Mind.forth: What does it do?

Mentifex mentifex at scn.org
Sat Dec 5 17:15:43 EST 1998


In response to an e-mail which may or may not also be a post in the
comp.robotics.misc newsgroup, it is asked:  What does Mind.forth do?

Since very truly yours Mentifex/Arthur here am eager to answer the
question but want to post the information, I am extracting the points
raised and answering them one by one.

True, the initial version of Mind.forth being assembled gradually at
http://www.geocities.com/Athens/Agora/7256/m-forth.html  takes in
three-word sentences (and doubly stores each word:  as an auditory
engram and as a lexical concept.  Although the current December 1998
version does not yet issue an output, the output will be a simple
sentence "thought up" by the Mind.forth program).

The only knowledge base ("KB") in Mind.forth is the sum of all its
experience.  You tell it things (please don't lie) and it stores
your factual statements as associative relationships *among* its
concept fibers.  The building up of the KB constitutes learning.

Since the e-mail asks about the documentation of Mind.forth, here
is the plan.  Mind.forth is being written into a 30K data space,
which corresponds roughly both to the file editing limit on Geo-
Cities.com and to the 1024 X 30 size of thirty MVP-Forth screens.
Part of the plan is to keep 30K Mind.forth small enough for robots.

However, each of the thirty Mind.forth screens already has its own
http://www.geocities.com/Athens/Agora/7256/screen15.html etc. Web
page with a 30K space for thorough explanation and code-comments.
The plan is to document each screen of code as it gets written,
so that eventually absolutely nothing in Mind.forth is unclear.

Your e-mail laments the choice of Forth, but I have compelling
reasons to do this work not first in Forth but *second* in Forth
after having already released the program on 26nov1994 as Mind.REXX.

Here is my starry-eyed, gosh-a-rootee scenario in emulation of
Linus Torvalds and his Microsoft-killer all-app Linux software:

Keep working on Mind.forth and releasing it on the Web.  At first,
only Amiga owners can key in the actual program using MVP-Forth
as found on the public-domain Fred Fish Amiga Library Disk #977.

But other Forth programmers can adapt the code to *any* version
of Forth and get it working -- if my own Mind.forth works well.

It's those other, brighter Forth programmers whom I am counting
on to port Mind.forth into programming languages that you prefer.
They won't make the effort until they see functionality developing
in Mind.forth -- but even the Mind.rexx program *did* function.

Let's just wait and see.  I am creating *linguistic* thought;
if it works, it will be a platform onto which clever persons
can attach machine vision, robot mobility, real speech, etc.

  /^^^^^^^^^^^\ Syntax Strings Together a Thought /^^^^^^^^^^^\
 /visual memory\           ________   semantic   /  auditory   \
|      /--------|-------\ / syntax \  memory    |episodic memory|
|      |  recog-|nition | \________/------------|-------------\ |
|   ___|___     |       |     |flush-vector     |    _______  | |
|  /image  \    |     __|__  / \  _______       |   /stored \ | |
| / percept \   |    /     \/   \/ Verbs \------|--/ phonemes\| |
| \ engrams /---|---/ Nouns \    \_______/      |  \ of words/  |
|  \_______/    |   \_______/-------------------|---\_______/   |



More information about the Neur-sci mailing list