Attention enhances visual processing, leading to improved sensitivity (discrimination performance) and faster reaction times (RT). It can be deployed voluntarily (endogenous attention) and sustained over prolonged periods, or captured automatically (exogenous attention), where an early enhancement is followed by a quick decline that may even reverse into a cost. This latter effect –inhibition of return (IOR)– is typically observed with RT, but not with sensitivity measures.

We investigated the temporal dynamics of exogenous attention on visual sensitivity. Participants performed a non-speeded orientation discrimination task. Tilted Gabor patches were presented at either one of four isoeccentric cardinal locations, or at one of two opposing locations along the horizontal or vertical meridian (blocked). A salient, non-predictive cue was randomly presented between 450ms before and 200ms after the test Gabor, either at the same location (valid), or the mirrored location opposite to fixation (invalid).

Results revealed that the effect of exogenous attention (valid vs. invalid discrimination performance) critically depends on the number of monitored locations: With two opposing locations, consistent with previous studies, the benefit peaked shortly after the cue and returned to baseline within 200ms. In contrast, with four possible locations, the benefit persisted beyond 450ms. Crucially, sensitivity at the cued location was never inferior to that at the invalid location, regardless of test location number. These findings suggest that IOR reflects motor inhibition, and is thus typically observed in RT measures. Perception, however, is not inhibited and –depending on the number of monitored locations– can continue to benefit from exogenous attention.