Mental records as " "

Whether the content of a representation is conceived as simply its reference (Russell) or as its reference under a certain mode of presentation (Frege), this content must be distinguished from the representation itself, conceived as that which conveys this content. In language, this distinction corresponds to that between the syntactic and semantic dimensions, between form and meaning; but it applies to thought as much as to language, since both involve representations endowed with content. Mental files are such representations. They are vehicles with content , but not to be confused with the content they convey. Yet this distinction opens up the possibility that it is the vehicles, and not any particular aspect of their content, that play the role of mode of presentation. Indeed, if a subject apprehends the same object through two distinct representations, he may, even if he is rational, fail to realize that both times he is dealing with the same object. Content, from Fodor's perspective, may well conform to the "Fido"-Fido theory, i.e. purely referential, since what plays the role assigned by Frege to the mode of presentation is in fact a "syntactic" entity, a vehicle, rather than a specific level of content.

However, even if Fodor is right and what plays the role of mode of presentation are mental files considered as particular, and therefore vehicles, the fact remains that these vehicles are characterized (typed) by the infogenerative relations they function to exploit. Each type of file conveys a presupposition, namely that the subject is in a certain infogenerative relationship with an entity in the external world (the file reference), the one that characterizes that type of file. This presupposition, which is associated with the file as a type and therefore independent of the environment (unlike the reference), plays a crucial role in explaining behavior, and comes under the heading of what philosophers of mind call the "narrow content" of thoughts, as distinct from the "broad content" that involves aspects of content that depend on the environment. We thus find the need for a two-level semantics inspired by Freghe.