This is one reason it bothers me when empirical philosophers don't treat priori methods *as* methods. E.g., "Swampman is too ridiculous so it doesn't matter whether he is/isn't representing." OK. But is he a reliable source of inferences about representational explanation? Which inferences?
One addresses questions one finds interesting using whatever information and methods of argument one has available and which one can reasonably hope will be informative re those questions. The questions, one's ingenuity, and the resources one can marshal, ultimately decide what methods one adopts.