next up previous
Next: Representation of Invisible Up: The Predictivity Principle: Previous: Predictivity Implies Prediction

Prediction of Disocclusion Implies Representation of Invisible Objects

How can a visual system predict disocclusion events? Somehow it must maintain a representation of the information that it will need. The representation must carry the properties of the occluded object, such as its position, velocity, and shape. Since no information about the occluded object is directly available while it is occluded, this representation must originate before it becomes occluded. This representation must interact with information about the occluder so that when the occluded object reaches the end of the occluder, a prediction that it will become visible again is generated.

If the occluded object is part of a group of objects, some of which are visible, then the motion of the visible objects in the group can control the perceived motion of the invisible object [2,14]. Therefore, we argue that the representation of the occluded object can be controlled (e.g., its direction altered) in mid-course, even while the object is invisible. This argument favors a persistent form of representation (the inferred invisible object is represented as a real object, with properties of motion, shape, position, etc.), rather than a ballistic form (with a representation impervious to alteration once it has been launched) [9].


next up previous
Next: Representation of Invisible Up: The Predictivity Principle: Previous: Predictivity Implies Prediction