|
Hi Tim,
Yes, it would seem that your views diverge from my father's in
this. His approach was that "entailment" was the underlying pattern of
organization of a system, the pattern of relations
created/constrained/maintained by the organization. Entailment is the innate
quality, the causal potential. Causality is what is expressed; what happens in
the universe. So, Aristotle's Categories of Causation is a form of analysis
intended to help science get at the entailment underneath causality; "Why does
this happen?" or What causes this causality? And there, I think, is the
confusion-- Cause, causal, causality, causation... they are not quite all the
same. In particular, "causality" refers to all the action in the universe; all
the behaviors and observables, all the properties of systems as expressed. These
are effects of entailment relations. What he wanted to understand was the
underlying entailment patterns. Thus, Rosennean Complexity Theory describes
entailment patterns which, he believed, are responsible for
causality.
We can only observe behaviors or properties of systems if they are
expressed in some way accessible to our perceptive capabilities. So, we can
safely assume that we are observing only a limited set of all possible behaviors
or properties of systems and we can further safely assume that whatever
behaviors or properties being expressed while we are observing are only a
limited set of all the observable behaviors and properties any
given system is capable of. Thus, entailment is the larger category and
causality refers to the observables (the effects).
Mind you, these are all names we give to various aspects of
scientific investigation and therefore are not "real". So it's not a big deal if
you define them differently. The only problem will be in trying to interpret
what my father is describing, if you have your definitions reversed in relation
to his. Entailment, in his work, describes the interactive potential of a
specific relation; A relation is entailed by, and entails, other
relational effects. The pattern of relations and relational effects in the
organization of a system is what makes it capable of behaving the way it
does and expressing the properties it expresses, even when it is not expressing
them. Causality refers to those expressed behaviors and properties. Do you
see? A newborn baby clearly has reproductive organs, for example. So the
entailment for reproduction can be glimpsed, even though reproductive behavior
won't be manifested for many years.
The reason why we can have a modeling relation that commutes
is due to entailment; the logical consequences of relational interaction in
specific situations (i.e. as constrained by other relations). The commutative
property is entailment; Causal entailment in the real system and inferential
entailment in the model. A model having an entailment pattern which commutes
with the real system it models can provide predictions about the behavior
(causality) of the system in interaction with changes in
environmental behavior. The predictions are the way we test our models for
commutativity-- we are testing the entailment of our models-- and this
predictive capacity is also how living organisms are able to exhibit
"anticipatory behavior". That's what he (RR) means when he speaks of
"model-based behavior" with regards to organisms.
Judith
----- Original Message -----
Sent: Monday, February 14, 2005 6:55
PM
Subject: Re: [ROSEN] Causality vs.
Entailment
Judith,
Hmm, after
going back and re-reading your comments on Ayten's post, I think we
fundamentally disagree. You said:
Entailment describes relations, including potential,
in totality. Causality describes what happens. In this way,
causality can be seen as a subset of the total entailment potential of
systems and can be used to try to learn about the entailment relations
within system organizations. It is entailment that Robert Rosen was
most seeking to understand because it would include causality (and
inference, as well-- which describes the version of entailment
expressed in modeling/simulations) within it. Entailment is the more
comprehensive category.
As I see
it, entailment is meaningful only as it occurs in either
inferential or causal structures. If entailment is, as in the OED, "To
bring on by way of necessary consequence", then this presupposes and requires
that entailments can only occur where there exists systems of necessary
relations by which "necessary consequences" can be formed.
Entailment is thus not a more comprehensive category, but
instead more like an organizational property occurring in
a realm with necessary relations. Nor does causality just describe
"what happens". Causality in my view is the entire locus
of necessary (i.e., causal) relations between phenomena in the external
world, whether a particular incident has occurred yet or not. Likewise,
inference is not merely a description of what happens in the formal
world, but is instead the entire locus of necessary relations between
formal entities, whether a particular inferential relation has been calculated
yet or not.
We can establish synonymies between these two modes of
entailment, but I argue that it is not possible to abstract entailments away
from these modes. This would be attempting to abstract the
organization of necessary consequential structures away from the system
of necessary relations (causal or inferential) which makes it work. We can
analyze (e.g., Aristotelian analysis) either of these causal or inferential
entailment structures, but that does not abstract entailment away
from the mode being analyzed:
"Indeed, inferential entailment (between propositions)
and causal entailments (between external events) are the only two modes of
entailment we know about. We can in fact deal with both of them in
exactly parallel Aristotelian terms, by asking why." [EL p.
89]
Where
you say below:
In other words, the
behaviors we observe are "causality" but the entailment relations created by
the organization are what drive causality.
But
either these "entailment relations" are causal entailments or they are
inferential entailments. Otherwise, this posits some third kind of entailment
mode, which conflicts with the quote from Essays above. If it is a material
system, then these must be causal entailments.
Where
you say below:
JR: I would agree with that, but
at the same time, it will always be the case that there are far more
entailments than what are being illustrated causally.
I would say in
reply:
If there are indeed far
more entailments than what are being illustrated causally, this tells us
that our current formulations of causality are highly
inadequate.
This is what I
meant by:
Tim G. wrote: And that if there are perceived
to exist entailments which exceed a particular representation of
causality, then it is the representation(s) of causality which need to be
expanded/revised so that such entailments can be accounted for in causal
terms.
As you say, we
import causal entailments into formal counterparts, to ask "why?" of
them. The formalisms of these formal counterparts are what I called "the
representation(s) of causality" in the above quote. I think we can
study entailments only as they occur in either causal realms or
inferential realms. I do not think that we ever
study (or ever can study) entailments divorced
from these realms.
The reason I focused on
causality (as opposed to inference) is that I agree with Rosen that
processes in the external world is the realm of effective
processes[EL p. 160], and thus, entailments in the causal realm are the
place to look to determine the adequacy (or not) of our current formal
representations of causality, and thus more importantly, the adequacy of our
formal representations of entailments therein and generally.
Regards,
Tim
Tim Gwinn
wrote: I view causality as the means by which entailments occur in the
external world.
I would agree with that, but at the same time, it will always
be the case that there are far more entailments than what are being
illustrated causally. One of the purposes of conducting behavioral
experiments, for example, is to try and trigger causal "episodes" so
that we can try to get a glimpse into the entailment relations
which make such episodes happen (to "cause" the
episodes).
Tim G. wrote:
And that if there are perceived to exist entailments which exceed a
particular representation of causality, then it is the representation(s) of
causality which need to be expanded/revised so that such entailments can be
accounted for in causal terms.
So this seems not quite
accurate. Our models are built with inferential entailments which are
intended to mirror the causal entailments of the real system we are
modeling. In other words, the behaviors we observe are "causality" but the
entailment relations created by the organization are what drive
causality.
I think part of the confusion
may have to do with the different uses of the word "cause". Aristotelian
analysis is based on the four categories of "causation" which can actually
be viewed as entailment, rather than "causality". Inferential entailment is
the entailment which drives the inferences we can make in a modeling
situation and causal entailment is the entailment which drives
causality.
So, to reword (my additions in
brackets) your assessment of Aristotelian analysis:
Tim wrote:
So, for example, Aristotelian
[analysis of] causa[tion] offers temporal
simultaneity of causal [entailment] factors unlike a
linear, chain-like, view of [the sources and flow of]
causality, and Aristotelian [analysis] offers
[explanations for] causal potential in
the category of final cause
[/causation].
I would equate entailment in a system with what you called
"causal potential".
Tim wrote: But
the Aristotelian picture falls short in other ways, and so seems to me to
lead to the opportunity to create yet more comprehensive formalisms of
causality.
I agree. I think my father would also. In fact, I believe his
work is an attempt, on multiple fronts, to begin doing exactly that; with
one little change to your wording... On page 10 of Life,
Itself:
Robert Rosen wrote: "There is, as
yet, no comprehensive investigation of the ideas I have sketched in the
course of the discussion above; they are too new. But it seems that such
ideas, or ideas like them, are necessary in many ways. I would in particular
draw attention to the way such ideas ultimately rest on entailment alone, on
systems of entailment in the material world (causal entailment) and in the
world of formalisms or mathematics (inferential entailment), and on
comparisons or congruences between such entailment systems. I have come to
believe that the concept of entailment provides a reliable anchorage for the
scientific enterprise itself, and I accordingly recommend it to your
attention."
Judith
----- Original Message -----
Sent: Monday, February 14, 2005 9:01
AM
Subject: Re: [ROSEN] Causality vs.
Entailment
Regards,
Tim
Ayten's notion of entailment is accurate in depicting
entailment as sort of a form of "causal potential" although there
is more here than this. All causality is just a tiny fraction of the
total potential entailed by systems and their interactions. Entailment
is also independent of time or directionality, in that
separate things (parts, components, stages, interactivity,
etc) can both entail/be-entailed-by one another in totality,
but not necessarily directly or specifically (like the chicken
and the egg). Causality is dependent on time flow, among other things,
where entailment is not.
Entailment describes relations, including potential, in
totality. Causality describes what happens. In this way,
causality can be seen as a subset of the total entailment potential of
systems and can be used to try to learn about the entailment
relations within system organizations. It is entailment that
Robert Rosen was most seeking to understand because it would include
causality (and inference, as well-- which describes the version of
entailment expressed in modeling/simulations) within it. Entailment is
the more comprehensive category.
It's sort of like a situation where your pet dog
reacts to some new stimulus in ways you've never seen it behave
before. Extreme heat, extreme cold, earthquake, weather,
whatever... The entailment was always present, but this is the
first causal example.
There have been cases of organisms that were thought to be
fully understood, only to have it suddenly become manifest that what
we've seen is only a very long larval stage and the organism undergoes a
sudden metamorphosis to reach some new stage (which we presume is the
adult). The Axolotl, of Mexico, was one such organism. It developed the
capability of breeding while still in its larval stage, a fairly
well-known phenomenon (neotony). In this case, the lakes where this
salamander species lived was at very high altitude. When taken to lower
altitudes, the axolotls began to metamorphose into a fairly ordinary
salamander species.
My father was fascinated by the axolotl because he said it
had far more biological capability in its larval stage, which it lost
when it metamorphosed. For example, in the larval stage it can breathe
air or use its gills in an aquatic environment. It can completely
regenerate an entire limb if cut off. In its natural environment,
the axolotl never actually fulfills its entailed metamorphosis from
larval stage to adult stage, living its entire life in the larval stage.
Thus, entailment and causality diverge in the same
species.
My father actually wrote a fictional short story based on
this phenomenon, entitled "What Really Happened To Jeff" or something
like that, where the main character is a college student whose roommate
was experimenting on axolotls and developed a theory that our current
human form may just be a case of arrested development in
the larval stage. He figured out how to trigger it and disappeared
without a trace with no clues as to what could have happened to
him (except for the dessicated sea squirt found lying on the floor
of the lab...).
Entailment is what is responsible for side effects; the so
called "hidden variables" or unknown relations within a system being
studied.
Judith
|