Moral realism and program explanation

Abstract

Alexander Miller has recently considered an ingenious extension of Frank Jackson and Philip Pettit's account of ‘program explanation’ as a way of defending non-reductive naturalist versions of moral realism against Harman's explanatory criticism. Despite the ingenuity of this extension, Miller concludes that program explanation cannot help such moral realists in their attempt to defend moral properties. Specifically, he argues that such moral program explanations are dispensable from an epistemically unlimited point of view. I show that Miller's argument for this negative claim is inadequate, and that he has, in spite of himself, identified a promising defence of moral realism.

Notes

¹A referee for this journal has pointed out that this characterization of Harman's conclusion could be misleading. Arguably, Harman thinks that moral properties do exist, but are just relativistic properties. His argument should be taken as a challenge to moral properties (sui generis or otherwise) that are more robust and less relativistic than his.

²Actually, it would be enough to show that they figure in the best explanation of anything that needs explaining. Given the naturalistic, empiricist terms of the debate, however, this is tantamount to showing that they figure in the best explanation of things that show up in experience.

³Jackson and Pettit and Miller use the terms ‘higher-level’ and ‘lower-level’ at some times, and ‘higher-order’ and ‘lower-order’ at others. I shall restrict myself to talk about ‘levels’ so as to avoid confusion with the logical or semantic senses of higher or lower ‘order’.

⁴Miller's argument concerns Cornell Realism in particular, but the issue at stake involves naturalistic non-reductive moral realism generally. I am not especially concerned to defend the whole bundle of doctrines (including causal theories of reference and knowledge) commonly associated with the ‘Cornell’ variant of that view.

⁵Conclusion (M3) is identical to (H5).

⁶For a discussion of the importance of program explanation in contexts of epistemic limitation, see Coppock 1999.

⁷It appears that explanation is not closed under deduction: if p explains q, and q logically entails r, it does not follow that p explains r. This is relevant, because one may be tempted to suppose that, if God knows a lower-level proposition such as (g2), ‘Molecules u, v, w were next in the queue to strike on the molecular bonds at place p_n at time t_n+1, with momentum m_n’, and since (g2) (arguably) explains the counterfactual (e**), ‘If molecules x, y, and z had not struck the molecular bonds at place p_n at time t_n, with momentum m_n, the glass still would have cracked, due to the striking of u, v, and w, place p_n at time t_n+1, with momentum m_n’, and (e**) entails (e*), then God thereby has an explanation of (e*). But this does not follow, any more than the claim that (g2) explains ‘2 + 2 = 4’ follows from the facts that (g2) explains (e**), and (e**) logically entails ‘2 + 2 = 4’. I suspect that an analogue of an old school-yard jibe holds here, ‘It takes one to explain one!’, but I have not tried to argue for that broader point.

⁸A referee for this journal pressed both these points.

⁹Jackson and Pettit do not state this explicitly, but it is suggested by what they say about fragility. Over the course of their paper, Jackson and Pettit test their account against four examples involving: the shattering of a fragile glass, a square peg's not fitting in a round hole, the cracking of our sealed jar containing boiling water, and a piece of uranium's emitting radiation. The fragility case, they argue, ‘is one where the claim about [program explanation] fails: under natural background assumptions, we can say that someone who understands the molecular structure which accounts for the breaking of the glass understands all that is grasped by someone who offers the fragility account. The reason is that all it means to be fragile is to be such—say, to have such a molecular structure—that breaking occurs under the relevant sort of knock. But the fragility case is exceptional' [1990: 116 – 17].

¹⁰Eugenio Lombardo helped me understand this line of argument.

¹¹A referee for this journal has made it clear that much of importance hangs on whether properties such as temperature reduce to lower-level properties such as mean kinetic energy, and that these matters are complex and contentious. Such matters go far beyond the scope of this paper, so I shall try to construct my argument to be as neutral as possible on them. More importantly, in the context of the present dialectic, to assume that moral properties do reduce cleanly to physical properties simply begs the question against our moral realist, who was assumed at the outset to be a non-reductive naturalist.

¹²It guarantees it, of course, only in conjunction with laws relating momentum to pressure, stating the maximum pressure that this jar can withstand, and so on. Also, it may be observed that this sort of guaranteeing is broadly conceptual or a priori, and not causal. This may make it unattractive to those moral realists who are committed to making do with only causal explanations, but it may also show why some moral realists are prepared to abandon that commitment. Alex Miller helped me to see this.

¹³Here the generality of program explanation comes together with its higher-level character as part of the reason why it succeeds where process explanations fail. It is not the generality of universal generalization, however, but the generality involved in existential quantification: ‘If this jar of water is boiling, then there exist some water molecules such that … .’

¹⁴As Todd Buras has pointed out, if Jackson and Pettit's argument works in the case of the cracking glass, a fortiori it should work in the case of morality. If a higher-level proposition contains information not available in the totality of lower-level propositions, where the connection between levels is close and tight (as between heat properties and kinetic properties), they are much more likely to contain information not available in lower-level propositions where the connection is not so close and tight (as between moral properties and subvenient natural properties). If program explanations are genuinely indispensable in the case of the cracking glass, a fortiori they will be indispensable in moral cases.

¹⁵This may be the better way to interpret Harman's challenge and Miller's answer to it. As a referee for this journal has pointed out, Harman also wants the proponent of sui generis, non-relativistic moral properties to give an account of how these properties fit into a naturalistic worldview—which may be construed as going above and beyond merely demonstrating their explanatory potency.

¹⁶The research and writing of this essay were supported by the Institute of Faith & Learning at Baylor University. I thank the Institute and the Baylor Philosophy Department for their support, and David Alexander, Todd Buras, Alex Miller, and Margaret Watkins Tate for helpful discussion and comments. An early draft of the essay was presented at the University of Leeds; I thank Brad Hooker, Matthew Kieran, Andrew McGonigal, Joseph Melia, Roger White, and especially Scott Shalkowski and Eugenio Lombardo for comments on that occasion, and two anonymous referees for this journal for exceptionally detailed and helpful comments.