Cognitive scientists in the 1980s were interested in the semantics of psychological state ascriptions: both realists and eliminativists agreed that the truth of such ascriptions depended on facts about cognitive architecture. But the semantics and the science of mental states have since gone their separate ways. Most semantic approaches to mental states (e.g. Stanley and Williamson 2001 on know-how) are neutral with respect to the science of cognition, while scientifically-inclined theories of the mind (e.g. the predictive processing theories of Clark 2013 and Hohwy 2013) generally have nothing to say about the truthmakers of psychological predicates. This separation may be ending, however: recent work on Literalism (Figdor 2018) and Fictionalism (Demeter 2013, Toon 2016) are attempts to reconcile the semantic and scientific elements of philosophy of mind. In this talk, Drayson explores the previous separation of semantics and science, and the challenges facing these recent attempts to bring them back together.