ABSTRACT
Quasi-realists face the challenge of providing a plausible analysis of acknowledgments of moral fallibility (for example, ‘I believe that lying is wrong, but I might be mistaken’). This paper develops a new analysis of these acknowledgements, according to which they express moral uncertainty. After advertising the advantages of this analysis, I take up the question of how to understand moral uncertainty in expressivist terms.
Disclosure Statement
No potential conflict of interest was reported by the author.
Notes
1 Beddor and Goldstein [Citation2018] call these conjunctions ‘concessive belief attributions’ based on their resemblance to concessive knowledge attributions (sentences of the form I know ϕ, but I might be mistaken). Interestingly, while such CKAs are typically judged infelicitous, their belief-involving counterparts sound perfectly natural. For discussion of CKAs, see, among others, Lewis [Citation1996], Stanley [Citation2005], and Hetherington [Citation2013].
2 See Blackburn [Citation1971, Citation1998], as well as Horgan and Timmons [Citation2015] and Lam [Citationforthcoming]. See also Ridge [Citation2015] for a variant of this analysis, cashed out in a hybrid expressivist framework.
3 This idea traces back to Ramsey [Citation1926]. For contemporary developments, see Yalcin [Citation2007, Citation2011], Rothschild [Citation2012], and Moss [Citation2013].
4 My statement of this worry departs in a number of respects from Egan’s formulation, in part so as to sidestep features of his formulation that Blackburn [Citation2009] deems objectionable. In particular, Blackburn insists that the notion of ‘idealisation’ is normative, and hence should be understood in expressivist fashion. My formulation of the challenge is consistent with this point.
5 Some might suggest that expressivists shouldn’t try to offer a non-circular analysis of acknowledgements of moral fallibility (cf. Blackburn [Citation2009]). However, as Köhler [Citation2015] and Ridge [Citation2015] emphasise, one goal of the quasi-realist program is to explain the mental states that we express when we make various realist-sounding claims. So, quasi-realists need to give some story about what mental state Ava expresses when she utters (1b). Without some analysis of acknowledgements of moral fallibility, it’s unclear how the quasi-realist can fulfil this commitment.
6 See, among others, Kratzer [Citation1981] and Dowell [Citation2011].
7 See also Rothschild [Citation2012] and Moss [Citation2013].
8 See Stevenson [Citation1937] and Schroeder [Citation2010: 69–70].
9 What is the relation between belief and credence? For our purposes, we need not take a stand on this. One could embrace a ‘Lockean’ view, according to which beliefs are just credences that exceed some threshold. Alternatively, one could allow that belief entails high credence but deny that belief is reducible to credence; perhaps belief is a sui generis state. A further question concerns what sort of mental state is expressed by assertions of atomic sentences (e.g. eating meat is wrong). The simplest answer would be that asserting ϕ expresses a belief in ϕ. However, given that belief does not require certainty, this raises the question of why ‘epistemic contradictions’ of the form ϕ and might not-ϕ sound incoherent, unlike (8). In response to this problem, one option is to hold that assertions express a state stronger than belief, such as certainty. Another option is to retain the idea that assertions express beliefs, but to adopt an expressivist analysis on which the conjunction ϕ and might not-ϕ is semantically inconsistent. (See Beddor and Goldstein [Citation2018] for relevant discussion.)
10 See Bykvist and Olson [Citation2012] and Staffel [Citationforthcoming].
11 While I aim to show that there is a plausible way of developing an expressivist account of moral credence, nothing in the Credal Analysis hinges on the details of this account. Those who prefer some alternative expressivist treatment of moral credences can still help themselves to the Credal Analysis in order to solve the Fallibility Problem. For examples of alternative treatments, see Eriksson and Olinder [Citation2016] and Ridge [Citationforthcoming].
12 For a different approach to the Commonality Challenge, see Beddor [Citation2019].
13 Here’s another normative similarity. Reasons for belief can be weighed against one another: a reason to believe p can be outweighed by a stronger reason to believe p. Maguire [Citation2018] argues that the reasons for adopting certain affective attitudes don’t work like this. If a suffering loved one dies, the reasons for being distressed do not seem to be outweighed by the reasons in favour of feeling relief. (Indeed, Maguire takes this to cast doubt on the very idea that there are reasons for affective attitudes.) Here, too, planning seems to pattern with belief. The reasons in favouring of planning to go librarywards can be outweighed by the reasons in favour of planning to stay at home.
14 See e.g., Velleman [Citation1989] and Marušić and Schwenkler [Citation2018].
15 See Marušić and Schwenkler [Citation2018] for a somewhat different way of developing this idea.
16 Goldstein also provides an independent argument that degrees of planning are subject to probabilistic requirements. This argument involves extending accuracy dominance arguments for probabilism to degrees of intentions. I defer discussion of this argument to another occasion. For related discussion, see Staffel [Citationforthcoming].
17 Thanks to a referee for raising this point.
18 Cf. Woods [Citation2014], who questions whether expressivists can explain the coherence of I believe eating meat is wrong, but I don’t disapprove of it.
19 For helpful comments and discussion, I am grateful to Andy Egan, Simon Goldstein, Stephanie Leary, Daniel Wodak, the NUS Philosophy reading group, an audience at Lingnan University, the editors of AJP, and two anonymous referees.