Sep 4, 2009




Robert Hecht-Nielsen from UCSD developed something maybe extreme. Yesterday I came across the "Confabulation theory". I don't seriously know anything about it, just:
Assuming, that thinking is an offspring of moving - evolutionary and functionally -, this theory applies principles of the neuro-muscular system to the brain. And they are using network modelling. And it is a general theory of mind/brain. Impressive is following application of the theory:
An "empty" system (program build according to the theory) was fed with text of a newspaper. After that it was able to produce (output):
1.) grammatically correct sentences. (Without being taught grammar)
2.) If you give him a (random meaningful) sentence, his output is...
...well, check out your self. One example:

The program read:
“He started his goodbyes with a morning audience with Queen Elizabeth II at Buckingham Palace, sharing coffee, tea, cookies and his desire for a golf rematch with her son, Prince Andrew. The visit came after Clinton made the rounds through Ireland and Northern Ireland to offer support for the flagging peace process there.”


The program then generated following sentence:
“The two leaders also discussed bilateral cooperation in various fields.” The output is the conclusion the program draws on bases of the (random-news-paper-)-feed and resembles - in order - the words, it calculates as most expedient to follow the last two sentence it was input.

I thought, wow.

I tried to find more information but all I got, is:
He wrote a book and a few publications, there is a little bit in xyPedia, he seem to be still at UCSD and is (still?) Vice President of a company, that develops and applies technologies for decision management. (?!)

"Confabulation theory" produces no search results in
- web of knowledge
- IEEE
- ACM

Well, maybe we will hear more about him in the future, maybe not.
Who is interested, a talk he gave at IBM:



Btw., more about network modelling and general theory of mind: Jeff Hawkins

Followers