then the question "What is your evidence for God?" is no more appropriate than the question "What is your argument for the existence of the external world?" Plantinga acknowledged that properly basic beliefs can be defeated — overridden by sufficiently strong counter-evidence — and he devoted substantial attention to potential defeaters for theistic belief, including the problem of evil and the challenge of religious diversity. But the default position, on his view, is that the theist is within her epistemic rights to believe without argument, and the burden falls on the critic to produce a successful defeater.^{5, 13}

Objections to the reformed epistemology response

The most persistent objection to Plantinga's approach is the "Great Pumpkin" objection, originally raised by Flew and others. If belief in God can be properly basic, what prevents someone from claiming that belief in the Great Pumpkin, or in voodoo, or in any arbitrary entity is properly basic? Plantinga responded that the question of which beliefs are genuinely properly basic is an empirical matter: it depends on whether the relevant belief-producing faculty exists and is functioning properly. The sensus divinitatis either exists or it does not, and if it does, it produces warranted beliefs; if it does not, then theistic belief is not properly basic. This answer, however, has struck critics as question-begging: whether the sensus divinitatis exists is precisely what is at issue, and Plantinga's model is explicitly conditional on God's existence. If God exists, theistic belief is probably warranted; if God does not exist, it is probably not warranted. The model thus defends the rationality of theistic belief only on the assumption that theism is true, which limits its persuasive force for those who do not already accept the conclusion.^{7, 14}

J. L. Mackie, in The Miracle of Theism, argued that the very notion of properly basic theistic belief is incoherent in a pluralistic context. Adherents of mutually incompatible religions — Christians, Muslims, Hindus, devotees of countless local and historical religions — could each claim that their beliefs are properly basic, produced by their own cognitive faculties in appropriate circumstances. Since these beliefs contradict one another, they cannot all be warranted, yet Plantinga's framework provides no neutral ground from which to adjudicate between them. The reformed epistemologist can say that only one set of religious beliefs is produced by properly functioning faculties and that the others result from cognitive malfunction (Plantinga attributes non-Christian religious belief to the noetic effects of sin), but critics see this as an ad hoc move that presupposes the very conclusion it is meant to establish.^{7, 5}

A further objection concerns the conditional nature of Plantinga's defense. His model demonstrates that if theism is true, then theistic belief is likely warranted. But this conditional conclusion leaves the truth of theism entirely open. An atheist can consistently accept Plantinga's conditional while denying its antecedent: "Yes, if God existed and had designed us with a sensus divinitatis, our theistic beliefs would be warranted — but God does not exist, and no such faculty exists." Plantinga's project is thus primarily defensive rather than offensive: it shows that there is no successful philosophical argument that theistic belief is irrational, but it does not provide a positive reason for anyone who does not already believe to begin believing. The burden-of-proof question, for those who find the evidentialist framework compelling, remains unanswered.^{5, 9, 14}

The contemporary landscape

The burden-of-proof question in philosophy of religion remains unresolved, in part because it depends on prior commitments in epistemology that are themselves contested. Those who accept some form of evidentialism — including most atheist and agnostic philosophers — continue to insist that theism bears a burden of proof that has not been adequately discharged. The arguments of natural theology, they maintain, are individually flawed and collectively insufficient: the cosmological arguments face the problem of why God would not also need a cause, the design arguments are undermined by Darwinian evolution and the problem of natural evil, the moral argument presupposes a contested metaethical theory, and religious experience is too variable and culture-dependent to serve as reliable evidence.^{7, 8}

Those who follow Plantinga and the reformed epistemology tradition reject the premise that theistic belief requires evidential support in the first place. On their view, the burden-of-proof demand reflects an epistemological imperialism — the illegitimate extension of evidentialist standards beyond their proper domain. William Alston's work on the epistemology of religious experience has reinforced this position by arguing that mystical perception can be a genuine source of justified belief, analogous to sense perception, even if it cannot be independently validated by non-mystical means.¹²

A third position, represented by philosophers like Nicholas Rescher, holds that the concept of burden of proof is context-dependent and pragmatic rather than fixed by any single epistemological principle. In some conversational contexts — a philosophy seminar, a public debate — the theist may appropriately bear the burden because she is asserting an existence claim. In other contexts — a community where theism is the default — the burden may fall on the atheist who dissents from the prevailing consensus. Rescher argues that presumptions are pragmatic instruments for managing inquiry under uncertainty, not timeless logical truths, and that the question of who bears the burden of proof is always relative to the specific dialectical situation.¹¹

The persistence of the debate reflects a genuine philosophical impasse. Evidentialism and reformed epistemology represent two fundamentally different pictures of what rational belief requires, and neither can refute the other without presupposing its own framework. The burden-of-proof argument remains one of the most consequential issues in philosophy of religion because it determines not just who must argue for what but what the entire enterprise of evaluating theistic claims can reasonably be expected to accomplish. If the evidentialist is right, then the absence of conclusive arguments for God's existence is itself a significant datum — a reason to withhold belief. If the reformed epistemologist is right, then the demand for such arguments was misguided from the start, and the rationality of theistic belief does not depend on whether philosophers can produce compelling proofs.^{5, 9, 15, 16}