In this paper I will propose to revisit and connect the work I did with Benjamin Spector on the interaction of attitude verbs and their lexical semantics with embedded questions (Spector and Egré 2007, 2014) with some of my earlier work on the question-selection behavior of attitude verbs (Egré 2008). One assumption Spector and I are making is that a question denotes some potential complete answer to a question, though not necessarily the actual complete answer. In my earlier work, I sketched an account of when a verb taking declarative complements might select questions, but based on an essentially incompatible premise, which is that the question should denote its actual complete answer, in order for the embedding verb to guarantee a particular inference from that-clauses to wh-clauses. The paper will give an overview of both problems and of the way they may be related.