In this paper we present a solution to Saul Kripke's Puzzle About Belief (in: Margalit (ed) Meaning and use, Dordrecht, 1979) based on Kaplan's metaphysical picture of words (Proc Aristotelian Soc 64:93-119, 1990, J Philos 11:504-529, 2011). Although it is widely accepted that providing such a solution was one of the main incentives for the development of Kaplan's theory, it was never presented by Kaplan in a systematic manner and was regarded by many as unsatisfactory. We agree with these critiques, and develop an extension of Kaplan's theory by introducing the notion of "conservative" word uses, as well as arguing for the restriction of Kripke's Disquotational Principle only to such uses. This restriction allows us to solve Kripke's Puzzle by arguing that the assent of the puzzle's protagonist does not allow disquotation and ascription of corresponding inconsistent belief, as it contains the non-conservative use of a proper name. At the end of the paper, we defend two of Kaplan's theses that we find essential for our argument to work: that individuation of word tokens involves taking into account the intention of the speaker, and that different uniquely referring proper names of the same shape should be regarded as different homonymous words.
