Theory of possible worlds

IRIS Wiki - Narrative Theories - Theory of possible worlds

Authors

Thomas Pavel, Marie-Laure Ryan, among others.

Histo-geographical placement

Last quarter of the 20th century.

Type of story

All types, including complex stories.

Parent Theories

Theory of Possible Worlds in Analytic Philosophy.

Brief Description

Possible worlds constitute a set of narratological theories that describe narrative in terms of several worlds. Each possible world (PW) is a set of elements that constitute a system. Worlds are interconnected to each other by various hierarchical relations (for example an element of a PW, a character, creates another PW by telling a story). Possible worlds have been formalized in logico-mathematical terms, with the particular goal to formalize the relations between these PWs. For example, what is the relation between W₀ and W_D? What is the accessibility between one possible world to the other? What is the logical truth of an element of W_D? Etc.

Several kinds of PWs can be found in narrative:

• The actual world (noted W₀), in which the reader reads a book, the movie audience sees a movie and the audience and actors gather inside a theater.
• The fictional or dramatic world (noted W_D), that is described by the narrative, inhabited by characters, places, objects, and where narrative actions take places.
• Spectators' subworlds, which are worlds that the reader constructs mentally as possible/anticipated worlds. They correspond to inferential walks described by U. Eco in his textual cooperation theory.
• Characters' subworlds, generated by “world creating” propositions such as wish, hope, or imagine.
• Alternate fictional worlds, which are parallel to W_D, when the story explicitly tells several variations on a story so that to create several alternate fictional worlds (It's a wonderful Life (F. Capra), Smoking / No Smoking (A. Resnais), etc.).

Relation with Interactive Storytelling

For Interactive Storytelling, formalizing these possible worlds is a natural option. Artificial Intelligence research provides useful tools for that purpose. For example, the Expert System JESS offers the possibility to apply reasoning rules specific to a “world”, called a module. The very principle of planning is an exploration of possible worlds, in order to generate a solution (in W_D) that satisfies some given constraints (such as enabling to reach a certain story goal). This relationship is well established in the book from Marie-Laure Ryan's research (Ryan, 1991). In agent-based planning, the plan explore character's subworlds. However, despite this strong relation between the theory of possible worlds and Artificial Intelligence, current AI algorithms make limited use of the possible worlds they generate. Generated possible worlds reman internal to the algorithms. What is being shown is a single world, W_D, the best world according to the exploration of other possible worlds. For example, a character, based on possible world's exploration, could say: "If we let it in, the ship could be infected." (Alien movie, R. Scott); or "I you tell her the truth, she will be made at you and tell everything to your boss". Beyond planning mechanisms, the very fact that a single story can embed a multitude of possible worlds of different natures, each these subworlds being candidate to some degree of interactivity, has been largely overlooked. Here follows three suggestions for new possibilities of IS, based on the narrative theory of possible worlds. More details can be found in Szilas et al. (2012).

• Embedded story: One of the most visible “world creating” proposition is the fact that a character in a narrative tells a story. Such an embedded story is a fiction in a fiction. Within a first fictional world, the embedding world, a boundary is crossed and the reader/audience is transported into another world, the embedded world (Ryan, 1991). Applied to IS, embedded stories would consist in letting the user act in both the embedding and the embedded worlds. A Non Playing Character would "tell" an embedded story in a participative way, since the user would play one character in the embedded story. Several cases can be distinguished, depending on what crosses the boundary between the two worlds:
  • Worlds are strictly separate;
  • Some knowledge crosses the world;
  • Some material entities (characters, objects) can travel between worlds, creating strange effects such as the disappearance of one character.
• Multiple actualization: With current IS systems, although the user imagines a multitude of possible worlds and is given choices at several points within the story (choice points), only one route is finally explored. If the user replays, he can explore another route, but each reading is distinct, and previous readings are not accessible during the current reading. The theory of possible worlds opens the possibility that when interacting with a story via a choice point, the user can, not only actualize one of his choices, but also explore other alternative routes. Explorations of these several alternative possible worlds occurs within the same reading session. Navigation between alternative possible worlds can be designed in various manners, depending on:
  • Who decide where the bifurcation points are, the user or the computer;
  • When the navigation is enabled;
  • Where, in a possible world, the user arrives (at the end of his previous exploration, wherever he wants, etc.).
• Optative interactive world: This type of world represents what a character desires. If interactive, such possible worlds would be similar to an alternative actualization (see above), except that this possible world would be biased by the desires of the character creating this subworld.

Systems/Tools using this theory

All systems based on planning implicitly use possible worlds: I-Storytelling, Fearnot, Moe, Gadin, etc., but as explained above, explicit interaction with several possible worlds is still lacking in IDS.

Links

• Possible worlds, described by Marie-Laure Ryan in the living handbook of narratology.

References

• Pavel, T. G. (1975). “Possible Worlds in Literary Semantics.” Journal of Aesthetics and Art Criticism 34, 165–76.
• Ryan, M.-L. (1991). Possible Worlds, Artificial Intelligence, and Narrative Theory. Indiana University Press.
• Szilas, N., Axelrad, M., Richle, U. (2012). Propositions for Innovative Forms of Digital Interactive Storytelling Based on Narrative Theories and Practices. In Z. Pan et al. (Eds.): Transactions on Edutainment VII,  LNCS 7145,  pp. 161--179. Springer,  Heidelberg (2012).