DPRG List  

DPRG: Deep Thoughts and rambling

Subject: DPRG: Deep Thoughts and rambling
From: S Sunder sunder at rocketmail.com
Date: Fri Jan 9 04:41:12 CST 1998

- ---Jim Brown <jgbrown at spd.dsccc.com> wrote:
> I've been thinking about an approach:
> Case 1:
> A baby knows nothing, it just lays there and looks around.

Why do you says knows nothing?

> Then, they realize that they're hungry and scream out in pain.  (natural
reflex, output)

Hunder is inside sense and screem is not the foist idea to
get milk. Screem is there as an intutive action in the baby
perhaps for muscle excercise. It gets coupled to the 
demands when it learns that the mother offers milk only with
this signal. Success is a pleasure and the failure a pain.

> Mommy comes around and gives them a bottle. (extenal input)


> They start sucking the bottle. (natural reflex)

That is not so simple. A drop of milk in the mouth triggers
many harmone related process ans the baby learns to 
accelerate the process.

> They feel better so they quit crying. (pain gone, reflex not necessary)

Yes. The pain of hunger may go this way. Whem other pain 
comes then the baby is forced into that state and the baby
is actually not looking for the mother but to getrid of it
with inferior knowledge.

You cry in a hopeless state and that is intutive.

There is a sequence in which the learning abilities develop.
Like you have first standard and then second standard
education. You don't just all up unless you are alien or
something like that.

Shyam India

> After many times, the baby grows accustom to the familiar pattern  (pattern
> Case 2:
> During idle time the baby tries to figure out how to move it's muscles. 
> The baby figures out that it can roll over and crawl (repituar building)
> Case 3:
> The baby sees a toy and has to have it (want)
> The baby figures that if it crawls to it, it'll have it.  (cognition)
> 	(what it can do with what it knows so far)
> Case 4:
> The baby puts everything in it's mouth.  (experimentation)
> Some things taste sweet (covet)
> some no taste (indifference)
> some bland.  (satisfactory)
> Some things taste sour.  (rejection)
> Case 5:
> The baby goes near a power socket (want)
> Mommy spanks baby (painful deterrant)
> Baby cries in pain (reflex action)
> maybe happens a few more times and then
> the baby learns not to go near the socket (deturred/dislike)
> So from the above cases, to make a robot as smart as a baby it would have to
> have or be able to do the following:
> 	experiment,
> 	react (reflex),
> 	have input & output,
> 	pattern recognition,
> 	build a list of learned abilities
> 	generate wants
> 	figure out how to achieve those wants
> 	have likes (which want is more desirable or to be coveted)
> 	have dislikes (which want is less desirable or to be avoided)
> In my opinion, a simple pattern matching net will not acheive these goals.
> The system would need to be able to somehow have a thread of execution
> that could generate a procedure.   It would then need
> to be able to alter that procedure for (1) whether that procedure worked
> or didn't work, and (2) for efficiency, and (3) go back and revise the
> procedure based on wiser discernment.  These repituar of procedures would
> be remembered in a list of useful procedures to accomplish a task.
> So in this system, there would need to be a set of base procedures that
> the system can make simple procedures with.  There would also be a base set
> of positive and negitave feedback such as hungry or full and feeling fine
> and feeling pain, and getting a toy and needing a toy.  The system would
> try to build procedures, possibly ramdomly at first, to see if it can acheieve
> any desireable results.  After the results have been met, it then tries to
> see if efficiency or a better way of doing things can be achieved.  The
> procedures would be stored as both the list of instructions that were
> ran, and the feedback that those results gave.  Later the procedure
> could be compared to the results to build.
> From the procedures it creates, it should also be able to run them in some
> sort of simulation mode from responses that it already expects.  This mode
> would allow the system to test procedures without actually moving about
> before actually deciding on the best plan of action.
> These are just observations and trying to figure out what an intelligent
> system would need.
> - - ____ - - - - - ___ - - - - - - - - - - - - - - - - - - - - - - - - -
>     \/\_\@ ____   /  /\ __  ___          ___  http://www.dprg.org (Jan 17)
>     / //\ / / /\ /--/ //\_\/\_/\ /\/\/\ /\_/\ http://users.why.net/jbrown
> /__/ // // / / //__/ // / /__/ //_/_/ // // /(972)519-2868, (972)495-3821
> \__\/ \/ \/\/\/ \__\/ \/  \__\/ \_\_\/ \/ \/jgbrown at spdmail.spd.dsccc.com
> My employer won't claim these opinions so I'm giving them away for free.

Get your free @yahoo.com address at http://mail.yahoo.com


More information about the DPRG mailing list