The thesis investigated if the SGOMS architecture is the default structure for processing expert knowledge. We compared an SGOMS model implemented in ACT-R to a model using ACT-R alone. The task had no interruptions but the SGOMS/ACT-R model had processes to deal with interruptions. As result, the SGOMS/ACT-R model predicted slower processing times than the ACT-R alone model, which did not have these extra processes. The task was a well-practiced memory game. The results showed that, although the perceptual motor strategies of the two participants were very different, their cognitive processing for most of the task was virtually identical and in line with the SGOMS predictions. Overall, the results suggest that people use the SGOMS mechanisms by default but can deliberately train themselves to avoid some SGOMS mechanisms for at least some parts of the task.