As Prior was perhaps the first to note in print, Hesperus-Phosphorus cases aren't the only kind of substitution-puzzle that attitude ascriptions give rise to. It's commonly held that "that-p" and the corresponding propositional description "the proposition that p" stand for the same proposition. But "Holmes suspects that Moriarty has returned" and "Holmes suspects the proposition that Moriarty has returned" mean something quite different. (The phenomenon repeats itself with a wide range of attitude verbs that have both clausal and transitive forms.)

In this paper I trace the difference to a difference in the thematic grid of clausal "suspects" and transitive "suspects". In a neo- Davidsonian framework in which action and state verbs introduce quantification over events, the difference emerges starkly in the semantics as an explicit difference in thematic relations. So substitution of "that-p" for the corresponding propositional description "the proposition that p" or vice-versa is ruled out by the "no truth-condition-altering side-effects"requirement on correct applications of Identity Elimination. The point is that substitution effects a change in syntactic category of the attitude verb, which in turn triggers the side-effect of changing thematic relations: when the transitive verb is used, it is the theme of the attitude-state or event that is identified, but when the clausal verb is used, it is the content of the state that is identified.