Participants were induced to concentrate preparatory attention at a central location, to identify a letter there, to identify a 2nd letter to the extreme left or right of a central horizontal range of 5 locations, and then to identify a 3rd letter at 1 of the central 5 locations. Analog and discrete versions of the moving-spotlight model predict that response times to the 3rd letter will be most rapid at the location of the 2nd letter, whereas an activity-distribution model predicts that the most rapid responses to the 3rd letter will be at the central location, where preparatory attention is strongest. The data from 3 experiments, taken together, are inconsistent with the moving-spotlight models and are consistent with the activity-distribution model.