In this contribution, we address the issues concerning the semantic value of Wittgenstein’s subject “I”, as in (i) “I have a toothache”, resulting from the use of predicates that involve first-person knowledge of the mental states to which they refer. As is well-known, these contexts give rise to the phenomenon of ‘immunity to error through misidentification’ (IEM): the utterer of (i) cannot be mistaken as to whether he is the person having a toothache. We provide a series of arguments in favor of a principled distinction between a de facto IEM, grounded in perceptual and proprioceptive judgments, and a de iure IEM, grounded in experience reports whereby the experience wears the experiencer on its sleeve. From this perspective, the no-referent account of subject “I” advocated by Wittgenstein/Anscombe is correct. In fact, we show how this analysis can be made compatible with a Kaplanian account of first-person indexicals, by identifying the speaker in the context of utterance with the person who has access to the reported private experience.