An operational theory is developed for the generalization of relational productions in the presence of background information. This provides a unified theoretical solution for a broad class of intelligence paradigms, including letter sequence completion, visual analogies, and the learning of operators from before-and-after situation pairs or from situation sequences. Relational productions are used as a formal model for operators. The generalization process is given access to a common body of

more »

... ckground information. Background literals which have association chains to foreground literals may be incorporated into the generalization. This theory has been computer implemented in SN0B0L4, and examples illustrate its application in the four paradigms mentioned above.