Accurate visual detection of layout change requires a stable observer position

Purpose. Scene perception and recognition rely on representations of both the identity and layout of objects. The present studies assess the accuracy and stability of such representations by examining the effect of observer position on the detection of identity and layout changes. Methods. On each trial, subjects viewed (for 2 seconds) a simulated 3-dimensional room containing 5 objects. After a -second delay, they then saw either a) an identical room, b) a room in which one object had moved to a new position (layout change), or c) a room in which one object was replaced by a different object (object change). Subjects indicated, as rapidly as possible, whether or not the first and second rooms were identical. Results. In line with previous studies using 2-dimensional arrays of objects, when the simulated observer position was stable throughout the trial, subjects detected layout changes quickly and accurately but detected object changes more slowly and less accurately. When the simulated observer position shifted approximately 35 degrees between the first and second part of each trial, subjects detected substantially fewer layout changes. Conclusions. A brief glance at a 3 dimensional scene allows an accurate representation of the relative positions of objects with respect to the observer. However, the representations seem to be specific to the observer position; unlike conditions with a stable observer, large changes to the observer position necessitate effortful processing to detect layout changes. The stability of layout representations across smaller shifts in observer position is also considered.