Abstract
Despite the fact that humans can readily recognize visual objects from one moment to the next, recent work has shown that we have detailed information about only a handful of objects at any one time. One interesting phenomena that highlights this limitation is ‘change blindness’. This is the striking phenomenon whereby individuals have difficulty detecting changes to visual stimuli. In a series of experiments, a visual search task was imbedded within a flicker or change detection paradigm. The target was defined by a change across two visual displays separated by a blank temporal gap. Each display contained identical items at each location, except for at the target location, which contained different items. The sequence of visual displays and gaps were cycled until observers detected the changing target item. This paradigm is particularly useful because the type of change, at the target location, can be manipulated. In the current series of experiments the type of change was varied in terms of features (i.e., the number of features changing at the target location) and familiarity of the change (e.g., a changing familiar object vs. unfamiliar object). In addition, we investigated the extent to which change detection performance varied as a function of processing time (i.e., display duration) and practice (i.e., training sessions). The results provide strong support for the idea that visual changes can be detected using both featural-level information (number of features) and object-level information (familiar vs. unfamiliar object). The results are also discussed in terms of the shift from feature-based to object-based processing, and the degree to which change detection performance improved as a function of learning.