The Review of Metaphysics

I

Consider the following case:

Perhaps the best interpretation of this dialogue is that Sinan has attempted either to issue an injunction to Ray—"Call the police!"—or to remind him of some obligation that he already possesses. And in response, it appears that Ray has either resisted that injunction or in fact denied that he has the obligation. In either case, he has apparently done so on the grounds that it is not possible for him to do it. That is, it would seem that Ray has reasoned:

1. (1). An agent ought to φ only if she can φ. [Premise]

2. (2). I cannot φ. [Premise]

3. (3). And so, it is not the case that I ought to φ. [1,2 AEE-2]

Ray's reasoning here would seem to be sound.

Premise (1) is an instance of the moral principle that ought implies can (OIC). OIC (or something like it) is often assumed without argument in normative discourse. Cases such as Case 1 lend support to its intuitive plausibility. Moreover, OIC has a prestigious pedigree, stretching back at least to Kant (and sometimes referred to as "Kant's dictum"),² [End Page 779] possibly Aquinas,³ but perhaps even as far back as Roman law (ad impossibilia nemo tenetur; impossibilium nulla obligatio est).⁴ Nonetheless, OIC has recently been met with criticism. Some critics have mounted indirect attacks on it, either by attempting to undermine any reasoning in its favor⁵ or by attempting to show that ordinary intuition does not support it as much as might be supposed.⁶ Others have mounted direct attacks by attempting to provide genuine counterexamples of impossible obligations, cases in which some agent ought to do what cannot possibly be done in his circumstances.⁷ In response, friends of OIC have constructed defenses on both fronts.

In this article I am concerned to adjudicate the latter dispute. The question I ask is: Can OIC be defended from the purported [End Page 780] counterexamples offered against it? I argue that it can. However, I will say straightaway that the question posed is somewhat misleading. As King has recently explicated, OIC is not so much a single principle as it is a family of related principles.⁸ Different varieties of the principle are determined by specifying (at least) the following six parameters: (1) what "ought" means; (2) what "implies" means; (3) what "can" means; (4) what the subjects of the "ought" and "can" are; (5) what the objects of the "ought" and "can" are; and, very crucially, (6) how to temporarily index these different elements. Once fully determined, some OIC-type principles are defensible; some, on the other hand, are not.

My preferred cognate OIC principle is that ought implies can or could have (OICCH). In section 2 I attempt to state exactly how I understand OICCH by specifying how it addresses King's six decision-points, as well as a couple more. In section 3 I show how OICCH is able to escape all of the best counterexamples that have been posed to OIC. And in section 4 I compare the virtues of OICCH with those of other, similar accounts and argue that it is plausibly one of the best OIC-type principles to date. Since I will be considering not absolute reasons in its support but only relative ones, the upshot of this article is that those who already find OIC prima facie plausible should considering adopting OICCH instead.

II