BOCK@INTELLICORP.ARPA (Conrad Bock) (06/14/88)
Date: Mon, 13 Jun 88 01:27 EDT From: Conrad Bock <BOCK@INTELLICORP.ARPA> Subject: Resource limitation applied to hypostatization and consensus reality To: ailist@ai.ai.mit.edu I would not entirely recommend Winograd and Flores' book, but a way occurred to me to make it (and consensus reality) more intelligible from a computer scientific viewpoint. If we agree that our minds are constructing theoretical entities from patterns in our input-output data, then we might also agree that there are so many theoretical entities that only a few of them can be open to revision at any one time, given resource limitations. This is hypostatization (ie, taking concepts to be reality) as a computer scientist might express it. Since many of the concepts we use are learned from other people, we might assume that many of our hypostatized concepts (which are part of our reality) are due to social interaction (as Hayes suggested). Hence, reality is partly social. A computer scientist might say the concepts are in, or have been put into, the hardware or at least a lower level language. Winograd and Flores might call these concepts (I'm interpreting now) ``practice'' or ``background''. That's the proposal. There's already a hole in it as far as Winograd and Flores go: since we as computer scientists build our machines, we don't have as much interest in situations where the machine was already built before we got here; that's natural science. I think Winograd and Flores are concerned with the situation where we are the machines that are already built (practice is ``already doing''), so the causality is from background to concepts, not the other way around. Conrad Bock -------