[From Chris Cherpas (960909.1302 PT)]
_Next step & selection_
The "control of perception" is difficult to learn to control for, partly,
because controlling for a non-circular perception of causation is already so
easy/errorless in the physical/non-biological domain -- even if it's
sometimes hard to justify such a perception (e.g., a la Hume).
When it comes to biological/control systems (e.g., ourselves), we can
indeed perceive the part of the loop that runs through the environment as
just more non-circular causation. But our concept of causation runs into
trouble when we try to explain our behavior this way, as Bill Powers has
discovered and explained extensively.
I suggest there is a similar problem in learning to control for selectionist
causality (do I need to define this here?), and further say that control and
selection are related. Within a control loop, the variations in actions are
not directly "caused by the loop;" but neither are they "uncaused" or
capricious. Is there value in saying that a control loop "selects"
via an internal reference?
In other words, are control loops examples of the "vicarious selectors"
of Donald Campbell's "Evolutionary Epistemology" in the sense that that
they are representations from a lineage of control systems that had to
survive a relatively direct, non-vicarious selection process?
The _sense_ I wish to convey is that there is a relationship between
a selectionist view of causality and a control view, and that people have
trouble learning to control for either for the same reason. However,
the _force_ of what I am saying, if I may invoke that distinction, is not
that we should therefore "reduce control to selection." I think "selection
from within" (control) has a structural/functional coherence that makes the
control perspective more useful for understanding on-going adaptive
behavior than what by now may sound like selectionism-pushed-to-the-limit.
To paraphrase the quote which started this particular subject, control systems
are simpler to understand than "vicarious selection systems."
But if selectionism cannot contribute to our understanding of behavior the
way "controlism" can, it is not because of the deleterious consequences of
controlling for linear cause-effect. It may also be the case that
"selection" allows notions of agency that are "less committed" or specific
than that denoted by "control."
_Uncertainty_
Anybody have a PCT-based interpretation of how we perceive uncertainty?
What can (subjectivist) Bayesians learn from PCT?
_Where's Perception?_
What, exactly, is the referent for the term, "perception?" The input function
which is controlled? Is how PCT talks about perception as an "object?" When
it comes to the "process" of perception, is it the whole loop which is being
named?
Best regards,
Philosophical, yet Pedantic, in the SF Bay