In Egly, Driver, and Rafal’s (1994) seminal study, an attentional precue appeared either at the target location (valid), a different location within the same object (invalid-same), or on another object (invalid-different). Performance was best in the valid condition, reflecting the advance allocation of spatial-attention. In addition, performance was better in the invalid-same than invalid-different condition, reflecting object-based attention allocation. However, previous studies that used this paradigm did not include a baseline condition in which neither a specific object nor a specific location was indicated. It is, therefore, not clear whether this object-based effect reflects a ‘genuine’ performance benefit over baseline, or a reduction of the cost inflicted by allocating spatial attention to the wrong location. To examine these possibilities, the authors performed 3 experiments in which they added a neutral condition to the classical paradigm. The typical results were replicated, but performance was worse in the invalid-same than neutral condition. Hence, attending an object only reduced the cost of allocating attention to the wrong location. Importantly, because the different theoretical accounts of object-based effects generate different predictions regarding performance in the neutral condition, these findings pose various constraints on the different accounts.