ABSTRACT
This paper sketches an approach to the computational modeling of perceptual organization (PO). I see PO as an integral part of cognition. This means viewing PO as a process of object recognition, in which a generic model of the collective properties of objects is used, instead of a specific model of individual objects. That is, PO is part of the process of visual memory. I also take the view that processes analogous to PO can play a similar role in nonvisual memory. Memory is an interaction between some novel situation ormemory cue and prior knowledge. I propose thatmemory proceeds by first using genericmodels to analyze the current situation. The results then trigger more specific items in memory. This process can be repeated until a good fit is found between specificmemory items and the current cue. In this view, PO is the first step in the process of bringing our prior knowledge of the world into alignment with a current image of it.
