is the transformation over time of a well defined, enduring structure [1
]. Based on this definition, Simons and Rensink defined change blindness
as the striking failure to detect a large change that would typically be detected easily [2
]. Thus an individual experiencing change blindness fails to detect that an item has undergone change, perceiving the item as maintaining a well-defined, enduring structure, that is, perceiving the item to be the same as it had previously appeared. It might be assumed that the phenomenon of change blindness reflects the failure to detect the presence
of change or the absence
of change. In several studies, Rensink found that detecting the absence of change is much more difficult than detecting the presence of change, making the individual prone to perceiving an unchanged item as a changed item [1
]. This finding implies that in the detection of the presence of change, individuals easily make the mistake of perceiving a changed item to be the same as it had previously appeared, but also easily make the mistake of perceiving an unchanged item as a changed item in the detection of the absence of change. Such competing implications suggest that different forms of perceptual processing underlie these two types of change detection. Despite such findings, as well as the knowledge that systematic comparison of the two types of detection is necessary to identify the differential perceptual processing employed in detection of the presence and the absence of change, little research has focused on the perceptual processing underlying detection of the absence of change.
Some works on the perceptual processing of subjects experiencing change blindness reported that even when subjects do not perceive a change, they retain representations of the scenes before and after the change [6
]. This finding indicates that detection failure or change blindness may be attributable to a failure to retrieve or compare scenes before or after a change, suggesting that differences in detection of the presence and the absence of change may be due to differences in the retrieval/comparison process employed in these two types of detection [9
]. In the natural environment, rapid detection of change is required for the identification of both dangerous entities (e.g., predators) and beneficial objects (e.g., food and water). Thus, it is plausible to assume that the change detection mechanism would place greater weight on the detection of the presence of change than the absence of change, as a lack of change would not be of interest in this context. Acceptance of such an assumption implies that retrieval/comparison processing for the detection of the presence of a change differs from that for the detection of the absence of a change. This, taken together, implies that the presence of change may be anomalous information for the visual system and that it may be processed preferentially regardless of the circumstance of that change such as what change occurred or how the change was made. If so, global information about the presence of change takes priority over detailed identification of features such as the shape or color of each item. To investigate this possibility, this study examined the nature of the detection of the presence and absence of change by using the same items but interchanging the target and distractors between both detection tasks. Static displays of both detection tasks were identical, but global dynamic information about the presence of change between displays was different. Because some top-down factors, such as a subject’s interest, knowledge or expectations, also affect the detection of change [11
], we used the same display and varying instructions in order to exclude potential effects of top-down strategies by subjects, thus testing the intrinsic effects of the properties of stimuli. It was supposed that as set size increased, retrieval/comparison processing of each item would also increase, resulting in different retrieval/comparison load effects on performance in the detection of change with different set sizes. If the retrieval/comparison processes are different for the detection of the presence and of the absence of change, subject performance should differ with set size. We selected the set sizes based on consideration of the difficulty of both detection tasks. Previous research has shown that detection of the absence of change is more difficult than detection of the presence of change [1
]. When we tested both detection tasks using the same set sizes in the pilot study, we found that the detection of the absence of change was too difficult to perform, and in contrast, the detection of the presence of change was too easy to allow for an investigation of changes in performance. Nevertheless, the use of the same set sizes in both experiments, if it had been possible, would have made the results easier to navigate. For example, presentation time could have been manipulated to vary task difficulty, allowing the use of the same set sizes in both tasks. Doing so, however, would have introduced the additional factor of presentation time, which would have prevented a direct comparison of the results of the two tasks.