Theism, Atheism, and Agnosticism

Two people call themselves atheists. The first holds that the claim that God exists is false, and can say why. The second holds nothing at all on the matter and simply declines to believe, on the ground that no adequate case has been made.

When a theist argues with them, only the first is in the argument. The second has taken up a position that cannot be refuted, because it does not assert anything. Whether that is intellectual hygiene or an evasion is one of the live questions in this article, and settling it requires being clear about what these three words pick out.

Positions as claims

Notice what these definitions are definitions of. They pick out propositions, which are the sort of thing that can be true or false, argued for, and used as premises. They do not pick out states of a person.

That choice is deliberate and it is the standard one in academic philosophy of religion, for reasons worth stating rather than assuming.

It keeps the two positions symmetrical. Theism is almost never defined as a psychological condition, and defining its opposite as one produces a pair that cannot be compared.

It matches what the disputes are actually about. When someone argues that suffering counts against God's existence, the conclusion of that argument is a proposition, not a report of the arguer's state of mind.

And it keeps the words usable in arguments. A psychological state cannot be a premise or a conclusion, cannot be true, and cannot be supported by evidence. If atheism is defined as the absence of a belief, then no argument for atheism is possible even in principle, which is a strange result for a position with a large literature arguing for it.

Nothing in this rules out the other usage. It means that when the other usage is intended, this tier will say so, using different words.

Positions as states of a person

The pair comes from Michael Martin, whose Atheism: A Philosophical Justification of 1990 organises its case around the distinction, and the underlying idea is older than the labels. Antony Flew defended defining atheism as the mere absence of theistic belief in the early 1970s, and his reason for preferring that definition was an argument about who is obliged to make a case, which the last article in this tier takes up in full.

The distinction is real and it matters. Not believing that something is so and believing that it is not so are different conditions, and a great deal of ordinary reasoning depends on keeping them apart. A jury that returns a verdict of not guilty has not found the defendant innocent; it has found that the case was not made. Someone who has never heard of a proposition does not believe it, and does not thereby believe its negation.

Suspending judgement, and saying nobody can know

The two readings are packed into one word and separating them prevents most of the confusion around it.

On the first reading, agnosticism is a report. Someone describes themselves as neither believing nor disbelieving. Nothing has been claimed about what anyone else should think, or about the evidence, and there is nothing to argue with.

On the second reading, agnosticism is a substantive epistemological thesis. It says that the evidence does not support either position, and therefore that theists and atheists alike hold their views without adequate warrant. This is a strong claim about the state of the arguments, it is precisely as much in need of defence as theism or atheism, and it can be wrong.

Thomas Henry Huxley coined the word in 1869, at a meeting of the Metaphysical Society in London where every other member had a label to hand and he found he had none. His formation was built against the gnostic of church history, who claimed knowledge of exactly the matters Huxley took himself to be ignorant of. What he intended was closer to the second reading than the first: not a mood, but a rule about not professing certainty beyond the evidence.

The overlap with the previous section is the source of most of the arguing. Someone who suspends judgement is, on the psychological definitions, both an agnostic and a weak atheist. The two labels are not competing descriptions of the same person's beliefs; they are two different partitions of the same territory, and asking which one is correct is asking a question with no answer.

One God or all of them

The definitions above were stated for the God of classical theism, and the restriction is not a technicality.

Denying that one specific being exists is a much smaller undertaking than denying that anything anyone has ever meant by a god exists. The arguments that get made in this subject are almost all directed at the first. An argument from the distribution of suffering, for instance, is aimed at a being who is at once all-powerful and perfectly good, and it has no force at all against a limited deity, an indifferent one, or a non-personal ultimate of the kind found in several major traditions.

This cuts against overreach on both sides. Someone who has answered the arguments for the God of classical theism has not shown that no such being exists. Someone who has refuted classical theism has not shown that every conception of the divine is empty, and would need either to work through the alternatives or to defend a general account of what the world contains, which is a much larger project.

The honest form of most positions in this subject is therefore narrower than the label suggests, and saying which conception is in view is part of stating the position at all.

Who has to argue

This is where the definitional work pays off, because how the words are defined determines who is obliged to do what.

If atheism is the proposition that God does not exist, then it is a claim about how things are, and the person asserting it owes reasons in the way the theist does. Two parties assert incompatible propositions, and neither gets a discount.

If atheism is the absence of belief, the shape changes. The atheist has asserted nothing, so there is nothing for them to defend, and the entire burden falls on the theist. Failure of the theistic arguments is then sufficient for the atheist's position to stand, without any positive case being made at all.

That is why the definitional dispute is not merely verbal. It settles in advance which side loses a stalemate.

Three cautions apply to any use of the notion.

The courtroom analogy is imperfect. A trial has to end in a verdict, so a default is institutionally required, and the presumption of innocence is a rule about what to do when the evidence is inconclusive rather than a claim about what is probable. Inquiry has no such requirement. Nothing forces a conclusion at the end of an inconclusive investigation, and suspending judgement is always available.

Assigning the burden to the party making the positive claim sounds neutral and is not. Both theism and its denial can be stated either positively or negatively depending on how the sentence is arranged, and any rule sensitive to that is a rule about grammar.

And the notion can be used to avoid the work. A party who never states a position, never defends one, and simply requires the other side to satisfy an unspecified standard has not participated in an inquiry. The reasonable version of the appeal is a claim about prior plausibility, namely that one hypothesis starts out likelier than the other and that inconclusive evidence therefore leaves it ahead. That is a real argument with a real premise, and it is where the last article in this tier picks the question up.

Next

The positions are now separated, and the question of who has to argue has been raised without being answered. Before it can be answered, two further things are needed. The next article fills in what theism claims about the relation between God and everything else, since a great deal of the argument turns on whether that relation is a single past event or a continuing one. The two articles after that ask what faith contributes to belief, and what kind of case for a claim like this is even possible.