A previously developed ACT-R/threaded cognition model of dual-task interference (Borst, Taatgen & Van Rijn, 2009) was used to predict neuroimaging data in four brain areas. These predictions were tested in an fMRI experiment, which confirmed the predictions in three of the areas. The fourth area, the intraparietal sulcus, showed a different pattern than predicted. To account for this, a new mapping of an ACT-R module onto a brain area was introduced: It was assumed that activation in the intraparietal sulcus not only depends on the problem state module, as is customary, but also on the visual-location module. The resulting model fit well to the human data, confirming the model’s assumptions of dual-task interference.