[From Rick Marken (940825.1440)]
Avery Andrews (940826.1528) --
Nice post to comp.ai.phil.
I do have a couple comments, though I'm not relaying them to comp.ai.phil
(might interfere with there discussion of "free will in electrons" or
whatever they're into now) . You say:
One important issue for the Powers viewpoint is whether it is possible
for organisms to stabilize an ESV [ecologically significant variable]
without perceiving and 'controlling' (in his sense) a correlated CEV
[complex environmental variable].
Earlier in your post you define a CEV as follows: "Each perceptual signal is
associated with an indefinitely large collection of `complex environmental
variables'" and you define an ESV as "features of the situation which we in
our presumed wisdom take to be critical for the survival or economic
viability of the system in question."
This definition of ESV seems to leave open the possibility that such
variables may NOT actually be crucial for survival (hence, they are not what
we call "intrinsic" variables; if an intrisic variable is not kept under
control the organism dies). Your definition makes me wonder why anyone would
care whether organisms can control ESVs or not. It also makes me wonder what
you mean by the following:
If ESV's are often stabilized in some manner without the perception
(veridical indication) of approximating CEV's, then Powers is basically
wrong.
If ANY variable is controlled (not stabilized, _controlled_) without
perception then PCT is completely and utterly wrong. If a variable is
stabilized without perception then there is no problem for Powers or PCT;
stability and control are two completely different (though superficially
similar) phenomena. If you meant "controlled" rather than "stabilized" then
why single out ESVs as a problem if they are controlled without being
perceived? It would be just as big of a problem if a CEV could be controlled
without being perceived. If, on the other hand, you actually meant
"stabilized" rather than "controlled", then why would this be a problem
at all, whether it's an ESV or a CEV that is stabilized. For example, when
I stop a swinging pendulum I am controlling the position of the bob; if I
stop the bob (by clutching it) at the bottom of its swing and then release
it (and stop looking at it, so I no longer perceive it), the position of the
bob will remain stable (as a result of my controlling) but it is no longer
controlled (because I'm not holding it any longer; someone could now come by
and apply a force disturbance, resulting in a change in the position of the
bob; if I were still holding the bob and controlling it's position such a
distubance would have little effect on the position of the bob).
Here's a better example, perhaps (still using the pendulum but now its on a
rubber band) becuase it is more consistent with your claim about ESV's: When
I control the bob, keeping it at ANY position, I am also _stabilizing_ the
rubber band, keeping it straight. If the rubber band is covered (so that I
cannot perceive it) I am stabilizing the rubber band without a perception
of an approximating CEV (the bob). But I am not controlling the straightness
(stable position) of the rubber band. If someone pulls on the middle of the
rubber band, I will resist the disturbance only to the extent that it affects
the controlled position of the bob; the "bend" remains in the rubber band
while the "pull" disturbance is in force. So the straightness of the rubber
band is stabilized (but not controlled) without being perceived. No problem
for PCT that I can see. Did you mean something different?
Martin Taylor (940824 18:20) --
I believe the environmental coherence argument to be correct in principle,
as I have acknowledged.
It permits passive statistical analysis of the scene, and allows efficient
use of the channel capacity of the optic nerve.
Have you seen any of this "passive statistical analysis" taking place? Could
you describe what kind of statistical analysis is being done and explain how
we could all observe it too?
These are the transform templates I discussed in the Pattern and Transform
working paper. Each one potentially represents a degree of freedom, if
they are orthogonal in respect of the sensory input. Their outputs
may well not be orthogonal, if the changing environment is (as we both
argue) coherent over space and time. But that is NOT the issue.
The issue at hand is in the selection of which degrees of perceptual
freedom to control at any moment.
Not quite. The issue is whether there are substantially more input degrees of
freedom than output degrees of freedom -- an issue which, in your long post,
you address only at the end:
But is it likely that the number of lowest-level PIFs is so small? [How
small? RM] That's a matter of opinion, until we know some way of
identifying individual PIFs >within the nervous system.
Yep. And at the moment, in my opinion (and, apparently in Bill Powers'
opinion too) the number of input df that can be controlled simulataneously
is no greater than the number of output df available to control them.
Is there any behavioral evidence that there is a degrees of freedom problem
for human control systems? What is the evidence?
Best
Rick