My specific contributions are seen in the claims which my work made. I have demonstrated that THE INTEGRATION OF MEMORY RETRIEVAL, ANALOGICAL MAPPING, BASE-CONSTRUCTIVE ANALOGY, AND PROBLEM REFORMULATION ARE SUFFICIENT AND NECESSARY FOR THE UNDERSTANDING OF THE NOVEL CONCEPT CONTINUUM. The necessity portion was through an argument which appealed to the processes that would be necessary for each of the novelty types to be understood by a reasoner. The sufficiency claim was supported by the computational model. Thus, I have contributed a specific algorithm which enables the understanding of novel concepts to occur, and I have shown the tasks involved to be both necessary and sufficient.
The process of creative understanding, however, can lead to bizarre understandings resulting. The overall process is cyclic; more novel concepts will require more cycles to be executed. But, the risk of this is that each additional cycle increases the risk that an unusable understanding will result; i.e., a bizarre understanding will be achieved. My next contribution addresses this problem by explaining one important way which the understanding process may be bounded. I did not wish to produce a set of limited and potentially ad-hoc heuristics which would bound the process; instead, I wanted the process to be inherently underconstrained. This means that the necessary bounding must come from another source. My claim was that THE ONTOLOGICAL REQUIREMENTS AND THE READING DOMAIN PROVIDE SUFFICIENT CONSTRAINTS ON THE UNDERSTANDING PROCESS. The ontology provides bounding by allowing specific types of transformations to be cognitively ``easier'' than others. At the same time, my research has supported the ideas that both satisfaction and interest need to play a role in the bounding of the process. Since these conclusions are not original to my work, nor are they the focus of the bounding process, I have not elevated them to the level of claims.
In addition to the above constraints, the reading domain contains a communicative agreement between the author and the reader which allows a reader to be extremely conservative in the understanding process--as soon as a sufficient understanding is achieved, the process is stopped. This does mean that a number of understanding attempts will stop ``too soon.'' However, since the author will eventually reveal information that will allow the reader to realize this, the conservative approach prohibits bizarre understandings from occurring.
Similarly, the reading process allows implementation of the most difficult of the four tasks which make up the understanding process--problem reformulation. It is difficult in the ``real'' world to decide when to abandon a current understanding for a new one. However, I claimed that THE READING PROCESS CONSTRAINS THE PROBLEM OF IDENTIFYING UNDERSTANDING FAILURES, THEREBY ALLOWING PROBLEM REFORMULATION TO WORK. This was supported by my research which showed that problem reformulation could occur in a principled fashion. Since the communicative agreement is known to exist between author and reader, it is possible that a misunderstanding will be expressly seen by the reader as the text is read.
Taken together, these contributions show that A SET OF TASKS (MADE UP OF MEMORY RETRIEVAL, ANALOGICAL MAPPING, BASE-CONSTRUCTIVE ANALOGY, AND PROBLEM REFORMULATION) ACT TOGETHER TO PRODUCE UNDERSTANDING. This conclusion, along with the contributions concerning knowledge representation, support the higher-level claim that THE ACT OF UNDERSTANDING INVOLVES CREATIVE BEHAVIOR WHICH REQUIRES A SET OF TASKS AND A KNOWLEDGE THEORY TO EXPLAIN. It is impossible to separate these two aspects--the tasks actually ``produce'' the understanding which is achieved; however, the workings of the tasks are dependent on the form and organization of the knowledge. Another organization scheme or another method of representation would probably not allow these specific tasks to function.