In Propositional Attitudes, Mark Richard claims that some natural and formal language sentences of the form(∀ x)(∀ y)(x = y → α → α [y/x])are false. He suggests a substitution for α that is sensitive to certain ancillary features of the variable letter as well as the assignment, and then argues that this substitution generates a false instance of the above-mentioned schema. I reject Richard's argument and argue further that the sentence is not an instance of that schema. I then argue that his putative natural language example fails as well. Finally, I suggest that although Richard's mistake here does not present any technical problem for his semantics for attitude ascriptions, it undermines his claim that his semantic theory is better able to respect the surface form of attitude ascriptions in natural languages than competing theories.