The mentioning of error-possibilities makes us less likely to ascribe knowledge. This paper offers a novel psychological account of this data. The account appeals to “subadditivity,” a well-known psychological tendency to judge possibilities as more likely when they are disjunctively described.