Abstract
Does statistical regularity affect inattentional blindness for natural scenes? Here we presented participants with statistically regular (i.e. good exemplars of four scene categories) or statistically irregular photos (i.e. bad exemplars or inverted good exemplars). We adapted Cohen et al.’s (2011) inattentional blindness paradigm and asked participants to perform an attentionally demanding RSVP task. Their task was to count the number of times a digit was presented in a stream of letters placed against a colorful checkboard background. Every 150 ms, the background changed, and with each change, a new digit or letter was presented. On the fifth (critical) trial, a scene image unexpectedly appeared behind the letter task, replacing the second-to-last checkerboard. Participants were asked if they noticed anything different on that trial and to describe the scene if they saw it. They were then asked to choose which of four scene photos was presented. Subsequently, participants completed three control trials that were similar to the critical trial except participants were told to ignore the letters and digits and only pay attention to the background. On the critical trial, 50% of the participants who received upright good exemplars experienced inattentional blindness, whereas 67% and 86% of the participants experienced inattentional blindness when the scene was a bad exemplar or inverted, respectively. Participants were also more accurate in choosing which photo appeared when the photo was a good exemplar. On the control trials, participants accurately selected the presented images on 94% of trials when the scenes were upright, whereas 75% and 78% of the bad exemplars and inverted images were accurately selected. Thus, participants were less likely to experience inattentional blindness for statistically regular scenes and were better at recognizing an image as having been presented in control trials when it was statistically regular (e.g. good exemplars) as opposed to irregular.