The claim of cognitive semiotics to offer something new to semiotics rests on the ambition to bring together the research traditions of semiotics and cognitive science. Our focus has been on using the empirical approach of cognitive science in investigating semiotic issues. At the same time, however, phenomenological description plays a major part in preparing the studies and integrating their results, which is what is offered here. Eco has claimed that the mirror is not a sign, but once the notion of sign is specified, the mirror image is seen to be a perfect instance of it. It is no accident that the Gallup test, which is supposed to demonstrate the emergence of the self, starts having a positive result concurrently with the picture understanding. In contrast, mental images are not images and thus not signs. They are presentifications, i.e., a means for making something present, in the sense characterized by Husserl, and by such followers as Marbach and Thompson. We however argue that Husserl’s model of picture consciousness is incomplete, and that Thompson’s study of mental images lacks clarity because of the absence of any real comparison to pictures.