We document three stylized facts about household inflation expectations that stood out during the 2021-23 inflation surge: i) a temporary dislocation in the term structure of expectations, ii) an earlier peak at longer horizons, and iii) a sharp increase in the share of households expecting deflation at medium and long horizons. We show that these stylized facts are consistent with households’ narratives attributing the inflation surge to temporary labor and supply disruptions. We also argue that they pose an empirical challenge to existing models of expectation formation. Implications for theory and expectations surveys are discussed.