When describing players’ actions in videogame playing, we often use sentences that seem to say that a player performs her action within a fictional world. But it seems to be impossible for a real agent to act in a fictional location. This is puzzling. This paper's goal is to analyze such sentences and solve this ontological puzzle.

There are some possible answers. Immersionism takes the sentences as expressing the players’ experiences of immersion. Virtualism assumes that they refer to players’ actions within not fictional but virtual worlds. Fictionalism claims that they are not true but fictionally true. Realism says that allegedly fictional actions are really real. I show that all of these are far from satisfactory. More sophisticated theories like Grant Tavinor’s interactive fiction theory and John Sageng’s reference shift theory also have some crucial flaws.

My own institutionalism takes a realist position and aims to fill the shortage. My view is this: Actions in videogames are as real as institutional facts. They are specified by graphical symbols, and these symbols are individuated by their fictional contents. Because of this, we use vocabularies originally for fictional contents to describe actions specified by on-screen symbols that represent the fictional contents.