Working Memory in the Acquisition of Complex Cognitive Skills

This paper reports three experiments using the secondary task methodology of working memory, in the task analysis of a complex computer game, ‘SPACE FORTRESS’. Unlike traditional studies of working memory, the primary task relies on perceptual-motor skills and accurate timing of responses as well as short- and long-term strategic decisions. In experiment 1, highly trained game performance was affected by the requirement to generate concurrent, paced responses and by concurrent loads on working memory, but not by the requirement to produce a vocal or a tapping response to a secondary stimulus. In experiment 2, expert performance was substantially affected by secondary tasks which had high visuo-spatial or verbal cognitive processing loads, but was not contingent upon the nature (verbal or visuo-spatial) of the processing requirement. In experiment 3, subjects were tested on dual-task performance after only 3 hours practice on Space Fortress, and again after a further five hours practice on the game. Early in training, paced generation of responses had very little effect on game performance. Game performance was affected by general working memory load, but an analysis of component measures showed that a wider range and rather different aspects of performance were disrupted by a visuo-spatial memory load than were affected by a secondary verbal load. With further training this pattern changed such that the differential nature of the disruption by a secondary visuo-spatial task was much reduced. Also, paced generation of responses had a small effect on game performance. However the disruption was not as dramatic as that shown for expert players. Subjective ratings of task difficulty were poor predictors of performance in all of the three experiments. These results suggested that general working memory load was an important aspect of performance at all level