Change perception using visual transients: object substitution and deletion.

In three experiments we studied change detection and identification when no extraneous transients were present in the image at the time of change. Each image consisted of 12 different objects, sorted by color into three different levels of probability of change. In Experiment 1, change of one object was detected and identified frequently in objects having the highest probability of change (central interest), which we hypothesize were mainly visited by attention. Changes in other objects with a lower probability of change (marginal interest), however, although detected efficiently were unlikely to be identified. Identification improved for less attended objects if the changed stimulus simply disappeared, allowing visual persistence to hold information about the object until attention could be shifted to it (Experiment 2). Contrary to previous findings showing that response times (RTs) for luminance change detection in a multi-element display are not altered by attention, we found changes in objects of central interest to be detected faster than in objects of marginal interest when objects' identity was to be held in working memory. However, no differences in RTs emerged in the same change detection task when objects' identity was not stored in working memory (Experiment 3).