Jerry A. Fodor, "On Knowing What We Would Say"

Speaker intuitions in cases where they are asked to imagine cases radically different from what we know to be the case are systematically unreliable, because it isn't possible to predict what other beliefs would change given change in some of our basic beliefs. That means it isn't possible to cite speakers' intuitions to isolate a difference between a mere empirical feature of the meaning of an expression (a "symptom") and a "logical" feature of the meaning of an expression (one the absence of which would entail a change in the meaning of the expression--a "criterion").