1.
Theory of
AI Textbook Algorithm Steps: Code the Reify Mind-Module
/^^^^^^^^^^^\ Reify Gives Names to Abstractions /^^^^^^^^^^^\ / EYE \ CONCEPTS in SEMANTIC MEMORY / EAR \ | | | | | _______ | ________ | | _______ | | | | / \ | /"katzen"\ | | /old \!!!!|!!!|!| | / German \ | / "essen" \ | | / image \---|---|-+ | \ syntax / | \ "fische" / | | \ recog / | |c| | \_______/ | \________/ | | \_______/ | |a| | _______ | _________ | | | |t| | / \ | /"feles" \ | | | |s| | / Latin \ | / "pisces" \ | | visual | e| | | \ syntax / | \"manducant"/ | | | a| | | \_______/ | \_________/ | | memory | t| | | _______ | | | | | |f| / \ | | | channel | | |i| / English \---|-------------\ | | | | |s| \ syntax / | | | | | | |h| \_______/ | | | | _______ | | | | flush-vector| | ________ | | | /new \ | |_|_|_ __V____ | / "cats" \ | | | / percept \ | / Psi \---------/ En \---|-/ "eat" \| | | \ engram /---|-/concepts\ reify / lexicon \ | \ "fish" / | | \_______/ | \________/-------\_________/--|--\________/ |http://mind.sourceforge.net/diagrams.html shows a Theory of Mind.
The Reify operation, as shown in the brain-mind diagram above,
is a function of the semantic memory of the sentient mindgrid,
where a linguistic superstructure of Chomskyan syntax flushes
out the momentarily activated deep mindcore Psi concepts so
that the corresponding lexical concept fibers are activated for
any language spoken and understood by the Robot AI Mind.
2. A Rose is a Rose
is a Rose
A British playwright once said, "A rose by any other name doth
smell as sweet," meaning perhaps that the concept of rose may be
expressed by a variety of words in a variety of human languages.
In a biological brain, the associative-tag fibers carry the
activation from the mindcore concept fibers to the lexical
control fibers, which in turn (under the guidance and control
of syntax) re-activate words and morphemes stored as memories
of sounds in the self-perceiving auditory memory channel --
where the mind "hears itself think" and thus perceives itself.
In the primitive Robot Mind, software flags function
in lieu of associative-tag fibers to flush out activations
from the Psi mindcore into the semantic memory lexicon.
http://mind.sourceforge.net/ai4u_157.html
is an overview of Mind.
// Reify() is called by nounPhrase or verbPhrase to flush abstract
// Psi concepts into the real names of English language reality.
function Reify() { // ATM 21jun2002; or your ID & date.
act = 0;
// In the next code, we drop the start of
// the search back from "t" to time-of-voice "tov"
// so that only pre-existing engram nodes prior to
// current generation will govern the generation.
for (i = tov; i>midway; --i) {
Psi[i].psiExam(); // inspect psi1 (act)ivation
if (psi1 > 0) {
act = psi1; // get the activation level and...
enx = psi6; // ... the transfer-to-English flag.
if (enx > 0) {
for (j = (t + 1); j>midway; --j) {
enLexicon[j].enExam(); // inspect English nodes
if (en0 == enx) {
if (act >= en1) { // Moncur p. 87.
enLexicon[j] = new enNode(en0,act,en2,en3,en4,en5);
} // end of fourth-deep if-clause
} // end of third-deep if-clause
} // end of inner backwards loop through Psi
enx = 0;
act = 0;
} // end of second-deep if-clause
} // end of outer if-clause
} // end of outer backwards loop through Psi
act = 0;
} // End of Reify(); return to nounPhrase or verbPhrase.
\ REIFY is called by nounPhrase or verbPhrase to flush abstract
\ Psi concepts into the real names of English language reality.
: REIFY \ ATM 22jul2002; or your ID & date.
0 act !
midway @ tov @ DO \ Search before the time-of-voice.
I 1 psi{ @ 0 > IF \ If any psi has a positive "act"
I 1 psi{ @ act ! \ get the activation level and
I 6 psi{ @ enx ! \ the transfer-to-English flag.
THEN
enx @ 0 > IF \ If the "enx" flag is positive,
midway @ t @ 1+ DO \ inspect English vocabulary nodes
I 0 en{ @ enx @ = IF \ for a match with "enx".
\ For any concept transfer only the highest `act':
act @ I 1 en{ @ > IF \ If a lower "act" is found
act @ I 1 en{ ! \ transfer the higher "act".
THEN \ end of check for lower activations
THEN \ end of check for Psi-to-En matches
-1 +LOOP \ end of backwards search in English lexicon
THEN \ end of check for positive "enx" in Psi
0 enx ! \ Reset the transfer-to-English flag.
0 act ! \ Reset the act(ivation) level.
-1 +LOOP \ end of backwards sweep through Psi array
0 act ! \ Reset the act(ivation) level.
; \ End of reification; return to nounPhrase or verbPhrase.
5. Analysis of the Modus Operandi
The AI source code finds active concepts in the deep mindcore Psi,
then uses the "transfer-to-English" variable "enx" to activate
the related "fex" concept in the English lexical array En.
In machine translation (MT), the AI Mind may pass activation
from the deep mindcore Psi up into the lexicons of two or more
separate languages in which the Robot AI Mind is ready to think.
6. Troubleshooting and Robotic Psychosurgery
Typically an AI programmer will troubleshoot the Reify module
by temporarily inserting code to show what activations are being
transferred from the deep "Psi" concept layer and what activations
are appearing as a result in the shallow lexical layer of language.
In the JavaScript AI Mind, a JavaScript "alert box" will do the
trick for both the Psi array and the En(glish) lexical array.
In Forth, a more ad hoc approach may be required.
Although the AI programmer is looking for the up-transfer of
activation from particular concepts being used in a test case,
it should be kept in mind that all currently activated Psi concepts
will be sending at least a modicum of activation up into the
shallow lexical array. Such a general upwards transfer of
conceptual activation plays a role in the functioning of
concepts still lurking in the subconscious after playing a role
in the generation of a sentence of thought. The train of thought
in an artificial mind may "jump the track" from one topic to another
if sufficient subconscious activation builds up on concepts that
have not yet recently been thought about. Such abrupt changes of
the topic of thought may play a role for insight and creativity.
7. Reification Resources for
Seed AI Germination and Evolution
