One thing the objectivist theory of meaning bypasses completely is how the universe gets divided up into concepts (the cosmic cookie cutter problem [Lenat and Guha, 1990]). We look at the color spectrum and divide a continuous space into discrete regions, which we call ``red'', ``blue'', etc. We look at orientations, and come up with three main concepts, ``vertical'', ``horizontal'', and ``diagonal''. We have the ability to see an army troop as a single unit, and a car as a collection of interacting components. We look at a room, and see a chair, a table, a person etc. A chemist could see the room as a bunch of molecules, or a particle physicist could see the same room as continuous waves. Who knows how a fly sees it? Our divisions are in no way universal or most natural. Cyc, on the other hand does not look at a universe and try to make sense out of it. It passes the concept acquisition problem completely to its designers, and its units of thinking are predetermined.
One could use the human with a teletype picture to argue that what is being built in Cyc is the model of a mind frozen after reaching adulthood. So, the problems of concept acquisition has been solved. The system has learnt to see the room in terms of tables and chairs. All we need to do now is understand new information in terms of these concepts and make inference over them.
The argument would be valid if the machinery that supports perception and concept acquisition does not support learning and inference later. As I will try to show in the following sections, they seem to play a very important role.