Abstract
Imagine you are looking at a cat and make the following inference: That cat sneezed; That cat is missing an ear; thus There exists a sneezing cat missing an ear. Such an inference is valid only if there is no equivocation on the term “that cat.” If “that cat” in refers to Puss, but in refers to Midnight, then the inference is invalid. This much is elementary. Now imagine that Puss is the cat in front of you when you think, but that a nefarious semanticist quickly substitutes similar-looking Midnight for Puss, so that when you think, the cat in front of you is Midnight, though you believe it is Puss. If it is true that the switch changes the meaning of the term “that cat” from Puss to Midnight, then your inference is invalid although this fact is hidden from you. Two questions are then raised. Do we ever really know that our inferences are valid, given the uncertainty of apparent coreference? And, is it ever reasonable to think that our inferences are valid, given this uncertainty? Lawlor’s answers are “no” and “yes,” respectively.