Communication between the supertasks

  The supertask organization is only one way in which task activity is coordinated. The other coordination claim of the research is that   COMMUNICATION IS REQUIRED BETWEEN THE SUPERTASKS. This allows the supertasks to support one another in the process of reading. Again, this claim is powerful only when the precise set of communication possibilities are described; this is possible through exploration of the ISAAC model.   One type of communication is simply having the results of one supertask given to another. For example, if the story structure comprehension supertask discovers that Professor Hugo is a character in Zoo, this information can be given to the scenario comprehension supertask in order to aid it in its search for agents. My research has resulted in the claim that INFORMATION DISCOVERED BY ONE SUPERTASK NEEDS TO BE AVAILABLE TO OTHER SUPERTASKS. This information is made available through the construction of three output representations which any of the supertasks may access. So, when a supertask adds information to one of these three representations, the other supertasks can view the addition and take advantage of it for their own processing.

  Another form of communication between supertasks is more explicit. It is possible that a supertask will discover knowledge during the reading process which it believes indicates that certain additional information is forthcoming. For example, a reader processing a text expects to see a title and an author. As a result, they will specifically look for this information. Consider Zoo again. If the reader of the story is familiar with other Hoch tales, they may predict that certain events will occur; perhaps the reader knows that all Hoch short stories contain an ironic twist at the end. Therefore, I claim that   EXPLICIT COMMUNICATION REQUESTS PASS BETWEEN THE SUPERTASKS IN ORDER TO FACILITATE COORDINATION OF PROCESSING. As Chapter 5 will describe,         I make use of four types of explicit messages: explanations, suggestions, anticipations, and requests. They differ in the amount of response a supertask is required to give upon receiving one; in other words, they represent different levels of urgency for the system. Expectations are the most urgent, in that they require the supertask receiving them to alter its processing in order to handle them. Requests are the most benign, with respect to processing changes; these simply request that information be given back whenever it happens to be discovered.