The argument stated at its strongest
The argument is not simply that God is hard to imagine. Its serious form is semantic and conceptual. It says that before we ask whether there is evidence for God, we must ask whether the word "God" has been given enough determinate content to make "God exists" a truth-evaluable sentence at all.
Proponents often point to the way classical theists describe God: eternal, immaterial, uncaused, self-sufficient, simple, necessary, infinite, unlike anything in the world, not composed of parts, not located in space, not subject to time, not dependent on anything else. The complaint is that these terms are mainly negative or maximal. They tell us what God is not, or that God has ordinary perfections without limit, but they do not supply a clear positive concept of what God is. If no intelligible content is attached to the term, then the sentence "God exists" may function emotionally, socially, or devotionally, but it does not state a proposition capable of being true.
This family of objections has several sources. A. J. Ayer, working within logical positivism, argued that many metaphysical and theological claims were not cognitively significant if no possible experience could count for or against them. John Wisdom's parable of the invisible gardener, later used by Antony Flew in "Theology and Falsification," pressed the related question of what would count against a theological claim; if every apparent counterexample is explained away, the claim seems to die "the death of a thousand qualifications." Kai Nielsen argued that a radically transcendent personal being is not a coherent concept. Michael Martin developed detailed arguments that some sets of divine attributes, when made precise, generate contradictions. J. L. Mackie argued that central theistic theses face deep tensions, especially around omnipotence and evil, and that the problem of evil is not a superficial objection but a major challenge to traditional theism.
The strongest version of the argument should be separated from weaker online versions. It is not enough to say, "I cannot picture an immaterial mind, therefore the concept is meaningless." Many real things are not picturable. Nor is it enough to mock technical theological vocabulary. Serious philosophers are entitled to use technical terms if those terms do explanatory work. The stronger charge is this: the theist has not supplied stable criteria of application for the term "God," or enough inferential role to distinguish assertion from mere sound. The critic need not demand a laboratory test or a mental picture. The demand is that we should have some grasp of what would make the term apply, what would make it fail to apply, and what follows from applying it. Without that, "God exists" is not like a disputed scientific hypothesis; it begins to resemble "a square circle exists," where the words do not successfully pick out a possible object.
Where the objection has real force
The objection has force against vague, shifting, or purely evasive God-talk. If "God" means whatever preserves belief from criticism, then the term has lost content. If God is said to be loving, but no conceivable pattern of suffering could count against that; intelligent, but no relation to intelligible order is required; good, but in a sense wholly detached from anything we mean by good; personal, but with none of the features that make personality intelligible; then the critic is right to ask whether anything has been asserted.
The objection also has force against some combinations of attributes. It is not anti-religious prejudice to ask whether the following can be jointly maintained: timelessness and responsiveness, simplicity and knowledge of many truths, immutability and agency, necessary existence and personality, perfect goodness and permission of apparently gratuitous suffering. These are not trivial puzzles. Professional theistic philosophers have spent centuries refining accounts of divine eternity, simplicity, knowledge, and action precisely because the tensions are real.
There is also a legitimate challenge about analogy. If the theist says that God is good, wise, and knowing, but immediately adds that these terms do not apply to God in the way they apply to creatures, the critic may ask what remains of the claim. If the meaning is wholly different, the words equivocate. If it is exactly the same, theism risks making God just one more item in the universe, only larger. Classical theology tries to occupy a middle position, but the position needs explanation, not mere assertion.
The strongest reply
The best reply begins by denying that a term must provide a complete positive essence, or direct mental image, in order to be meaningful. Many terms do useful work by description, causal role, theoretical role, or contrast before we possess a full grasp of the thing's intrinsic nature. Biologists spoke of genes as hereditary units before knowing DNA's structure. Cosmologists can meaningfully use "dark matter" as a theoretical term for whatever, if anything, explains certain gravitational phenomena in the relevant models, even while its underlying nature remains disputed. Ordinary speakers referred to water before molecular chemistry. These examples do not prove that God exists; they only show that incomplete understanding does not by itself make reference or truth-evaluable discourse impossible.
Of course, God is not a scientific posit in the same way dark matter is. The comparison only establishes a modest semantic point: incomplete or indirect description is not the same as meaninglessness. A concept may be thin, theoretical, or analogical and still be intelligible enough to be debated.
A theist can offer a determinate candidate description, even if not an exhaustive essence: God is the necessarily existent, non-dependent source of all contingent reality, possessing intellect and will in a non-bodily manner, and not composed of parts. "Intellect" and "will" here should not mean a human brain, emotions, or bodily decision-making; they mean, at minimum, knowledge-like and purposive features sufficient to distinguish God from an impersonal principle. That description may be false, and each element needs defense. But it is not obviously empty. It tells us what would not count as God: a finite spirit, a powerful alien, a demiurge inside the universe, a being dependent on matter, a composite object, or a contingent cause among other causes.
Nor are negative predicates automatically vacuous. We often identify things by negation: non-Euclidean geometry, inertial motion, indivisible particles in early atomic theory, non-human intelligence, non-local correlations. Negative language can narrow a concept. "Uncaused" excludes dependent beings. "Immaterial" excludes spatially extended physical objects. "Necessary" excludes beings that might not have existed. "Simple" excludes entities built out of metaphysical parts. Whether these predicates can all be coherently combined is a further question, but each is not meaningless merely because it is negative.
The demand for "testability" also needs care. If testability means direct empirical falsifiability, then many meaningful claims in philosophy, modal reasoning, ethics, and parts of theoretical science would be excluded; mathematics and logic would have to be treated separately as formal or analytic rather than empirical. This is why the strict verificationism associated with early logical positivism is now widely regarded, including by many atheist philosophers, as too strong. It appears difficult to verify by its own standard, and it threatens to exclude too much meaningful discourse. Ayer himself later acknowledged serious problems with the original formulation.
Flew's falsification challenge is more durable, but it is not the same as theological non-cognitivism. A claim can be meaningful yet so insulated from evidence that it is unreasonable to believe. "There is intelligent life in a galaxy beyond our observable horizon" may be difficult or impossible for us to test, but it is not meaningless. Similarly, if some formulations of theism are unfalsifiable in practice, that may be an epistemic problem, not automatically a semantic one.
Does attribute incoherence prove non-cognitivism?
The critic may shift from "God is meaningless" to "God is incoherent." This is a stronger route. A square circle is not merely unobserved; its defining predicates contradict one another. If the divine attributes do the same, then "God exists" is false in the way "a square circle exists" is false.
Some alleged contradictions are weaker than they first appear. The omnipotence paradox - can God make a stone too heavy for God to lift? - is commonly answered by defining omnipotence as the power to do all logically possible things. Making a stone that an omnipotent being cannot lift is not a possible task but an incoherent description. Many atheist philosophers accept that this simple paradox is not decisive.
Other tensions are more serious. Divine foreknowledge and libertarian freedom remain heavily disputed. Timeless knowledge of temporal events is difficult to model. Divine simplicity raises hard questions about how God can know many distinct truths without internal complexity. Most importantly, perfect goodness and the scale, distribution, and apparent gratuity of suffering raise the problem of evil, one of the strongest arguments against traditional theism. This is not merely a request for clearer definitions; if the argument from evil succeeds, it gives reason to think that an all-powerful, all-knowing, perfectly good creator does not exist. These issues may make certain versions of theism less plausible, or perhaps incoherent if no defensible account can be given.
But this still differs from the global claim that "God" has no meaning. To show that one model of God is incoherent is not to show that every theistic model is meaningless. Classical theists, personalist theists such as Richard Swinburne, perfect-being theologians in the tradition of Anselm, and analytic philosophers of religion such as Alvin Plantinga, William Alston, Eleonore Stump, Brian Leftow, and others do not all define God in exactly the same way. Their disagreements do not prove that any version of theism is true or coherent. They do show, however, that many versions have stable enough content for precise objections, counterexamples, and proposed repairs. That is enough to resist the blanket non-cognitivist claim, while leaving the substantive arguments to be assessed on their merits.
What the argument does and does not establish
The argument succeeds as a warning against empty theological language. A theist should not be allowed to move between incompatible meanings, retreat into mystery whenever challenged, or use superlatives as substitutes for concepts. If "God" is never defined beyond "something beyond everything," then the atheist is right to withhold assent and to ask for a clearer claim.
The argument may also succeed against particular doctrines if their defenders cannot explain how their attributes fit together. If timelessness, personhood, simplicity, knowledge, will, and moral perfection are each affirmed but never reconciled, then the charge of incoherence remains live. The burden is on the theist to show that the proposed concept is possible.
But the argument usually does not establish theological non-cognitivism in its strongest form. The sentence "God exists" is not automatically meaningless because God is described partly by negation, analogy, or metaphysical role. Nor does lack of empirical testability by itself make a sentence non-truth-evaluable. Modern philosophy has largely moved beyond the strict verificationist standard that once made this conclusion seem easy.
Nor does the argument, by itself, establish atheism in the ordinary strong sense: the proposition that no God exists. If "God exists" were literally meaningless, then "God does not exist" would also be meaningless, unless atheism is being used more broadly to mean withholding belief from insufficiently defined God-talk. The more natural conclusion would be ignosticism: no belief should be formed until a coherent concept is supplied. That is a serious position, and it may be reasonable in response to vague or shifting theology, but it is not the same as demonstrating that no God exists.
The fair conclusion is narrower. Some God-concepts are too vague to evaluate. Some may be internally incoherent. Classical theism faces genuine conceptual burdens, especially concerning divine simplicity, eternity, agency, and evil. Yet the general claim that the concept of God is too ill-defined to name any possible object overreaches. A theist can state a sufficiently determinate hypothesis for philosophical debate: a necessary, non-dependent source of contingent reality with intellect and will. Whether that hypothesis is true, supported by evidence, or ultimately coherent remains a further question. The non-cognitivist objection is valuable when it demands clarity; it is less convincing when it declares, in advance, that no clarity has been or could be achieved.