Abstract
We investigated object- and location-based components of the inhibition of the return (IOR) of attention. In Exp 1, we drew attention first to one of four boxes arranged around a center box and then back to the center. We introduced blank gaps (35, 70, 105 or 140 ms) before presenting a target in either the cued box or an uncued box. (In all experiments, targets appeared equally often in all peripheral boxes.) Following Gibson & Yantis (1994), we expected that the boxes would appear to be new objects after gaps longer than 100 ms, and therefore that the object-based component of IOR would disappear following long gaps. Latencies were longer to detect targets appearing at the cued location than at uncued locations (p< 0.05), but this difference was unaffected by gap length. Thus, as indexed by the gap size manipulation with static displays, IOR was not object-based. In Exp 2, we added color to each of the four peripheral boxes, so they could be distinguished as different objects. We drew attention to one of the peripheral boxes and then back to the center. Next, we rotated the peripheral boxes by 90° and then introduced a gap (0, 35, or 140 ms) before presenting the target. IOR was observed only at the cued location, not at the location to which the cued object moved (p< 0.05). Thus, Exp 2 again revealed location-based, but not object-based IOR. Is attention more likely to track the cued object if it is not drawn away before the cued object moves? In Exp 3 we rotated the peripheral boxes by 90° after cueing a peripheral box, but before drawing attention back to the center. After the motion stopped, we drew attention back to the center before presenting the target. No blank gaps occurred. Again, we observed IOR at the original cued location but not at the location to which the object moved (p<0.01). We consider the possibility that under most circumstances observers allocate attention to locations rather than to objects; hence IOR is predominantly location-based.