Change Blindness

Change blindness (CB) refers to the finding that people, in certain circumstances, are surprisingly poor at detecting large changes in visual scenes. This entry will present the methods used to produce CB and discuss how CB findings have informed thinking about the role of attention in scene viewing and about the complexity of visual representations.

In a classic demonstration of CB, observers attempted to detect a change while a photograph of a jet on a runway alternated with a photograph in which the engines had been removed. When a brief (~80 ms) blank frame was inserted between the two images, the change became very difficult to detect, requiring up to 40 alterations between the pictures for successful detection. Once the large change was detected, most people were amazed that they had failed to notice it sooner. This method of producing CB by inserting a blank frame between the two images is called the flicker technique. CB can also be produced when the two images are separated by an eyeblink, an eye movement (saccade), or a camera cut in a video; when an occluding event momentarily blocks the view of the change; when the change co-occurs with the appearance of a number of “mudsplashes;” or when the original scene slowly morphs into the altered image. These varied methods all have one critical aspect in common: When they are implemented, the change fails to produce a clear and isolated motion transient (i.e., a burst of neural activity in response to a change of retinal stimulation). Changes that produce a single clear transient are readily detected, but when people are unable to rely on a transient and must instead rely on their knowledge of the contents of the scene, CB often results.

In the 1990s, these CB findings garnered a great deal of interest. The use of real-world scenes combined with the surprising results revived interest in scene viewing and led to questions about the role [Page 146]that attention plays in the process. A number of CB findings support the conclusion that change detection in these circumstances requires focal attention. For instance, changes to items that are of central interest in the scene are more easily detected than changes to items of marginal interest. Given that attention should be preferentially allocated to central interest items, this finding is consistent with the view that attention is necessary for change detection. Second, changes become easy to detect if one is cued to attend to the location of the change prior to the change occurring. These findings suggest that attention is necessary for change detection.

Figure 1 Example of a flicker change detection paradigm

Other results, however, suggest that attention to an object may be necessary but not sufficient to detect a change. For instance, research finds that people often fail to notice when the main actor in a video is replaced with a different actor during a camera cut. Even in real-life interactions, people often fail to notice when the person they are giving directions to changes to a new individual, provided the direction giver's view of the swap is blocked by an occluding event. Clearly, in these situations the person who changes is the focus of attention, yet CB results. These observations suggest that one must not only attend to the object in order to detect the change, but must attend to the specific aspects of the object that would allow one to detect the change.

...