Israel developed and revised the Scalar Model of Polarity Items by applying some ideas of scalar reasoning to the analyses of polarity items (e.g. Israel 1996, 1998, 2001, 2004). However, he left three essential issues untouched. First, generally, why is the proposition that conforms to the cognitive schema unacceptable whereas the one that deviates from it can trigger the scalar implicature? Second, how do addressees accommodate the proposition to its opposite cognitive schema before they work out the scalar implicature? Third, how do the polarity items trigger and strengthen the particularized conversational implicatures in some contexts? This paper proposes that Relevance Theory (Sperber & Wilson 1986, 1995) can shed light on these issues, and concludes that contextual effects govern the felicitous uses of polarity items and that the complex interaction between propositions and contextual assumptions underlies the inference of both scalar implicature and particularized implicature in contexts.