![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
ABSTRACT
Atheists and agnostics have a vexed relationship. Atheists often regard agnostics as timid, or perhaps as disguised apologists. Agnostics often regard atheists as dogmatic hypocrites: they proclaim something on insufficient evidence, while accusing theists of this. This dynamic is familiar from the academic and popular literature. Here, I consider a more radical conflict between the two, based on Kripkean semantics for empty terms applied to atheism. Sorensen (2018. “Unicorn Atheism.” Noûs 52 (2): 373–388) christened the Kripke-inspired formulation of atheism ‘Unicorn Atheism’ and argued from there to the incoherence of agnosticism. But, I argue, the objection fails and instead presents an opportunity to reformulate agnosticism. By appreciating the relevance of Kripkean semantics to the issue, a better understanding of the two positions, and their conflict, emerges.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1 Bertrand Russell (Citation1969). See Sorensen (Citation2018, 378–790) for discussion.
3 See Dodd (Citation2017) for objections to the view that this combination of attitudes is coherent. See Hawthorne, Rothschild, and Spectre (Citation2016) for a more sympathetic discussion of such an attitude. Intuitively, it is sometimes reasonable to think (or say): ‘I believe p, but I might be wrong about that’. More on the importance of categorical belief states below.
4 See Friedman (Citation2013, section 3.1).
5 Likewise, a meta-linguistic theist about the name ‘God’ might believe that ‘God’ is non-empty. Note that one could be a theist without being a meta-linguistic theist, since one might hold that there are no words for gods in one's object language. Thanks to an anonymous referee for pointing this out.
6 Sorensen alludes to this idea on p. 378.
7 See van Fraassen (Citation1998, 216) for critical discussion. Hájek (Citation1998, 204) thinks that assigning probability .5 to a proposition amounts to agnosticism, and, as we will see, in the case of gods Sorensen has an objection to that view. Hajek also suggests that assigning a proposition a low probability can (at least in some contexts) be understood as believing that it is false. But a unicorn atheist rejects any such belief as well.
8 Sorensen intends his view to be about ‘god’ or ‘gods’ as a natural kind term, rather than ‘God’ as a proper name for some being, though he thinks the same argument applies to ‘God’ as a name (personal correspondence). See Johnston’s (Citation2009, 6–7) on whether the proper name ‘God’ is a description or honorific (such as ‘the highest one’) rather than a proper name. The argument I consider here, though, is about the predicate ‘god’, rather than the proper name ‘God’.
9 See Johnston (Citation2009).
10 None of what follows depends on the implausible claim that the negation of premise 2 is equivalent to, or specifies the meaning of, the sentence ‘God exists’ or ‘there are no gods’. See Kripke (Citation2011, 70). The point here, instead, is that one's attitude towards premise 2 constrains to some degree one's attitude about the purported proposition that God exists.
11 Perhaps Kripshe should say that, while she's an atheist, her atheism might turn out to be false. But that's not the sort of position Sorensen had in mind, since he wants his atheist to be dogmatic about there being no gods. Nevertheless, a less dogmatic position seems to be the best way to address the fictional floodgates from opening: one might worry that my objection to unicorn atheism about gods applies equally well to goblins, for example. One way to avoid accepting goblins as serious possibilities is to hold that, while we are not certain that there are no goblins, our settled view is that there are no goblins. This does not require us to be dogmatic about goblins, as Kripshe is dogmatic about gods. I discuss what having a settled view, or ending inquiry into whether ‘there is a god’, might amount to here in the last section.
12 See Christensen (Citation2010) and Horowitz (Citation2014) for discussion.
13 This is setting aside cases in which one is sure that something exists but unsure about what its name is. For the purposes of Sorensen's argument, we can all grant that if a god exists, we can call it ‘a god’. See Johnston (Citation2009) for some discussion.
14 Anticipating this functionalist point, Sorensen writes ‘Aside from a tiny percentage of action-oriented fanatics (such as fakirs and suicide bombers), the religious behave indistinguishably from atheists’ (377). But that seems false. They pray, are prone to be more hopeful about certain things (e.g. that good ultimately prevails over evil), and their betting behavior is based on their degrees of belief. Atheists, on the other hand, look towards different sources of solace, and different ways of explaining events in their lives, and display different betting behavior on things like premise 2. One response might be to loosen up the requirement that one must always understand the truth-conditions of one's bets, so that a unicorn atheist could still ‘bet’ some money on there being some god without understanding what possible state of affairs could make ‘god exists’ true. Full consideration of such a non-standard position would take us too far afield, and in any case, what if it succeeds? Then the unicorn atheist does think there is some probability that god exists after all (without understanding what that would take to be true), and the question arises what substantive difference there is between her and the agnostic.
15 Since I take it that the functional equivalence between the standard and unicorn atheist is sufficiently fine-grained to commit the unicorn atheist to a probability that there is a god, it might also be noted that, if Kripkean semantics is correct, then a standard atheist is likewise committed to a meta-linguistic position, namely the assignment of some (very low) probability that ‘there is a god’ expresses a (true) proposition.
16 Meta-linguistic agnosticism has the benefit of being clearly distinguishable from non-cognitivism about god, which is discussed by Sorensen on p. 378. The non-cognitivist holds that ‘god exists’ has no truth-value, while this agnostic is open to the possibility that it expresses a true proposition.
17 See Schellenberg (Citation2005, Chapter II.1) on the relation between the two.
18 Though I lack space to show this here, construing unicorn agnostic's categorical ‘suspense’ as the assignment of some probability seems unwise. For one, Friedman's counterexamples to the Russellian view, mentioned in section 1 above, may well apply to any such probabilistic, meta-linguistic construal of agnosticism as well. Second, reducing categorical states to degrees of confidence plausibly requires a contextual element, so that the required degree of confidence required for, say, belief, typically depends on the context. This would make a mess of the atheism-agnosticism debate. Better, then, to consider the agnostic's categorical state on its own terms. An alternative idea is that agnosticism is the state of categorically lacking belief, assent, or acceptance. See Flew (Citation1972), discussed in Draper (Citation2017).
19 Friedman (Citation2013, section 5) criticizes a similar view, which might be found in Crawford (Citation2004). For more resilient variants of the view, See Rosenkranz (Citation2007) and [omitted]. Also see Draper (Citation2017) on ‘epistemological’ agnosticism.