Future directions

  When studying a cognitive behavior as complex as reading, it is unlikely that the research will ever be ``finished.'' At this point in the work, there are several ways that the work could be extended. The most straightforward and least interesting, from a research perspective, is to expand the number of stories which the   ISAAC system is capable of reading. Although the creative understanding abilities of the model alleviates some of the efforts of adding the concepts necessary for a new story, there are still too few stories in the system to allow new stories to be added with no effort. New synopses can be added with little effort, since each takes place in the shared world of Star Trek. But, even then, new concepts do need to be added. Since ISAAC does no novel lexical understanding, each new word in a text will need to be added to the lexicon which ISAAC deals with.

This suggests another direction for future research. It should be possible to extend ISAAC's capabilities by increasing the coverage of the reading behavior afforded by the supertask breakdown. For example, a great deal of low-level reading behavior is ignored by the current theory and model, such as the fact that eye movement is a key component to understanding how humans read texts; yet this aspect of reading is below the level which my work is focusing on. However, it would be possible to extend the control and sentence processing supertasks to take more of this into account.

Similarly, it should be possible to extend the theory's coverage and   ISAAC's abilities to other genres of texts by including different story structure segments. I have hypothesized that the creative reading approach I have been studying is applicable to virtually any genre. For example, consider a newspaper story--it will likely contain concepts which are novel to the reader and must be understood in order to comprehend the article. A different style of story structure comprehension would be needed, but the other supertasks should be applicable.

Another area of future exploration I am interested in focuses on an aspect of the   creative understanding process I alluded to earlier--where should created concepts be placed in the ontological hierarchy? While it is easy in theory to say that a concept may shift across conceptual grid cell boundaries, it is sometimes difficult in practice to determine where its new location should be. Consider the concept of time travel, for a moment. A reasoner with no prior knowledge can utilize information concerning physical transport actions and the temporal column to transform a physical transport concept into a temporal transport actions; i.e., a horizontal shift from the physical action cell to the temporal action cell. Notice that both actions have physical agents for their initiating entities and physical objects for their transported objects. Now, consider a device capable of performing time travel, namely a time machine. Again, a reasoner can start with knowledge of physical transport machines and the temporal domain and develop the concept of a time machine. But, in this case, a physical object has remained a physical object. Both manipulations used similar knowledge; one resulted in a horizontal shift while the other resulted in an intracellular one. For the moment, the problem is circumvented by appealing to the minimum change heuristic, but it is certainly an area of research to be explored.

Finally, I am interested in exploring the application of the theory of creative reading to other domains. Specifically, many of the tasks involved should be highly applicable to creative invention. One aspect of creative invention is the ability to see new uses for existing artifacts; this is one specific area that a mechanism like   FMS could be directly applied.