Abstract
Viewers generally feel that they have a richly detailed perception of an entire scene, even though experiments have demonstrated that stimulus-processing capacity in unattended locations is relatively poor. Previous preliminary results from our lab suggests that when stimulus-processing capacity (d′) was matched, subjects were more liberal when detecting uncued than cued stimuli. They also reported higher levels of subjective visibility for uncued stimuli in a discrimination task when d′ was matched. In the current study, we manipulated attention by varying the the number of items presented. Subjects were asked to determine the tilt of a cued target and then rate the visibility of that target as low or high. The number of items on the screen was either two or four, and the subjects were given the cue/probe after the items had disappeared. When attention is divided between more items, performance decreases as expected. However, when we examined the relationship between performance and visibility ratings, we found it differs between the 4-item and 2-item conditions. At low performance levels, subjects rated visibility to be higher in the 4-item than in the 2-item condition, even though subjects were equally good at discriminating the targets. The results suggest that divided attention can lead to inflation of subjective visibility, especially at low contrast/performance levels. Taken together with other empirical results, we argue that perhaps the inflated visibility of weakly attended stimuli contributes to the fact that viewers are overconfident about what they can see in a single glance when looking at a complex scene.