Tim,
I re-read the references you gave to evolution in Rosen's work. Adaptation
is certainly a consequence of evolution, but I think Rosen is trying to
fit the process of evolution into his concept of anticipatory system,
which it is not. He says that, "a behavior or phenotype which is adaptive
necessarily is of an anticipatory character." Darwinian evolution is a
model-free process, hence no anticipatory action can exist. Also, Rosen's
description of evolution sounds like it works on individual dynamics.
Darwinian theory is about statistical biases only on population distributions.
Tim: as been several comments by listmembers that Don Mikulecky and others
had presented examples of such systems which are closed to efficient
causation but which are not alive, although I cannot find such examples at
the moment (anyone have any links or examples from those days?).
HP: I have always understood Rosen's phrase "closed to efficient
causation" as a formal condition, a closed topological loop in a diagram
of relations including abstract genetic and metabolic components of an
(M,R) system. A less abstract, more material view of this same condition I
have called "semantic closure," or more recently "semiotic closure," to
reflect the essential non-dynamic symbolic descriptive nature of the gene
as contrasted with the material dynamics of metabolism. (Rosen abstracts
away this distinction in his (M,R) systems.) I must note, meo periculo,
that this is essentially an elaboration of von Neumann's requirement for
evolution, i.e., the distinction between description and construction, or
(his analogy) of software and hardware. This is clearly only a necessary
condition and not a sufficient condition for evolution since viable
descriptions are rare in the space of all descriptions. Whether a daughter
cell with an immediately lethal gene is alive is just a matter of
definition, like how long must it last, or how many faulty proteins must
it synthesize to be called alive. For a discussion of semantic closure see:
http://www.ws.binghamton.edu/pattee/sem_clos.html
Tim: 2) I have always been puzzled by - what I consider to be - a deep
conflict between Rosen's proposing of sufficient condition in Q1 and the
"Godelian suggestions" (as Aloisius puts it) evident in Q2 that
"Sufficient conditions are harder; indeed, perhaps there are none".
HP: In my view, this is not a real conflict but a result of Rosen's
non-evolutionary, or what I call "synchronic," view of life. It is in the
nature of relational models that time is abstracted away. Rosen views life
as a kind of timeless order. As he says in LI, he can easily picture life
without evolution. It is also clear that a Darwin-like evolutionary
process is not computable in any Turing sense, first, because it depends
on mutations or errors in the description, and no formal system can
tolerate even one bit of error. Second, evolution never halts. Of course,
many aspects of evolution can be simulated but not in all details.
Also I find that Rosen's concept of error is not the same as the concept
of evolutionary mutation. Rosen's concept of error is limited to a
discrepancy between a system and its model and is therefore entirely an
epistemic problem (See AS Chap. 5.6, p. 307). A mutation changes the real
system itself by constructing a new system. Mutations are not errors in
Rosen's sense but changes in ontological reality. Rosen sees organisms as
only "inequivalent observers" (AL p. 319) instead of "inequivalent
constructors." Again the description/construction distinction and their
coded relation is the crucial condition for evolvable life.
Howard
http://www.ws.binghamton.edu/pattee/
http://www.c3.lanl.gov/~rocha/pattee/