Abstract
The human visual system typically tracks the position of objects as they move. However, when disruption occurs (e.g., as objects are occluded and disoccluded, an eye movement occurs, or when objects spontaneously disappear and reappear), we must determine whether or not the objects have moved. In most real-world perception, stable aspects of the surrounding environment provide landmarks for this recalibration process. Previous studies showed that when two objects are briefly viewed and then removed from view, the object reappearing first serves as a landmark for the object reappearing later. This results in the misperception that the second object has moved when, in fact, only the first one (the landmark) actually did. We explored whether this “landmark bias” was due to the objects' abrupt onset/offset by removing and revealing the objects more naturally. If the landmark bias represents a general process in which people treat the first object to reappear as the stable object, then observers should continue to see the second object as having moved. Alternatively, if the landmark bias results from a disrupted initial representation after sudden onset/offset, then the landmark bias should be eliminated. To test this hypothesis, two objects appeared side by side after which a moving occluder entered from one of the four edges of the screen, temporarily covering the objects before exiting. The objects could be occluded simultaneously (top/bottom entrance) or sequentially (left/right entrance), and revealed simultaneously (top/bottom exit) or sequentially (left/right exit) to mimic the traditional landmark test procedure. When the objects disappeared and reappeared more naturally, observers showed no landmark bias to misperceive the second object as having moved. This pattern also held for invisible occluders which provide no location cues, demonstrating that vanishing objects are treated differently than objects that gradually disappear.