[from Tracy B. Harms (2009-07-22 17:17 Pacific)]
Today I've been in conversation, via Twitter, with someone I know only
as "fustbariclation" over behavior, and esp. how to construe what is
accomplished at the neural level. It began with two "tweets" by him:
F: Logic is fundamental to our, & other animals, operation, but ¬ to
our conscious thinking, emotion is actually our driver. It helps to
know.
F: Logic is a useful moderator to our more instinctive operation a
wise person uses it as a tool, ¬master, but a valuable, servant.
My initial comment was, alas, rather opaque:
TH: You wrote "Logic is fundamental to our, & other animals,
operation". Can't imagine how this could be so.
He replied:
F: try catching a ball or chasing down a gazelle without logic.
F: Deciding to chase a gazelle without an implicit understanding of,
the law of the excluded middle (is it a gazelle ∨¬) leaves a hungry
lion.
My response to that was:
TH: Balls are caught without applying logic, like they're caught w/o calculus.
TH: It's possible to apply logic to ball-catching or gazelle-chasing,
but that's different.
TH: For a logic-free model of behavior, see Perceptual Control Theory.
http://www.perceptualcontroltheory.org/
Fustbariclation came back with this:
F: I'm aware of that. Perceptual Control Theory, to me, doesn't
exclude logic, just shows it embedded, which is my point, rather.
F: we're talking about logic in different senses. Clearly there's no
symbolic logic in animal behaviour, but it's implicitly applied
F: yes, balls are caught, as I've been saying, w/o conscious calculus,
but the brain performs calculus to do it.
F: brains, solve problems using logic & calculus as embedded
processes. We don't have conscious access to these mechanisms.
F: if we could access our own internal circuitry, we could think like
Mr Spock & calculate like savants, but, mainly, we don't.
TH: I recognize that they're mainstream, but I reject claims that the
brain performs calculus to coordinate movement.
TH: If true, that *would* imply that "if we could access our own
internal circuitry" we'd have computer-like calc. abilities.
F: yes, we would. I think that 'Idiot Savants' establish that this is
actually, surprisingly, the case.
TH: The processes don't, I'm confident, involve that manner of
calculation. Instead, we find error-dampening loops.
F: yeah, call them error dampening loops. These loops are cunning
methods of solving differential equations. prove me wrong.
F: what do you mean by 'performs calculus'? You can solve differential
equations with analogue computers - like brains.
F: now here's the real question: If the brain doesn't coordinate
movements through a calculus, how does it do it?
F: 'that manner of calculation'. No, of course ¬, no maths symbols
used for a start. But an isomorphic calculation by def. ≡ results
TH: Idea that nerve system calculates numeric results is akin to
thinking genes directly encode behavior.
F: analogue systems do calculate numerical results. They don't produce
spreadsheets, certainly, but that's not the correct measure.
F: maybe you should confess you're real objection. Robot & tiger catch
a ball; do you argue one magic, one calculation?
As things stand the dialog seems at risk of breaking down. CSGnet folk
will correctly anticipate that I don't see a different process for
("closed-loop") robots and tigers.I do recognize that what goes on
within an active control system is calculation. I thought (and still
think) he was asserting that what goes on in behavior is the
calculation of modeling, and that if we had some sort of privileged
perspective into the neural networks we'd be able to find the
correlates of values that represent qualities of something in the
environment *other than* perception of the controlled aspects of the
environment.
Perhaps I've dug myself into a hole and have no better course than to
accept that values are computed, implicitly and/or structurally, via
control systems. But, instead, I hope I can bring out my concern about
the flawed idea I think I see here: that when things are accomplished
it is by way of calculating values other than those indicated for PCT
components.
As I'm unsure how to try to get this point across, I'm interested in
any advice or criticism that might be offered here. Thanks!
Tracy