Consider the claim, derived from the ‘ought implies can’ principle, that:
If S ought not to have done A, S could have refrained from doing A.
My sense is that free will skeptics will typically have a hard time denying this claim, even if they are source incompatibilists and accept Frankfurt cases. Many compatibilists, such as Ish Haji (1998) and Dana Nelkin (2011) have a similar sense. But then, on the supposition that determinism rules out the ability to do otherwise, it also threatens 'ought' judgments.
One way of saving ‘ought’ claims from determinism involves dividing senses of ‘ought’ (C. D. Broad 1952). It’s standard to differentiate between ‘ought to do’ and ‘ought to be’ claims (e.g., Lloyd Humberstone 1971). For instance, Mark Schroeder’s (2011) distinguishes the action-related deliberative sense of ‘ought’, and the evaluative ‘ought’, as in ‘Larry ought to win the lottery’ where Larry has been subject to a series of undeserved misfortunes. Kate Manne (dissertation 2011) argues – plausibly to my mind -- that the evaluative ‘ought’ also applies to actions. She proposes that an evaluative ‘ought’ claim does not (at least directly) entail a ‘can’ claim, even when it concerns an action, while an ‘ought to do,’ which expresses a demand of an agent in a particular circumstance, does entail that she can perform the indicated action. We might call this last type an ‘ought’ of specific agent demand.
So here’s the thought. From my free will skeptical perspective, given determinism and that determinism precludes alternatives, when one tells an agent that he ought to refrain from performing some action in the future, the ‘ought’ of specific agent demand isn’t legitimately invoked, but the ‘ought’ of axiological evaluation can be. Such a use of ‘ought’ proposes to an agent as morally valuable a state of affairs in which he refrains from performing the action and recommends that he not perform it. We might call this the ‘ought’ of axiological recommendation. Such a prospective use of ‘ought’ is not at odds with determinism. Suppose it turns out that the agent performs the action anyway. If there was good reason to believe in advance that agent has or could develop the requisite motivation, and especially if there was good reason to think that articulation of the ‘ought’ judgment would contribute to producing it, this use of ‘ought’ would typically be legitimate.
Does this sound right? And how bad would it be if we had to settle for ‘oughts’ of axiological recommendation?