[FRIAM] not enough of Robert Rosen

[Glen E. P. Ropella]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Phil Henshaw on 01/09/2008 08:18 AM:
> The root problem is that equations are *so* self-sufficient that they
> have no environmental interaction at all, and that all natural systems
> are entirely different on that count.   Natural systems actually all
> directly *grow out of* the environments they will continue to interact
> with.  That equations don't do that and robots don't do that, except as
> the man-equation & man-robot couples they are, is the rub.  
> 
> That explaining why seems to be an ill-posed question, then, might it be
> considered highly useful information too, about what information can
> explain v. what we can only point to.

[grin]  Well, the way _I_ parse what you've said, I agree.  The problem
lies in the assumption of duality... the idea that we can _ever_ cleanly
separate inference from causality.

All models are always false.  The only access to truth is through
embedded interaction with the ambient context.

Hence, until/unless we build robots whose mind is directly and naturally
composed from the environment (like our minds are), the robots will be
fragile to ambiguity.

I take some exception to your use of the phrase "grow out of" and the
word "grown".  It conflates too many things.  For example, I think we
could, in principle, remove _time_ from the process and instantaneously
come up with a fully grounded robot mind.  I.e. I think it's logically
(but perhaps physically) possible.  Other examples of implications from
"growth" we might be able to remove are particular types of accumulation
like iteration.  It may not be necessary that each stage be fully
dependent on the stage that immediately precedes it.  I.e. perhaps s_n =
f(s_n-2, s_n-4, ..., s_0) so that only every other stage contributes to
the "grown" thing.

In any case, the term "growth" is too vague.

Also, "growth" may not be necessary at all.  The flaw in the mind/body
duality doesn't lie with how the system came about, necessarily.  The
flaw is that there is no such thing as a complete logical abstraction
layer.  One cannot completely separate thought from reality (and vice
versa), inference from causality, chemistry from physics.

Any "computer" where the software _cannot_ puncture the logical
abstraction layer and modify the hardware will be fragile to ambiguity.

- --
glen e. p. ropella, 971-219-3846, http://tempusdictum.com
If you disclose the solution to the mystery you are simply depriving the
other seekers of an important source of energy. -- Conchis, "The Magus"
by John Fowles

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHjOBSZeB+vOTnLkoRAu+sAKCVV+aZMyS8mYMwXKITnX519D+SJACffMRU
IC2E4ivbuyRz0PVJXsuBEdg=
=TN9N
-----END PGP SIGNATURE-----