Predicting interference in concurrent multitasking

Investigations into multitasking have shown that overlap in cognitive resource use between tasks leads to lower performance. Various computational models have been developed to explain this phenomenon. However, most concurrent multitasking models have been tested in dual-task situations only. In order to see if single-task models could explain dual-task performance, we developed a computational cognitive model that was fit on single-task performance. The model was used to generate a priori predictions of behavior for an experiment in which three tasks with various resource requirements were performed in isolation and in combination. The model predictions closely matched the behavioral data, indicating that models designed for single tasks can indeed account for performance degradation in dual-task situations. To further validate the model, a prediction of neuroimaging data in six regions of interest was generated, which was tested in an fMRI experiment. We achieved a partial fit for the neuroimaging data, indicating that some aspects of the BOLD response require further modeling.