Faith and the charge of epistemic arbitrariness
The argument under examination is familiar in atheist and sceptical circles, but it is not merely a slogan. In its strongest academic form it belongs to a wider evidentialist and anti-fideist tradition associated with thinkers such as W. K. Clifford, Bertrand Russell, Antony Flew, J. L. Mackie, and, in contemporary epistemology, the evidentialism of Richard Feldman and Earl Conee. Professional critics of theism such as Graham Oppy, Paul Draper, and J. L. Schellenberg usually press the issue through evidence, probability, explanatory power, and defeaters rather than by caricaturing believers. Popular critics of religion, including Matt Dillahunty and Peter Boghossian, have given the objection a sharper public form: if "faith" means believing when the evidence is absent or insufficient, then faith is not a reliable way of finding out what is true.
The core argument can be stated without assuming any one theory of knowledge. A way of forming or defending beliefs is intellectually respectable only if it is constrained by truth: it should either be responsive to evidence, reliably formed in the relevant conditions, or open to checks that expose error. Perception, memory, scientific inference, and testimony are not infallible, but they normally operate within networks of correction: they can be checked, compared, repeated where appropriate, challenged by contrary evidence, and revised. Faith, understood as belief beyond or without adequate evidence, appears to lack comparable constraints. One person can appeal to faith to affirm Christianity, another to Islam, another to Hinduism, another to a new private revelation, and another to a plainly absurd claim such as that the moon is made of cheese. If the same appeal can "justify" mutually exclusive conclusions with equal ease, then it does not discriminate between truth and falsehood. It settles nothing.
On this view, saying "I take it on faith" is not an explanation of why a belief is likely to be true. It is an admission that the usual standards have run out. If evidence is sufficient, faith is unnecessary; if evidence is insufficient, faith is epistemically irresponsible. The serious version of the argument is not that all religious people are stupid or dishonest. It is that faith, as a proposed route to knowledge, lacks reliability. A sincere believer may have psychological conviction, moral seriousness, and a coherent life built around that conviction, but sincerity is not truth-tracking. Contradictory beliefs can be held with equal sincerity.
What the argument gets right
This critique has real force against a common and weak use of the word "faith": belief held in defiance of available evidence, or belief adopted precisely because evidence is lacking. If a person says, "There is no good reason to think this is true, but I believe it by faith," then the atheist objection is not merely clever; it is decisive. That kind of faith cannot function as a public reason, and it cannot distinguish truth from fantasy.
It also has force against a common apologetic retreat. Sometimes a person gives arguments for a religious claim, but when those arguments are challenged, retreats to "faith" as though the word protects the belief from further scrutiny. This is epistemically unstable. If the belief was presented as evidentially supported, then its evidential support is open to examination. If, after criticism, the belief is defended only by faith, then the original evidential claim has been weakened or withdrawn. Faith cannot be used both as evidence and as immunity from evidential assessment.
The argument also exposes an important problem of religious diversity. The world's religious traditions do not merely differ in rituals and symbols; many make incompatible claims about God, the soul, salvation, revelation, incarnation, prophethood, karma, rebirth, final judgment, and ultimate reality. If members of each tradition appeal to faith in support of their own view, then faith alone cannot adjudicate the conflict. At most, it explains why people remain committed to their inherited or chosen traditions. It does not show which tradition, if any, is true.
Where the argument can overreach
The argument is strongest when faith is defined as belief without evidence. But not every serious religious philosopher accepts that definition, and careful atheist philosophers need not rely on it. Many religious thinkers use "faith" to mean something closer to trust, commitment, fidelity, or confidence based on a wider set of reasons. On that account, faith is not a substitute for evidence but a posture of trust within what the believer takes to be an evidentially meaningful context: religious experience, moral experience, historical testimony, metaphysical argument, communal practice, and perceived encounter with the divine. The atheist can grant this semantic point and still ask the harder question: are those alleged grounds actually strong enough, and do they survive known defeaters?
This matters because an argument against one definition of faith does not automatically defeat every religious epistemology. Alvin Plantinga, for example, does not defend faith as arbitrary belief. His Reformed epistemology argues that belief in God may be properly basic: not inferred from arguments, but rational or warranted in a way somewhat analogous to belief in other minds, the external world, or the past, provided it is formed by properly functioning cognitive faculties and is not defeated by counterevidence. William Alston defended the idea that religious experience might function as a socially established doxastic practice, somewhat like sense perception, while openly acknowledging that it is more controversial because its checks are less public and less uniform. William James, in "The Will to Believe," argued more modestly that in certain forced, living, and momentous options, where evidence cannot settle the matter in advance and refusing to choose is itself practically consequential, commitment may be permissible. None of these arguments says that merely attaching the word "faith" to a belief makes it rational.
One need not accept these views to see that they are not the same as "believe anything you like." They attempt to place religious belief within a broader theory of rationality. They may fail, but they do not fail simply because someone online can say "I have faith that the moon is cheese." Serious religious epistemologists are not usually claiming that the word "faith" magically justifies any proposition whatever.
The strongest reply to the atheist critique
The strongest reply is to challenge the assumption that every rational belief must be produced by explicit evidence or argument. Many of our ordinary beliefs are not held because we have inferential proofs for them. We trust memory before proving memory reliable. We trust perception before constructing a theory of perception. We believe other minds exist, that the world did not come into existence five minutes ago with false memories, and that the future will resemble the past in enough ways to make action possible. These beliefs are not irrational merely because they are not conclusions of arguments. They are part of the background from which arguments begin.
From here, the religious reply says: perhaps faith is not an alternative to reason, but a kind of basic trust. Interpersonal trust can be rational without being reducible to laboratory proof: a child may trust a parent, a friend may trust a friend, and a scientist may trust testimony from competent colleagues. But the analogy must be handled carefully. Trust normally presupposes some reason to think the trusted person exists and is reliable, so the analogy by itself cannot establish that God exists or that a particular revelation is true. Its more modest point is that trust is not automatically gullibility. It becomes gullibility when it ignores defeaters, refuses correction, or extends itself beyond its proper domain.
This reply succeeds in one important respect: it shows that the slogan "faith can justify anything" is too quick if aimed at all forms of religious faith. A defender of faith can say that genuine faith is not an all-purpose permission slip. It must be embedded in a larger framework of experience, character, testimony, practice, argument, and responsiveness to counterevidence. On this view, the absurd "moon is cheese" example is not parallel, because it lacks the kind of background practice, experience, testimony, and philosophical motivation that religious communities claim for their beliefs. That does not show those religious claims are true; it only blocks the cheap parody that every appeal to faith is epistemically identical to believing any random absurdity.
Why the reply is not complete
Still, the reply does not remove the problem. First, analogies with perception, memory, and testimony have limits. Perception and memory are fallible, but they are deeply integrated with shared checks. If I claim to see a tree and you do not, we can walk over, touch it, photograph it, compare angles, and test alternative explanations. Religious experience is often more resistant to such shared testing. That does not prove it is false, but it does make its public epistemic status weaker.
Second, properly basic belief faces what philosophers call the "Great Pumpkin" style objection, named after Plantinga's own discussion. If belief in God can be basic, why not belief in any strange entity or doctrine? Plantinga's answer is that not every belief is properly basic; a belief has warrant only when produced by cognitive faculties functioning properly, in the right environment, according to a truth-aimed design plan, and without undefeated defeaters. That is a substantive philosophical account, not a permission slip for arbitrary belief. Still, it is controversial as a public argument, because the design-plan element is naturally read in a theistic framework and will not be granted by many non-theists. Internally, the theory may be coherent; dialectically, it may not persuade those outside the worldview.
Third, religious disagreement remains a serious undercutting pressure. A Christian, a Muslim, a Hindu, a Buddhist, a Sikh, a Jew, and a philosophical theist may all report deep experiences, transformed lives, trusted authorities, and coherent practices. If these sources lead to incompatible metaphysical conclusions, then a reflective believer must ask whether their own confidence is partly explained by culture, upbringing, temperament, or social location. Disagreement does not automatically defeat a belief; experts disagree in many fields, and not all disagreements involve epistemic peers. But widespread, sincere, intelligent disagreement does weaken any simple claim that faith reliably tracks truth, unless the believer can explain why some faith-forming conditions are epistemically better than others.
What the critique does and does not establish
The atheist argument establishes this much: faith understood as belief without sufficient evidence, or as a retreat when evidence fails, is not a reliable route to truth. It cannot bear the weight often placed on it. It cannot, by itself, distinguish true religion from false religion, revelation from delusion, insight from wishful thinking, or loyalty from stubbornness. Used in that way, faith is epistemically defective.
It also establishes that religious believers who appeal to faith owe some account of what they mean by the term. If faith means trust, they must explain why the trust is warranted rather than merely comforting. If it means commitment, they must explain why commitment is appropriate before the disputed truth is settled. If it means a basic belief, they must explain why this basic belief is not arbitrary and how it responds to defeaters such as religious disagreement, divine hiddenness, the problem of evil, and plausible natural explanations of religious experience. Without such an account, the atheist criticism stands.
But the argument does not establish that theism is false. Nor does it show that all religious belief is irrational. It does not refute cosmological arguments, moral arguments, arguments from consciousness, historical arguments, or appeals to religious experience considered on their own merits. It does not show that naturalism is true. It shows, more modestly and more powerfully, that "faith" cannot be used as an epistemic blank cheque.
The best conclusion is therefore discriminating rather than dismissive. If faith is a name for believing regardless of evidence, it fails. If faith is a form of trust claimed to be grounded in experience, testimony, argument, and practice, then it must be assessed through those grounds. The atheist critique is most successful when it targets faith as arbitrary license. It is less successful when it treats every religious use of the word as though it meant that. A serious discussion should keep that distinction clear.