Perhaps the best known formalization of the creative process comes from the work of Wallas (creat:wallas). Wallas viewed creativity as being made up of four steps: preparation (the reasoner gains a rich knowledge of the problem), incubation (the reasoner's subconscious is working on the problem), illumination (a potential answer is consciously revealed in a flash of inspiration), and verification (where the reasoner examines, corrects, and refines the solution). This points out that creativity takes a great deal of knowledge and preparation and that the unconscious plays a large role in the process. Beyond that, it does not reveal much regarding how the process would be carried out or exactly what makes up each step. As a result, it provides a good description of the creative act but does not help us to understand it to the point where it may be modeled. Wallas' theory also differs from my creative understanding work in that his relies on the illumination stage--this stage is where much of the ``magic'' of creativity happens. My research, described in detail in Chapter 6, takes the approach that creative understanding depends on the same processes which so-called mundane understanding would. There is still a need for preparation, but it now happens on a continual basis rather than when creativity is triggered. There is also the idea of incubation but, once again, it is an always occurring facet of reasoning. Similarly, verification is required but as a general feature of the understanding process, not as a special aspect of the creativity task.
The most in-depth criticism of Wallas' work on creativity came from Weisberg (creat:weisberg). In his approach, creativity is an extension of normal problem solving:
Other researchers have taken a slightly different approach to the problem of creativity. Rather than seeing creative behavior as the application of creative techniques to a given problem, Getzels and Csikszentmihalyi (creat:getzels1) suggest that creative people are good problem finders, not just problem solvers. This approach allows more exploration of options before the direct tackling of the problem itself. In a similar fashion, Simon (creat:simon1) suggests that creativity is the process of reformulating a given problem so that the proper (and often creative-appearing solution) just ``jumps out'' at the reasoner. My theory of creative understanding also contains a problem reformulation element--if a current ``attack'' on understanding a novel concept is being unsuccessful, it may be advisable to reformulate the problem and try to understand the concept from a new perspective. Unlike the two previous theories, the problem reformulation in my work is not the central mechanism of creativity. The ISAAC model does not explicitly search for the proper interpretation of the problem prior to solving it; since ISAAC is an implementation of a creative understanding system, it is always searching for the ``proper'' interpretation of the concepts which are given to it. However, if ISAAC discovers that it has been utilizing an inaccurate interpretation, it will attempt to find another way to view the concept, a process I consider highly analogous to what Simon and Csikszentmihalyi and Getzels discussed. Also, the methods I use to implement my version are different, relying mainly on the fact that ISAAC is a reading system (see Chapter 6).
Finally, Clement (creat:clement1) and Nersessian (creat:nersessian1) have both proposed (among others) that creative behavior results from the ability to combine theories and concepts from domains which are generally separate. A creative person is someone who can use their knowledge of a certain domain in the solving of a problem in another domain. Thus, creativity is a highly analogical and knowledge-rich process (for further discussion of this topic, see [#!creat:ram2!#]). These ideas have directly led to one specific step of my creative understanding process--the ability to dynamically create a new concept from which to draw an analogy. Some of these ideas of dynamic creation are implemented in the work of Griffith, Goel, and Nersessian (science:griffith-goel-nersessian-1996). The ToRQUE model of scientific discovery manipulates generic models in order to problem solve. As one aspect, it can create new models ``on-the-fly'' when they are necessary by appealing to a set of generic transformations. The specific techniques I use to produce dynamically created concepts are where the differences exist between my work and this prior work.