Carnap took the content of a particular sentence or set of sentences to consist in the set of the consequences of the sentence or set. This claim equates meaning with inferential role, but it restricts the inferences to deductive or explicative ones. Here I reject a recent proposal by Robert Brandom, where inductive or ampliative inferences are also meant to confer contents on expressions. I argue that if Brandom's inferentialist picture is upheld, and both explicative and ampliative inferences confer meaning, one consequence of this is that the content of a sentence is to be read off from our ways of rationally altering our beliefs. Meaning and content then are largely concepts of pragmatics, with no clear theoretical interest. My critique affects certain aspects of Dummett's meaning-theoretic picture too, and the discussion also links up with the development of 'dynamic semantics'.