Abstract
IntroductionInattentional and change blindness are characterized by an inability to detect unattended events. Interestingly, subjects consistently overestimate their ability to detect such events and are surprised when informed that an undetected event has occurred. We hypothesized that this effect is due to the subjective visibility of unattended stimuli being disproportionately higher than what would be warranted by their objective information processing. We tested this hypothesis using the formal tools of Signal Detection Theory.
Methods and ResultsSubjects detected strongly and weakly attended gratings whose contrasts were adjusted online to produce the same discriminability (d'). We found that subjects were conservative in detecting the strongly attended gratings and liberal in detecting the weakly attended gratings. Further, we used a number of fixed contrasts for both the strongly and weakly attended stimuli. We found that in the weakly attended condition, subjects were close to optimal (i.e. unbiased) for all contrasts used, while in the strongly attended condition, subjects became very conservative for lower contrasts. Thus, surprisingly, subjects were less optimal in detecting the more strongly attended stimuli. We confirmed that this was still the case even when subjects were encouraged to be unbiased by explicitly specifying pay-offs, were informed about the prior probabilities of occurrence of the gratings, and were given feedback after each trial. Finally, when d' was matched in a discrimination experiment, participants gave higher subjective ratings of visibility for the weakly attended gratings, confirming that our previous results were due to differences in the subjective visibility of the stimuli rather than simple detection biases.
DiscussionWe demonstrated that the subjective visibility associated with weakly attended signals is much higher than what would be warranted by the quality of these signals. This effect can explain subjects' surprise at their bad performance in inattentional and change blindness experiments.