The notion that there are fictitional elements in science is not new, but historically the term has been used dismissively, as part of anti-realist positions about scientific theories. Recently, due to a growing philosophical focus on modeling, as well as the proliferation of computational modeling and simulation in scientific practice, the prospect of a progressivist account of the role of fictions in science (viz. one where fictions make a positive contribution to scientific progress) has been taken more seriously. This dissertation lays out a progressivist theory of fictional modeling that not only accommodates the representational character of scientific models, but their performative character as well. It distinguishes between a wide sense of scientific fiction, which I argue is an exaggeration of abstractions and idealizations that are arguably approximately true, and a narrow sense which is discontinuous with serious, non-fictional hypotheses and experimental practices. Even in this narrow sense, the inclusion of fictional elements in scientific models can solve pragmatic problems, facilitate accurate prediction, contribute to genuine discoveries, and perform explanatory functions; and they frequently do so not despite their fictional distortions of target phenomena, but precisely because of those distortions. Key to the satisfaction of these functions is the way that fictional models serve as props in acts of playful make-believe and are creatively re-interpreted in the course of virtual performances that can reveal novel features, structures, and possibilities for action in "serious," non-fictional contexts. This raises questions about how we might distinguish the intentional attitudes involved in make-belief from those involved in genuine belief, as well as normative questions about the extent to which effective scientific practice requires that we become fully absorbed in the make-believe of virtual scientific performances.