The training of complex task performance

To establish the rules that governs the effectivenness of part-task training in the learning of complex perceptual-motor tasks such as the 'Space Fortress' game, we compared two part-task training regimes with a control training regime basd on whole-task practice. In previous research, these part-task training regimes were shown to improve subjects' performance. We compared the rate of learning and the final performance of subjects trained with these regimes, as well as the extent to which the acquired skills were susceptible to inference by a battery of concurrently performed tasks. Thirty-three subjects participated in the study. The quality of subject performance varied with the training regime. The best performance was achieve by subjects trained with the hierarchical approach, which devotes part of the training time to practice on a series of sub-tasks presumed to develop the elements of the subjects' optimal strategy. Subjects trained with the integrated approach were continually exposed to the wholegame, while components of the game were emphasized by means of instructions and feedback. These subjects obtained lower scores than did subjects in the hierachical group, although they were superior to control subjects. However, the resistance to the interference produced by the concurrent tasks was most pronounced for subjects in the integrated group. Thus part-task training was superior to whole-task training. However, in the choice of training regimes it is important to consider which aspect of performance is to be optimized because robustness to inference in dual-task conditions is not necessarily correlated to higher levels of performance in single-task practice.