Much research supports location-based attentional selection, but J. Duncan (1984) presented data favoring object-based selection in a shape discrimination task. Does attention select objects or locations? We confirmed that Duncan's task elicits selection from spatially invariant object representations rather than from a grouped location-based representation. We next asked whether this finding was due to location-based filtering; the results again supported object-based selection. Finally, we demonstrated that when Duncan's objects were used in a cued detection task the results were consistent with location-based selection. These results suggest that there may not be a single attention mechanism, consistent with Duncan's original claim that object-based and location-based attentional selection are not mutually exclusive. Rather, attentional limitations may depend on the type of stimulus representation used in performing a given task.