I shall argue that there is not in general anything that can be called imperative inference. I do admit that there are certain logical relations between imperatives: these may be summed up in the fact that two imperatives may be said to be inconsistent, if and only if it is logically impossible that they should both be obeyed. What I deny is that this fact enables us in general to apply the notion of inference to imperatives.

By ‘an inference’ I mean a sequence of sentences of the form ‘A, B, …, so C’, such that (a) each of the sentences is actually used for its primary logical purpose – that is to say, where ‘A’ is an indicative sentence, ‘A’ is used to make a statement or assertion, and where ‘A’ is an imperative sentence, it is used to issue a command or order, tell someone what to do, etc.; (b) the final sentence is used to make a statement or issue a command which is arrived at or concluded from the previous statements or commands in virtue of a logical rule. This is not meant to be more than a very vague characterisation of an inference; to carry the characterisation further would involve anticipating some of the discussion that follows. One point, however, is important.