Abstract
Attention is biased from returning to recently-inspected locations, an effect known as Inhibition of Return (IOR). For IOR to facilitate visual search, it should be coded in spatial, not retinal, coordinates. Here we report two studies indicating that the reference frame of IOR is flexible and determined by task demands. When participants performed a manual target detection task with an intervening saccade between cue-offset and target-onset, we found retinotopic IOR. The cue occurred in one of four markers on the screen; this meant the target was more likely to occur in the spatiotopic location of the cue than in its retinotopic location. In a second experiment we inverted this pattern so that the target was less likely to occur in the retinotopic location of the cue. Now target detection was relatively facilitated in the retinotopic location and relatively inhibited in the spatiotopic location of the cue. However, spatiotopic IOR in this experiment was not as robust as retinotopic IOR in the previous experiment. These findings show that decreasing the probability of targets aligning with cues in retinotopic or spatiotopic coordinates increases IOR in that frame of reference, although IOR is more robustly retinotopic than spatiotopic in a traditional cue-target paradigm.