Abstract
How does the visual system keep track of the changes in a dynamic environment? In fact, does it keep track of these changes at all? Observers are usually poor at detecting prominent changes in a scene when the change occurs simultaneously with a brief disruption of the display. This so-called change blindness obviously points to a limit in our ability to represent and process visual scenes. But where exactly is the locus of this limit? One view suggests that the number of objects that can be represented at a time is strongly limited. A different view suggests that as long as the scene is in view, a large number of objects may be represented. However, due to limited stability, these representations may be volatile and easily overwritten by the post-change scene. Both accounts seem to agree that the representation of the changing object is lost under change blindness. In this study, we investigated the fate of object representations when changes go unnoticed. We presented scenes consisting of real world objects, one of which changed on each trial, while recording event-related potentials (ERPs). Participants were first asked to localize where the change had occurred. In an additional recognition task, participants discriminated old objects, either from the pre-change or the post-change scene, from entirely new objects. Participants performed poorly at recognizing pre-change objects compared to post-change objects. This result is consistent with the idea that pre-change object representations are overwritten by the post-change scene. However, ERPs differed between pre-change objects and new objects, indicating that a weak memory trace of the pre-change object had persisted. Importantly, this effect was found even when both localization and recognition failed. This finding suggests that in addition to limited capacity and overwriting, change blindness can also occur when the visual system simply ignores pre-change object representations.
Meeting abstract presented at VSS 2012