One important facet of design is the creative exploration involved in solving problems. We view invention and discovery as evolutionary design problems. For example, the invention of a coffee maker involves the incremental evolution of device models in an effort to eliminate the problems associated with past models. Similarly, the discovery of a physics principle is driven by anomalous predictions credited to that principle.
By our account, discovery and invention involve the adaptation of models in the course of problem solving. Adaptations are acheived by retrieving and using generic transformation models. These generic transformations are linked to models which capture patterns learned from past problem solving experiences. Thus far we have identified several types of patterns including: Generic Teleological Mechanisms, Generic Physical Principles, and Generic Topological Transformations. Each of these patterns involves an abstraction of some sort. They capture some abstract part of past problem solving experience. For example, GTMs can capture some abstract design mechansisms such as "feedback" or "cascading", GPPs capture principles such as "heat flow", and GTTs capture topological abstractions such as the "transformation from 3D to 2D" or the "transformation from one closed figure to another".
We contrast our view of invention and discovery with heuristic search based views (e.g. Langley & Simon, Kulkarni & Simon, Lenat, Valdes-Perez) which treat discovery either as an inductive method for formulating initial theories from data (Langley et al.) or as a method for exploring a large search space based on some set of operators (Lenat, 1988).
The Role of Generic Models in Conceptual Change, Todd W. Griffith, Nancy J. Nersessian, and Ashok Goel. In Proc. of the Eighteenth Annual Conference of the Cognitive Science Society. ftp://ftp.cc.gatech.edu/pub/ai/goel/griffith/cogsci96.ps
For links to a complete list of papers and some ftp'able versions of the above papers go to: Complete Listing