Verbalizable Knowledge of Skilled Task Performance

Abstract

It has recently been shown that variables which affect skilled performance of a task can have no effect on subjects' ability to answer questions about the task and that variables which affect question answering can have no effect on skilled performance. When seen in the same task, this is termed a ‘dissociation’. However, for those concerned with the elicitation of knowledge for engineering purposes, these results are disturbing. The present paper argues that judgments of dissociation may be premature if important parts of an expert's knowledge are not being probed. A series of experiments using a simulated city transport task shows that overall performance on more detailed questionnaires generally does not improve with practice on the transport task. However, there is noticeable improvement on some individual auestions. The reasons for this are discussed.