An extremely rough epistemic argument against the belief in any deity.

The following argument just came to me, and I’d like to see what you think about it. It strikes me as a persuasive argument against religion, but I came up with it literally ten minutes ago, have no idea if it already exists in the theological literature (Kenny? Are you reading?), etc.

I’m particularly sketchy on claim 3. I think it’s probably wrong or insufficient as written, but I intuit that something closely resembling it should work.

So this is your opportunity to help put an argument together or take it down. Have at it.


Claim 1: We ought to believe atheism rather than the claims of any existing organized religion, taken as a whole.


Many defenders of religion argue that atheism and religion are both matters of faith, in that there is neither evidence for the existence of a deity nor for the nonexistence of a deity. For the purposes of claim 1, I will assume this arguendo. However, the existing organized religions are all saddled with copious claims for which there is either overwhelming evidence to the contrary (e.g., fundamentalist Christian claims about the age of the earth), or which are wildly at odds with all the evidence we have about the way the world works (the parting of the Red Sea, reincarnation and karma in Hinduism, claims that entail the denial of evolution by natural selection). As atheist belief systems come attached to no such claims, we ought to prefer, in a series of pairwise comparisons, an atheist belief system to each existing organized religion.

Claim 2: We ought not to believe in the god of any particular organized religion, minus the demonstrably false claims of that religion.


Perhaps, then, we could have Allah, or Shiva, or whoever without Islam or Hinduism or whatever? But, in light of the fact that there is no evidence supporting the existence of any god, absent accepting the parcel of claims in an organized religion about what that god did and that god’s relation to oneself, there is no particular reason to accept any particular god over another. Why should one choose to believe in Shiva rather than Allah, other than that one happens to be a Hindu? At most, we could believe that atheism and the existence of each individual no-evidence deity are equally likely to be true, i.e., agnosticism.

Claim 3: We ought not to believe the set or some subset of organized religions or gods of organized religions, rather than atheism.


The next strategy for the believer is to suggest just that: that one can believe that there is some god, or that some organized religion is true, without picking between them. Here’s how such an argument might go:

“But Paul, let’s be Bayesians about this. Your previous two claims have established that we ought to choose to believe atheism over any individual organized religion — which we can translate to the claim that atheism is to be given more subjective probability than any individual organized religion. However, atheism may not be entitled to more subjective probability than the conjunction of each individual organized religion or god from an organized religion. It may be that our sense that the universe is designed, our difficulties with getting a complete cosmological theory, etc. give us good reason to assign atheism a subjective probability less than .5, and assign a higher subjective probability to the proposition ‘one of the organized religions is true’ or ‘one of the gods of one of the organized religions exists.’ ”

(from now on, I think I’ll just call “deities from organized religions” ORdeities)

There are two problems with this argument. First, I think it’s illegitimate under practical circumstances to aggregate beliefs this way. Suppose you assign .49 to the sky’s being blue, and .01 to each of 51 other possible colors for the sky. Strictly speaking, it is correct to suggest that you rationally believe that the sky is not blue, however, if you are asked to make your best guess for the color of the sky, you’d damn well better say “blue.”

Second, to the extent this kind of belief-aggregation is legitimate in a purely theoretical sense, it can at most compel agnosticism, not belief in a deity. For no matter what your subjective probability of “some ORdeity exists” is (the sum of your subjective probability in each ORdeity), your subjective probability, by claims 1 and 2, must be higher in “either some ORdeity exists or no deity exists at all.” (Indeed, if we exclude deities that are not ORdeities — hereafter DOdeities — this probability is 1). If it’s legitimate to aggregate the probability space such that we can make claims about the summed probability of all the ORdeities, why isn’t it legitimate to take the next step and make a claim about the summed probability of that plus atheism?

This summation procedure also yields some very weird claims. For it follows from the at-worst-equal subjective probability of no-deity and any particular ORdeity that if the set of ORdeities is more belief-worthy than no-deity, then the set that includes no-deity plus every ORdeity minus one is more belief-worthy than any individual deity. If we accept the argument that I imagined my interlocutor posing, we must also accept, for example, the claim that

a) “either there is no god, or the gods of Hinduism, Islam, Buddhism, [etc.] are real”

is more likely than

b) “Jesus is real”

thus, if the summation argument gives us reason to utter the phrase “the nonexistence of a deity is not belief-worthy,” it also gives us reason to utter that phrase about the existence of every other deity.

Claim 4: We ought not to hold beliefs in a specific deity that corresponds to no organized religion

Argument: this is basically a repeat of claim 2: sure, we could make up our own deities, but what possible reason would we have to believe them?

Claim 5 and argument: repeat of claim 3, but for DOdeities, for the same reasons.

And these claims seem to sum to the claim that we ought to accept no organized religion nor believe in any deity.


One Response to “An extremely rough epistemic argument against the belief in any deity.”

  1. Kenny Says:

    Yes, I’m still reading :)

    Personally, I think the religious believer makes a mistake in step 1. The definition of ‘faith’ you give is not one that I endorse (see here), though it is a common one among religious believers.

    It seems to me that a religious believer who starts where your imagined opponent starts must end in denying the rationality of religious belief (and so must become either an atheist or a fideist).

    The believer might try to escape by a Plantinga-like move. Suppose she adopts the following principle: when there is no evidence for or against a proposition, I ought not to revise my beliefs with respect to it. (This seems like a bad principle to me, because it seems to ignore Ockham’s Razor-type concerns, but it is, at any rate, not a crazy one.) So the believer goes on believing and the unbeliever goes on not believing.

    This might rescue a minimalistic deism, but surely not any religion which is full-bodied enough to actually effect one’s life. As you point out, any full-bodied religious system will surely have some evidence against it. (The same is, of course, true of scientific theories: surely there will be some observations to make the theory look false, however strong its overall evaluation comes out.) So if the believer starts from the position that there is absolutely no evidence for her belief, she will inevitably fail to defend the rationality of her belief, for the sorts of reasons you list. For the most widely held religious systems, this problem is pretty bad, as there is a lot of apparently contrary evidence.

    I think the moral of the story should be that the religious believer who wants to make room for ‘faith’ should hold that the evidence is inconclusive or equally balanced or something like that, and should then proceed to defend this claim. To admit that there is no evidence is to give up the game, for precisely the reasons you give.

Leave a Comment