An Evolutionary Perspective

[Evo-Mark]

Hello everyone. Many thanks to James Tauber for inviting me to this
group (at the HTM Workshop).

After reading many books on the subject of evolution, I consider
myself something of an expert. I hope to bring to this group a
perspective that reflects careful thinking from an evolutionary point
of view. In fact, I will start by positing that all forms of creative
intelligence emanate from some form of an evolutionary process. Thus,
human thinking reduces to an iterative and (mostly) deterministic
process whereby patterns of neural firings are able to rapidly evolve
toward whatever plan is expected to maximize future happiness (and
minimize future discomfort).

The HTM architecture is a crucial part of the process, because it
automatically captures spatio-temporal patterns (behaviors) in ways
that allow them to be easily combined into various plans. Linguistic
elements evolve so as to conveniently express the various elements of
those plans.

Any reactions? Or, should I elaborate further.

--Mark

[Rich]

I have been an engineer/computer programmer for quite a few years,
and I avoided anything like biology from junior high school all the
way
through two EE degrees and nearly a decade of programming. My first
jobs were with defense contractors, but then I went to NIH to write
programs that did statistical processing of fMRI images. But during
that
boring programming time on military contracts, I learned and argued a
lot about evolution over at the talk.origins world. The years of
studying
evolution and then working with people researching brain function
finally convinced me to come back and get a PhD in Neuroengineering.
It was nearly exactly 25 years between my last biology class in school
and my first neuroscience class at Emory University.

That's just a long way to say that I tend to approach all of this
from an evolutionary viewpoint too, although my primary
interest is in useful applications of this approach to current
engineering problems.

I lived in Finland for a couple of years, learned to speak
Finnish fluently, and got hooked on linguistics at the
library. That is extremely popular in Finland, since it's
both an extremely conservative language and a very
synthetic and agglutinative one. It might even be a nice
control case for any HTM process we come up with, since
it's a fairly well-studied language, and it's fairly suitable for
existing intelligent machine processing to compare it to.

rich

[ma...@frolix.com]

Rich,

I also have advanced degrees in EE. My profound interest in
evolution didn't happen until after I went back to school for an
MBA. Only when I read Richard Dawkins book, The Selfish
Gene, did I fully understand the origins of language, intelligence,
and everything else of value in the world. The two concepts
in that book, which allowed everything to fall into place for me,
were: (1) looking at evolution from the perspective of the replicators,
and (2) realizing that patterns of behavior (what Dawkins refers to as
memes) are just as valid a basis for evolution as patterns of DNA
(genes).

Dawkins allowed me to reconcile a long held belief of mine, that
the mind is a mechanistic (mostly deterministic) device, with the
evolutionary emergence of intelligence and all its products (including
and especially language). I now see clearly that there
is no source of creativity anywhere in the universe other than
evolution. I have become convinced that even the process of
creative thinking amounts to nothing more than a very rapid process
of neural firing patterns evolving.

It is my belief that no progress will be made on understanding language
until it is common for linguists to look at the human mind as
a mechanism for facilitating the evolution of behavior patterns (including
and especially patterns of word usage).

This is where the HTM model comes in. It can automatically collect
patterns of word usage, and patterns of patterns, and patterns of
patterns of patterns, etc. I have very definite ideas as to how these
hierarchical representations then interconnect by one mechanism at high
levels and by another mechanism at low levels to produce a semantic
web, and how neural firing patterns go up and down hierarchies,
jumping from one concept to another concept (from one hierarchy
to another hierarchy), by way of these high level and low level
connections. This forms the basis for what is known as the "train of
thought."

That's enough for now.

--Mark