Learning and Misperception: Implications for Price-Level Targeting