Abstract
An experiment on transfer of learning using text editors revealed significant differences in performance, based on the learning experience of the subjects. The set of commands of a text editor was divided into four subsets. Different groups of subjects learned these subsets in different orders. Depending on the order of learning, subjects formed different concepts of the editor as manifest by their choice of commands, their errors, and their model of the editor, elicited by a sorting task. Pragmatic production model approaches to transfer would need significant enhancement to accommodate this result.
