Abstract
We attempt in the present paper to investigate analogical reasoning (AR) detached from specific tasks and to formulate its general principles so that it may become a component of problem solving programs as much as the means-ends analysis has been shown to be one in the literature on GPS. After a brief survey of previous related work, we describe the role of AR in intelligent systems, establish a set of plausible working hypotheses, and outline the contributive and hierarchical models of AR. Further advanced processes are then discussed that can extend the scope of the above models. Some problems of implementation are followed by the description of practical results obtained in the explorations of a few simple tasks.