Cliff wrote:
>This I'm not clear on. The first sentence is very different from the
>second. Yes, the same formal system can be implemented (I like the
>term "manifested", connoting "made actual") by/in different physical
>systems, but the first sentence implies that a formal system can exist
>independently of ANY physical manifestation. This smacks of idealism.
>
>On the other hand, the fact that these two physical systems share
>something in common implies that there IS something they share in
>common, and that something is a priori not related to their physical
>aspect.
>
>This problem is true of all "relational" or "informational" type
>categories: it's a truism that a car isn't just a pile of parts, but a
>collection of parts in a certain special RELATION. What is the nature
>of this relation, if not ideal?
No, this relation is something we build ourselves. This differs from
idealism in that there is not an ontological domain required in which
'relation' is to exist. I would consider it in a nominalist way: 'relation'
is something an observer may assign to two or more objects. The fact that
we assign it is not enough ground for it to (really) exist (which would in
fact be the idealist's position).
For instance, if we recognize a certain pattern after having seen two cars,
we may say that the parts of the cars have a certain relation; or we may
say that there (really) is some relation between those parts. But this is
misleading language, I think. For these relations 'are' not there; we
impose them; and for good reasons at that! But, good reasons or not, the
relations are our own constructs. This holds for cars as well as for
mathematical relations, and also for symbolic operations.
Jeff also wrote:
>> 3) And there can be systems that have spects of both (matter-symbol
>> complementarity). It is this third type that Patee says open-ended
>> evolution can occur (and life).
Similarly, as to this third domain, a fully nominalist approach (as in 2)
does not satisfy, nor does a fully realist approach.
We cannot:
a) treat all observed phenomena as causally determined processes (realism)
nor
b) treat all assigned relations as observer's constructs (nominalism)
because:
ad a)
some processes seem to be of a symbol processing quality (e.g. dna
production)
and
ad b)
some relations seem to be read or constructed by the observed system itself
(e.g. the 'reading' of dna by a living system itself)
To maintain that 'laws' (causal processes) and 'constraints' (symbolic
processes) are complementary domains, as Pattee does, or to distinguish
causal from formal entailments, as does Rosen, is,I think, only half of the
story. The other half, I think, is _that these two cannot always be
distinguished_. This is how I interpret Rosen's cyclic models of the type:
f phi b
a---->b b-----> f f ----> phi
That is: there are double roles, and they are mixing up. (By the way: this
is also what I found interesting in Catharina's work, as far as I
understood it). But double roles are only a _mode of expressing_ our
inability to fully understand what happens. It is this mixing up, our no
longer distinguishing the two as separate events, that is both a limit to
our understanding and a feature of the phenomena studied!
(For me as a psychologist this is very important. I think that the _limits
of what we can understand_ with respect to autonomous processes and living
experience should not only be understood as a nasty inconvenience for doing
proper science, but rather as a source for better understanding the
phenomena themselves. But here another story would start...)
Well, enough so far.
Arno Goudsmit