[comp.ai.philosophy] Emergent properties in VERY simple systems

avr@cbnewsj.att.com (adam.v.reed) (10/06/90)

I'm rather surprised to find that everybody else's (at least on
this newsgroup) notions of "emergence" are so different from my
own. For years, I've been using the "emergence" of the ability
to store information over time (engineers say "memory", but some
biologists and psychologists find such usage unbearably
anthropomorphic) when two NOR gates are combined into a flip-flop
(separately, NOR gates do not have memory, a flip-flop does)
to help my students grasp the concept of "emergence". Now I find
the denizens of this newsgroup are talking of "emergence" as
something applicable only to very complex systems, or to
properties not predicted, or not understood, or not capable of
extrinsic definition. What's going on?

My own notion of an emergent property of a system would include
any property which is "qualitatively different" from any property
of its component parts. I think that the notion "qualitatively
different" here means "defined by criterial operations which
cannot be derived from the criterial operations defining the
properties of the component parts by simply changing the value(s)
of one or more parameters". Thus, when two one-gram masses are
combined into a two-gram mass, the latter's property of weighing
two grams is NOT an emergent property, since it differs from each
component's property of weighing one gram only in the value of a
parameter. Memory, on the other hand, does NOT <correspond except
for parameter values> to any property of separate NOR gates.
Hence, it fits my idea of an "emergent property".

Emergence, in this sense, is a very useful notion for separating
interesting and uninteresting properties that systems get by
virtue of the organization and interaction of their parts. Why do
some people have a need to make it into something esoteric and
mysterious?
					Adam_V_Reed@ATT.com

sarima@tdatirv.UUCP (Stanley Friesen) (10/09/90)

In article <1990Oct5.210737.4836@cbnewsj.att.com> avr@cbnewsj.att.com (adam.v.reed) writes:
>
>I'm rather surprised to find that everybody else's (at least on
>this newsgroup) notions of "emergence" are so different from my
>own. For years, I've been using the "emergence" of the ability
>to store information over time
> when two NOR gates are combined into a flip-flop
>(separately, NOR gates do not have memory, a flip-flop does)
>to help my students grasp the concept of "emergence".

This seems like a perfectly valid, and extremely simple, example of emergence
to me.  It may even be the best example yet for explaining what the word means.

> What's going on?

I think some people are confused by the fact that we often *assume* emergence
when we do not have an explanation for some system property.  Thus it might
appear that emergence applies to *inexplicable* properties only.  But, as I
see it, the assumption of emergence in this case is philosophical [i.e. there
is an explanation for everything, so we assume an explanation and take the most
obvious one as the first guess] [At least that is how I approach it, this is
probably a better assumption to make than to assume acausality or supernatural
causality].

>Emergence, in this sense, is a very useful notion for separating
>interesting and uninteresting properties that systems get by
>virtue of the organization and interaction of their parts.

Exactly.  The emergent properties of a system are the ones that must be studied
in the context of the system, bercause they do not exist at the lower levels.
Thus the appropriate domain of biology is the emergent properties of living
things, since the other properties are the domain of chemistry and physics.
The appropriate domain of AI research and cognitive science are the emergent
properties of cognitive processors, such as brains and AI software.
-- 
---------------
uunet!tdatirv!sarima				(Stanley Friesen)