-- Messages from file: [SUMEX-AIM]<LENAT>MESSAGE.TXT;1
		 -- Monday, February 5, 1979 15:17:49 --

   200  8 Jan  Pcohen                Make appointment for Thur. or Fri.
   203  8 Jan  Pcohen                Meeting Friday 11-12
   282 22 Jan  Pcohen                From Pcohen re:AM
   336 30 Jan  Pcohen                FROM Pcohen Jan 30.


200 -- ************************
Date:  8 Jan 1979 1002-PST
From: Pcohen
Subject: Make appointment for Thur. or Fri.
To:   lenat

Dr. Lenat, I am a student of Ed Feigenbaum's in a special program, and I would
like to talk to you about your work. Ed said you are working on a program
that has discovered properties of numbers, and that in general you are working
with heuristic-guided discovery of new knowledge(?)(If I have this wrong, it
isn't Ed's fault, but my own for not understanding properly.)
   My background is in cognitive psychology, but I am happy to disavow it all
because I think it produces murky results. The problem of how people discover
things has been thrown in the "intractable" bin by psychologists, I think,
and I'd welcome an opportunity to explore it.
   I hope I haven't misunderstood completely what you are doing. Could I speak 
to you later this week about it? Ed tells me you will be out of town until
Wednesday; could we get together after that? Thanks, Paul Cohen. (pcohen@sumex).
-------


203 -- ************************
Date:  8 Jan 1979 1216-PST
From: Pcohen
Subject: Meeting Friday 11-12
To:   lenat

I look forward to meeting (11-12) with you. Thanks, Paul.
-------


282 -- ************************
Date: 22 Jan 1979 1234-PST
From: Pcohen
Subject: From Pcohen re:AM
To:   lenat

Dear Doug, I have been reading your thesis following our conversation of
last week and one question in particular comes to mind:

	Are tasks on the agenda ever explicitly or implicitly
	(via the Worth rating) organized into "conceptual groups"?

It isn't a very clear question, I'm afraid, but let me clarify what I'm
getting at: AM explored losers and trivial concepts; in only two instances
did it explore something you thought was a loser but was actually a winner.
The point is that you have some pretty good intuitions about what is
worth exploring that AM lacks. Now, what follows may well be a loser, but
I'm wondering whether the agenda couldn't be organized more than it is 
now by some meta-function that corresponds to what we call intuition.
One can imagine a set of heuristics, provided by experts, of the following
sort: "Whenever I postulate an X I always try to instantiate it 
immediately", or "Whenever I postulate an X I try and derive it in another
way", etc.
   One thing stands clear from your thesis, and that is that "open-ended"
search is by no means unconstrained. I'd like to work on how tasks are
justified, because I think that provides in-roads to the psychology of
discovery that are lacking in your thesis. I think there are human criteria
for the evaluation of "interestingness" that might be instantiated in AM,
for example, Koestler's thesis that a concept is interesting if it is
the intersection of two "planes". (Plane and intersection are undefined,
but need not be.)
   Let me know if this sounds worthwhile. Thanks, Paul.
-------


336 -- ************************
Date: 30 Jan 1979 0924-PST
From: Pcohen
Subject: FROM Pcohen Jan 30.
To:   lenat

Dear Doug, I'm still waiting for you to reply to my message. Since then 
I've had some more ideas, this time about the development of heuristics.
At first thought it seems it may be worthwhile to look at the way 
heuristics develop in children. I was unable to tell from your thesis
whether the machine developed concepts in the same order as children do, but
but if not it may be because children develop heuristics over time while
AM had an initial set all of which were immediately available.
I would be interested in exploring the development of mathematical
concepts in children from an AM perspective with some people in psychology,
and if you think it's an interesting possibility, let's get
togetyher and discuss it. Then I can go talk to Flavell, or someone.
Please respond to this, if only to say you're not interested; I have
to put together a committee by May, and it's February already.
                   Thanks, Paul.
-------