Extending situation theory using narrative structure
Situation Theory was developed by Jon Barwise and John Perry at Stanford to aid contextual reasoning, initially funded to improve computational reasoning systems.
Keith Devlin developed a formalism of situation theory, known as Layered Formalism and Zooming, to represent how the truth of facts alters in different contexts. However, the structure on one side of the equation (see below) was a placeholder for the real-life situation it represented, so was not able to be implemented in a machine. The only formal depiction of instances of change were at the different moments of measurement.
Goranson and I positioned narrative structure in the philosophical slot where Devlin’s ‘real world’ focal situation would be, due to the similarity of it being an interpretive target. We replaced the non-computable placeholder on the left-hand side of the equation, as well as the ‘turnstile’ that joins the two sides. Instead, the structures of narrative transitions would be inserted in this space, so that we could formally capture the transition between fixed, generic knowledge structures. This is where my animated models fit in.
Details of a two-sorted reasoning method
Situation theory’s equational design has some unique features that accommodate this characteristic. Some of these are:
- LFZ’s equations are split down the centre, with the two sides separated by a turnstile (for a layperson, this turnstile is located where an equals sign might usually be). Below is a reproduction of an LFZ equation from Devlin’s Logic and Information:
The above statement says: Jon sees Mary (Devlin, 1995, p. 51).
- Each side of the turnstile represents a different method of reasoning. On the right-hand side are logical rules of association; on the left-hand side sits an S, which is simply a placeholder that represents some elements of the actual situation in which Jon sees Mary.
The result is a system that situates knowable facts (on the right hand side) into real-world situations (on the left hand side), and thereby acknowledges that between one logical statement and its consequence, many unrecorded factors might have altered in the real world. The fact statements within the chevrons are named infons.