"Many things that we regard as physical are actually psychological." In other words, many categories are defined functionally, not physically. (NB:This was a basic and early move by Miller and Johnson-Laird in their book on language and perception.)
Because of this, "we need to combine at least two different kinds of descriptions. On one side, we need structural descriptions for recognizing chairs when we see them. On the other side we need functioanl descriptions in order to know what we can do with chairs." But this is not simple association. Instead, there are presumed to be highly specific links between structural and functional descriptions. "With that knowledge we can do amazing things, like applying the concept of a chair to see how we could sit on a box, even though it has no legs or back!" This is why uniframes need to incorporate this kind of knowledge.
(NB: The purpose of detailed structure/function mappings is to provide generalization of knowledge. It is still not clear how these mappings are instantiated -- eg. by agents.)
Pearl Street | Society Of Mind Home Page | Dawson Home Page |