Abstract
Beck, Angelone, & Levin (2004) demonstrated that changes that are likely to occur in the real world (probable changes: e.g., a lamp changing from off to on) are detected more frequently than changes that are unlikely to occur in the real world (improbable changes; e.g., a blue lamp changing into a green lamp). Top-down knowledge may influence CD performance by directing attention to some aspects of the scene over others. Alternatively, top-down knowledge may influence the memory representations, the probability of comparing representations, and/or the probability that a change signal will be discounted during a post-perceptual decision process. We examined these possible explanations using the same stimuli from Beck, et. al., (2004). Participants' eye movements were monitored during a CD task for probable and improbable changes and during a memory test for the scenes. Eye movement data revealed that when a change was detected, more time was spent looking at the change area of the scene prior to the change than when the change was not detected. However, selective attention does not appear to be the cause of the change probability effect, because probable and improbable change areas were examined equally often even though probable changes were detected more frequently than improbable changes. Memory performance revealed equivalent memory representations for both types of changes, and explanations of false alarms to no-change scenes were no more likely to be descriptions of probable changes than improbable changes. Therefore, it appears that both types of changes are attended and represented equally and that a post-perceptual decision process is not the cause of the effect. In support of these findings we found implicit CD (longer fixation durations on the change area when the change was not explicitly detected) for only the probable changes. Therefore the change probability effect results from a failure to compare the pre- and post-change representations for improbable changes.