Knowledge structures

  Supertasks perform processing--taking in input and producing output. The output may be a specific concept or set of concepts which is returned to another supertask. Or, the output may be the altering of a structure in memory and not be an explicit return value. This is the case for the supertasks which produce the tri-representation discussed in Chapter 2; I have denoted these three supertasks as   primary supertasks. To fully represent a story, three large-scale structures are needed:  

• A representation which captures the actions which occur in the story; this representation interprets the story as a ``telling'' of some event or set of events. I call this the scenario representation; as will be described in Section 5.3.3.3, the scenario is my term for the events of a story.
• A representation which captures the fact that a story is a cultural artifact created by one agent (the author) for the purpose of conveying the scenario to another agent (the reader). This is referred to as the story structure representation.
• A representation which captures the internal reasoning activity of the reader. I call this trace of activity   the metareasoning representation.

These representational structures in memory are viewable by all supertasks, via requests given to the   memory supertask. Thus, if a supertask creates and alters a memory structure, the information stored there can be observed by other supertasks. As soon as information is discovered by a supertask, it becomes available to the reasoner as a whole. This form of communication is implicit in nature--any supertask may examine the output knowledge structures being built by the other supertasks.