Visual Analogy
J. Davies
Jim Davies
Georgia Institute of Technology,
College of Computing, Atlanta, Georgia.
jim@jimdavies.org
Analogical problem solving with visual representations is important in many situations. I propose to develop a computational theory of visual analogy that will 1) support the notion that visual abstractions are useful for analogical transfer in cases where there are symbolic mismatches at the non-visual level, and 2) provide a language of primitives that will prove useful for visually representing problems and problem solving procedures. Support for these endeavors will come from how the theory predicts and explains test data, comparison with published experimental participant data, comparison to non-visual accounts, implementation of the theory in a computer program, and experiments with it.