Abstract

The acquisition of expertise in formal problem solving has been assumed to involve either a shift from backwards to forwards inference, or a shift from unguided to guided forwards inference. In a longitudinal study, the acquisition of formal problem-solving expertise was investigated. Participants were tested as novices before undertaking controlled practice in the problem domain which involved transformation rule problems , and were finally tested as experts. The direction of inference in problem solutions was found to be inadequate to describe the strategic differences between novices and experts. Therefore, a new solution coding system was applied, based on atomic components of problem solution. Analysis of novice and expert solutions revealed no systematic strategy in the novice stage solutions were confused and contained unproductive steps and backtracking. Several strategies were found in the expert solutions, but they did not agree with previously reported results. It was therefore proposed that the acquisition of expertise does not involve a change from one specific solution strategy to another, but rather the development of an efficient strategy, which can differ between participants.