¹ Advocates of EMVT include Ross, Jacob, ‘Rejecting Ethical Deflationism’, Ethics 116 (2006), pp. 742–68CrossRefGoogle Scholar, and Sepielli, Andrew, ‘What to Do When You Don't Know What to Do’. Oxford Studies in Metaethics 4 (2009), pp. 5–28Google Scholar. Advocates of hedging more broadly construed include Oddie, Graham, ‘Moral Uncertainty and Human Embryo Experimentation’, Medicine and Moral Reasoning, ed. Fulford, K. (Cambridge, 1994), pp. 144–61CrossRefGoogle Scholar and Guerrero, Alexander, ‘Don't Know, Don't Kill: Moral Ignorance, Culpability, and Caution’, Philosophical Studies 136 (2007), pp. 59–97CrossRefGoogle Scholar.

I use the expression ‘moral value’ in a way that is meant to be neutral with respect to all first-order moral theories and their axiological commitments or lack thereof. To the extent that a moral theory makes value comparisons, such that one action is better or worse or equal to another, that theory assigns moral values, on my use of the term. Some authors prefer the expression ‘moral choice-worthiness’ instead of ‘moral value’ in an attempt to avoid any confusion on these grounds. See for example MacAskill, W., ‘Moral Uncertainty as a Voting Problem’, Mind 125 (2016), pp. 967–1004CrossRefGoogle Scholar. For purely aesthetic reasons, I opt for ‘moral value’ instead of the more cumbersome ‘moral choice-worthiness’.

² To hedge one must consider the potential disvalue of wrongly choosing the action you believe is probably right. But this consideration requires making inter-theoretic value comparisons – you must have a sense that it would be much worse to φ if one theory (the one you think less likely) is true than to not φ if another theory (the one you think more likely) is true.

³ MFT is advocated by Gracely, Edward, ‘On the Non-Comparability of Judgments Made by Different Ethical Theories’, Metaphilosophy 27 (1996), pp. 327–32CrossRefGoogle Scholar, and Gustafsson, Johan and Torpeman, Olle, ‘In Defence of My Favourite Theory’, Pacific Philosophy Quarterly (2014), pp. 159–74CrossRefGoogle Scholar. Other non-hedgers include MacAskill, ‘Voting Problem’.

⁴ Weatherson, Brian, ‘Running Risks Morally’, Philosophical Studies 176 (2014), pp. 141–63CrossRefGoogle Scholar; Harman, Elizabeth, ‘The Irrelevance of Moral Uncertainty’, Oxford Studies in Metaethics 10 (2015), pp. 53–79CrossRefGoogle Scholar.

⁵ Weatherson, ‘Running Risks’.

⁶ Harman, ‘Irrelevance’.

⁷ Authors have in fact independently advanced each of these responses. Andrew Sepielli, ‘How Moral Uncertaintism can be both True and Interesting’, Oxford Studies in Normative Ethics 7 (2017), pp. 98–116, responds to the exculpation problem by appealing to epistemic probabilities, while Krister Bykvist, ‘Evaluative Uncertainty, Environmental Ethics, and Consequentialism’, Consequentialism and Environmental Ethics, ed. A. Hiller, R. Ilea and L. Kahn (New York, 2014), pp. 122–35, appeals to weakening the normative domain of Internalism's prescriptions from the moral domain to the instrumentally rational domain (Bykvist's paper was published a year before Harman's and thus does not address her argument, but one similar enough for all intents and purposes). Though these authors’ arguments share certain core similarities to my response to Harman, I offer some important supplements and criticisms of their work. My article also takes a wider view of the problem, encompassing both of these responses, and ultimately offering novel substantive arguments for the rational reading of Internalism.

⁸ It may be objected that Harman is using a case of moral certainty to impugn theories for moral uncertainty – perhaps Internalist theories are meant to be silent on such cases. We can easily modify the case to navigate this objection. Suppose instead that the agent is morally uncertain, but her uncertainty is structured such that φ-ing strongly dominates all other actions. That is, according to every theory she is considering, φ-ing is better than all other actions. Every Internalist theory will prescribe φ-ing under these conditions. Now suppose that φ-ing is objectively wrong. In this case, all Internalist theories will direct a morally uncertain agent to act on her false moral beliefs.

⁹ Harman argues for this claim first in ‘Does Moral Ignorance Exculpate?’, Ratio 24 (2011), pp. 443–68, later applying it against Internalism in ‘Irrelevance’.

¹⁰ Harman, ‘Irrelevance’, p. 65.

¹¹ Harman, ‘Irrelevance’, p. 65.

¹² Harman asserts roughly the contrapositive of 1: ‘An agent is blameworthy for her behavior only if she acted as she subjectively should not have acted’ (‘Irrelevance’, p. 56). Harman uses ‘blameworthy’ and ‘culpable’ interchangeably, and her wording makes explicit what I leave implicit – that the ‘ought’ is subjective. I take 1 to be equivalent to Harman's formulation.

¹³ I do not mean to suggest that Harman asserts a general principle like 1, while failing to appreciate the danger of making it precise in the manner of 2. In fact, Harman specifically addresses the worry that Internalism's prescriptions might not be moral, which I will address shortly. I only introduce 2 to underscore the fact that ‘oughts’ and culpability are related in such a way that for an entailment like 1 to go through, they must agree with respect to their normative domain.

¹⁴ Oddie, ‘Embryo’.

¹⁵ Gustafsson and Torpman, ‘Defence’.

¹⁶ Ross, ‘Deflationism’, p. 755 (my italics).

¹⁷ Lockhart, Tedd, Moral Uncertainty and its Consequences (Oxford, 2000)Google Scholar.

¹⁸ Wedgwood, Ralph, ‘Akrasia and Uncertainty’, Organon 20 (2013), pp. 484–506Google Scholar, at 494–5 (my italics).

¹⁹ Bykvist, ‘Evaluative Uncertainty’.

²⁰ Harman, ‘Irrelevance’ p. 55.

²¹ Sepielli, ‘Uncertaintism’, p. 103.

²² A third strategy might be to construct a plausible and more robust epistemic backstory for both Max and Gail such that their moral ignorance does indeed seem justified. Suppose they were raised in a family and culture in which mob hits and revenge killings were widely believed to be justified, suppose all of the most articulate and authoritative sources of moral knowledge available to them also held these beliefs, and so on. I doubt that we could completely expunge our strong moral aversion to mob hits and revenge killings this way, but for what it's worth, I find my culpability-finding intuitions weakening considerably the more robust this backstory becomes. This, of course, tells in favour of my claim that our intuitions are tracking a lack of epistemic justification. Each of these strategies reinforces the others and underscores the flaws in Harman's original intuition pumps.

²³ I assume that abortion is an issue about which reasonable and well-informed people may disagree about the moral facts. It does not really matter for my purposes what the moral facts are about abortion. We could switch the details and make Glinda confident that abortion is wrong, when in fact it is not (adding the supposition that Glinda has some very strong moral reason to have the abortion – perhaps because attempting delivery would carry with it a high risk of maternal death, leaving Glinda's other children without a mother). I take it we would still find Glinda not culpable for whatever actions she takes so long as we also believe her to be epistemically justified.

²⁴ The failure of these cases extends beyond Harman's 2015 paper against Internalism, to her earlier paper, ‘Does Moral Ignorance Morally Exculpate?’ Ratio 24, (2011), pp. 443–68. In this paper, Harman is responding to sceptical worries raised by Gideon Rosen (‘Skepticism about Moral Responsibility’, Philosophical Perspectives 18 (2004), pp. 295–313) about the extent to which anyone is really morally culpable for anything. Harman argues that even a narrower thesis of Rosen's, roughly that justified moral ignorance exculpates, tells against our intuitions about culpability. Harman cites cases similar to Max and Gail along with several others. The cases of Matt and Glinda should undermine our intuitions in these cases as well.

²⁵ The inspiration for this case is obviously Judith Thomson's famous paper ‘A Defense of Abortion’, Philosophy and Public Affairs 1 (1971), pp. 47–66.

²⁶ Harman, ‘Irrelevance’, pp. 72–3.

²⁷ See e.g. Gustafsson and Torpman, ‘Defence’ and MacAskill, ‘Voting Problem’ on how non-hedging theories have an advantage on this issue.

²⁸ I again want to stress that Harman herself is not objecting to the rational reading of Internalism on these grounds. The target of her objection is closely enough related to the rational reading that it is worth considering, but I do not want to falsely attribute a weak objection here to Harman. Her worry about interestingness does strike me as a sound objection against the target she has in mind.

²⁹ I borrow this case from Weatherson, ‘Running Risks’.

³⁰ See e.g. Ross, ‘Deflationism’, and Brian Hedden, ‘Does MITE Make Right?’, Oxford Studies in Metaethics 11 (2016), pp. 102–28.

³¹ Weatherson, ‘Running Risks’.

³² This article has benefited from the thoughtful comments of many readers and audience members, including those at the University of Colorado Boulder and at the 2017 Eastern APA. I especially want to thank Graham Oddie, Jonathan Spelman, Macy Salzberger, two anonymous reviewers for this journal, and Krister Bykvist for their very helpful written comments.