@Article{KosslynShwartz1977,
author = {Kosslyn, Stephen M. and
Shwatrz, Steven P.},
title = {A simulation of visual imagery},
journal = {Cognitive Science},
year = {1977},
key = {},
volume = {1},
number = {},
pages = {265--295},
month = {},
note = {},
annote = {}
}
The imaging area was a polar-coordinate system where the edges were less active, beacuse mental images have boundaries. [269]
Kosslyn 1975: Subjects need more time to process larger images. [270]
The idea behind this, as related to imagination, is that when a person hears something like "a horse with a chicken head" (summary author's example) he or she is able to use the words to retrieve images, or parts of images, to create the hybrid in the visual buffer.
The program works with two images: a chair and a car. [271]
To image the wheel, PUT calls FIND to locate the wheel well in the imaged car skeleton. As detail is needed, parts are added, and the system needs to find out where to put them in the visual buffer. Size is adjusted relative to the skeleton [280]
In this implementation, parts are refreshed in the buffer as needed (when the reasoner must find a part.) [281]
SCAN puts the attended part in the center of the buffer. [Summary author's note: I think Kosslyn's theory has changed by the time he wrote "Image and Brain," because I think I recall there being an attentional window that can travel in that model.] [284, but see justification on p290]
"It seems to us that imagery would most often be used in memory retriveal when one did not have an explicit notation that some object had a sought property." [284]
At the borders of the buffer the image gets degraded-- things in that area might be too degraded for the FIND function to find.