This thesis synthezises theories from intermedia studies, semiotics, Gestalt psychology, cognitive linguistics, cognitive psychology, cognitive poetics, reader response criticism, narratology and possible worlds-theories adjusted to literary studies. The aim is to provide a transdisciplinary explanatory model of the transaction between text and reader during the reading process resulting in the reader experiencing a mental, virtual world. Departing from Mitchells statement that all media are mixed media, this thesis points to Peirce’s tricotomies of different types of signs and to the relation between representamen (sign), object and interpretant, which states that the interpretant can be developed into a more complex sign, for example from a symbolic to an iconic sign. This is explained in cognitive science by the fact that our perceptions are multimodal. We can easily connect sounds and symbolic signs to images. Our brain is highly active in finding structures and patterns, matching them with structures already stored in memory. Cognitive semantics holds that such structures and schematic mental images form the basis for our understanding of concepts. In cognitive linguistics Lakoff and Johnsons theories of conceptual metaphors show that our bodily experiences are fundamental in thought and language, and that abstract thought is concretized by a metaphorical system grounded in our bodily, spatial experiences.

Cognitive science has shown that we build situation models based on what the text describes. These mental models are simultaneously influenced by the reader’s personal world knowledge and earlier experiences. Reader response-theorists emphasize the number of gaps that a text leaves to the reader to fill in, using scripts. Eye tracking research reveals that people use mental imaging both when they are re-describing a previously seen picture and when their re-description is based purely on verbal information about a picture.

Mental spaces are small conceptual packets constructed as we think and talk. A story is built up by a large number of such spaces and the viewpoint and focus changes constantly. There are numerous possible combinations and relations of mental spaces. For the reader it is important to separate them as well as to connect them. Mental spaces can also be blended. In their integration network model Fauconnier and Turner describe four types of blending, where the structures of the input spaces are blended in different ways. A similar act of separation and fusion is needed dealing with different diegetic levels and focalizations, the question of who tells and who sees in the text. Ryan uses possible worlds-theories from modal logic to describe fictional worlds as both possible and parallel worlds. While fictional worlds are comparable to possible worlds if seen as mental constructions created within our actual world, they must also be treated as parallel worlds, with their own actual, reference world from which their own logic stems. As readers we must recenter ourselves into this fictional world to be able to deal with states of affairs that are logically impossible in our own actual world. The principle of minimal departure states that during our recentering, we only make the adjustments necessary due to explicit statements in the text.

I en narratologisk studie granskas spatialitet i skönlitterär prosatext. I texten undersöks några spatiala nivåer som kan länkas till extradiegetisk berättarinstans och "normal" perception kontra drömmar och visioner hos karaktärer inom diegesen. Till den narratologiska studien kopplas undersökningsmodeller hämtade från det som med en paraplyterm kallas kognitiv poetik och som behandlar både den litterära texten och läsarens perception och kognition i läsprocessen. I studien berörs också den fiktiva berättelsen betraktad som "värld" ur ett litteraturteoretiskt och filosofiskt resonemang kring begreppen "actual world", "possible worlds" och "fictional worlds".  Det övergripande syftet är att förklara hur text och läsarkognition samspelar för att skapa mentala, rumsliga "bilder" hos läsaren, bilder som kan skifta mellan olika läsare och som resulterar i den berättelsens "värld" som läsaren upplever. Som grundtexter i avhandlingsarbetet används Eyvind Johnsons Krilonsvit.