RE: Review of Novamente & a2i2

From: Justin Corwin (
Date: Sun May 05 2002 - 22:43:42 MDT

>In case it wasn't evident from context, this message was intended for

Maybe so, but it does raise an interesting point.

See, I've always been perplexed by something. Most people seem to identify
and categorize concepts and thought-complexes(what most people jargonize
into meme-plexes or paradigms) very very differently than I do.

I was on the sidelines for this exchange wondering to myself what Eliezer
would make of the fact that I spend much of my life with a 'silent' mind.
There aren't any named concepts or thought streams running through my head.
Just virtual-kinesthetic feelings. It's only when I switch on what I think
of as my analytic mind (or occasionally I turn on an internal narrator for
preciseness) that the inside of my head gets noisy.

I have repeatable used concepts that I can't think about in English without
confusing myself. For example, (I have to kind of explain around this, so
bear with me) in some situations a certain confluence of factors (people,
activities, objects) give a 'directionality'(for lack of a better term, to
events around me. I've learned to rely on this directionality in order to
skew social dynamics in my preferred direction. Strangely, the best metaphor
I can come up with is the way I can sense intent when sparring or fighting
with someone(particularly kendo and fencing).
This feeling actually had me convinced for a while that I was
psychic(embarrassing experiments in my early teens to follow).

Other people don't seem to have structures like this.

Yet others have structures that I have difficulty understanding. I have a
friend that can't think about music without categorizing it by it's 'shape'.
(Most Bach materializes as green polyhedrons, apparently) I have another
friend who has internal invented words for smells!!!

This sort of variance in internal thought life really creeps me out, and
makes me think that introspection on the subject of cognitive science may be
really really wrong.

Some time ago now(yipes I'm getting old, nostalgia, run! ) I devised a
metric for determining internal cognitive process based on some simple
selectors. It was intended to be plugged into customized learning protocols
I had designed. Some of the learning tests was surprisingly effective. We
had kids speaking russian and italian way ahead of median learning curves,
and I broke several people's problems with learning advanced geometry. We
didn't exactly had new Euclids running around, but it was impressive to our
investors, anyway.(One of the test cases was an investor's kid who
beforehand had been flunking algebra I)

What we learned from our experiments is that Internal modeling of concepts
is sometimes bizarrely varied between people, and that the modeling of said
concepts is often a bottleneck in learning. Translation is a bitch.

One of our more repeatable conclusions is that the best 'generalized'
learners were those who could use generic mental structures to the greatest
effect. Particularly those who used mental object analogy, and visual
categorization specifically.

Anyways, didn't mean to ramble too much, but this kind of plugs into the
wordless concept debate here.


-I did at one point have a bit of investor money lined up for an
experimental accelerated learning program for a small sample population,
based on my research, and some example projects the team had.

-Yes, I did this in high school. I was frustrated by my lack of ability in
some areas, and originally approached this as a self improvement project
that snowballed. I unfortunately was never able to helpfully categorize my
own mental gymnastics. So my learning is as slow as ever. I still believe
that if I could get someone competent trained in the system I designed, an
accurate profile could be designed, and my various learning disabilities and
disconnects could be corrected, But in the meantime, tough cookies.

-I apologize to anyone I've been corresponding to, and the people who've
been getting emotional and nonsensical emails, or none at all. I'm in the
midst of kind of a familial tragedy, and most of my projects and
correspondence have been derailed. (which is probably also why I'm rambling
about one of my most beloved and failed attempts.)(Spectacular failure,

Anyway, Questions to Ponder:

1. Do these mental-object differences imply significant neuroanatomical
differences between people? (we never really investigated it, we didn't have
the expertise or the time. We were more about what worked, than why)

   1A. Does that matter at all, if it does?

2. What impact on artificial minds does this difference entail?

  2A. Are these neccesary differences, or can they be ignored?

  2B. Do these differences convey significant advantages in cognitive
processing that an artifical minds may take advantage of?

alrighty, back to your normally scheduled Ben vs. Eli AI-athon:
> > -----Original Message-----
> > From: []On Behalf
> > Of Ben Goertzel
> > Sent: Sunday, May 05, 2002 8:51 PM
> > To:
> > Subject: RE: Review of Novamente & a2i2
> >
> >
> >
> > One more thing.
> >
> > I am quite perplexed by your notion that concepts need to be *named* to
> > useful.
> >
> > Don't you believe in content-addressable memory???
> >
> > This is one of the design principles that Novamente inherits from
> > attractor
> > neural nets (ANN's), actually.
> >
> > Semantic nets, which you dislike, involve addressing concepts by name.
> >
> > ANN's involve addressing concepts by *specifying part of their
> > contents* or
> > *specifying a collection of related entities*.
> >
> > In Novamente (an integrative design), *some* concepts are addressable by
> > name, but *all* concepts are addressable ANN-style (because concepts are
> > represented by "maps", which are much like attractors in ANN's).
> >
> > -- Ben
> >

Join the world’s largest e-mail service with MSN Hotmail.

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:38 MDT