Efficient pointing is crucial to graphical user interfaces, and input techniques that dynamically change their activation area may yield improvements over point cursors by making objects selectable at a distance. Input techniques that dynamically change their activation area include the bubble cursor, whose activation area always contains the closest object, and two variants of cell cursors, whose activation areas contain a set of objects in the vicinity of the cursor. We report two experiments that compare these techniques to a point cursor; in one experiment participants use a touchpad for operating the input techniques, in the other a mouse. In both experiments, the bubble cursor is fastest and participants make fewer errors with it. Participants also unanimously prefer this technique. For small targets, the cell cursors are generally more accurate than the point cursor; in the second experiment the box cursor is also faster. The cell cursors succeed in letting participants select objects while the cursor is far away from the target, but are relatively slow in the final phase of target acquisition. We discuss limitations and possible enhancements of input techniques with activation areas that contain multiple objects.