From: Ben Goertzel (ben@goertzel.org)
Date: Wed Jan 25 2006 - 19:28:20 MST
Hi,
In this vein, I have recently posted on the AGIRI Forum at the following link
a list of books pertaining to developmental psychology that are now on
my bookshelf.
I have not read them all yet so I can't vouch for them all, but I found
Tomassello -- Constructing a Language
to be a particularly good review of the modern language acquisition literature,
Shultz -- Computational Developmental Psychology
to be a good review of modern ill-conceived attempts to use
over-simplistic neural net and rule-based AI methods to emulate young
children's learning processes,
Thelen & Smith -- Dynamical Systems Approaches to the Development of
Cognition and Action
to be a good review of dynamical systems approaches to modeling motor
learning and infantile cognition, and
Noe -- Perception in Action
to be an interesting though in some respects overly extreme argument
in favor of the extremely tight integration of perception and action
in human neuropsychology.
-- Ben
On 1/25/06, Mike Dougherty <msd001@gmail.com> wrote:
> Game theory (if that's what it's called) - to understand the difference
> between the "zero-sum" game of most traditional models, and "infinite
> resource" (for lack of a better term opposite zero-sum) thinking that will
> result from a post-Singularity economy.
>
> I would recommend "the Art of War" (Sun Tzu) as it applies to game theory.
> Most of the war game strategies could be applied to building complex
> behavior from simple(r) atomic concepts. It might be a while before an AGI
> is directly consuming the high-level significance of the text, but if
> developers are writing probability competition algorithms - then "the Art of
> War" is a terse distillation of basic principles.
>
> How about a course in group dynamics? (or sociology, or whatever your
> university calls it) If the AI is modelled as a synergistic collection of
> individual cells, each with primitive motivating forces - then the macro
> level behaviors are influenced by micro-level relationships. Worth
> examining are the edge cases of "mob" psychology, where a group of
> individually good-natured people are swayed by peer-pressure to riot and
> destroy the venue they're inhabiting.
>
> I know there are those who would approach AGI from a computer science
> direction, but I believe that a multi-discipline background helps bring a
> wide range of new ideas to the table.
>
> I don't think you would be talking to yourself - in fact, i think this is a
> very worthwhile thread. I will continue thinking about it, so I can
> contribute more later.
>
>
> On 1/25/06, Richard Loosemore <rpwl@lightlink.com> wrote:
> > Does anyone else have ideas for an AGI syllabus, or would I be talking
> > to myself if I tried to start such a thread?
> >
>
>
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:55 MDT