A passage in Hume's Treatise has, by now, become one of the most cited in moral philosophy: the observation that writers on ethics tend to move from premises stated entirely in the indicative — what is or is not the case — to conclusions stated in the imperative — what ought or ought not to be done — without ever specifying the step by which the transition is made. The observation has been read, by some, as a refutation in itself: morality, the reading runs, cannot be derived from fact. It has been read, by others, as a more modest reminder that the derivation, if one is to be attempted, requires its own warrant.
The stronger reading has the merit of crispness but the cost of cutting too cleanly. Hume's own writing does not, in fact, give up on the project of grounding moral judgement in features of the world. His sentiment-based account locates the warrant of moral approval in the responses of beings constituted to care about certain things — suffering, fidelity, fairness — and not in arbitrary fiat. To make Hume's observation about the indicative-to-imperative move into a proof that no ethics can be naturalistic is to enlist him in a programme he did not, on a fair reading, undertake.
The weaker reading is more sustainable. To notice that one cannot, without further premises, derive an 'ought' from an 'is' is not to settle which further premises are admissible. A great deal of work in metaethics over the last century — naturalist, non-naturalist, expressivist — has consisted precisely in trying to specify what such further premises might be. Hume's observation, on the weaker reading, is a call for that specification rather than a barrier to it. The barrier interpretation, by contrast, treats the gap as unbridgeable in principle and so closes off the very inquiry the observation should have opened.
This is a case in which a much-used passage has acquired a load greater than the original lines bear. The passage is real, the observation is real, the gap it points to is real; what is also real is the long subsequent inquiry into how the gap might be bridged. To treat the passage as having settled that inquiry is to misuse it as an instrument of polemic rather than a prompt to argument. The careful reader hears in Hume what was actually said: that the move requires an account, not that no account can be given.
It can be inferred from the passage that the author would most likely regard a philosopher who cites Hume to argue that no naturalistic ethics is possible as: