[From Rick Marken (960818.1400)]
Me:
The accuracy of inertial guidance systems has nothing to do with it. All
you need is a circuit that does what the pigeon does: controls a
representation of the 2-D position of the ship. The input to the circuit is
an electronic representation of this variable; we know that this input is
available because it is being converted into a visual display for the pigeon.
Martin Taylor (960816 17:00) --
You are in the wrong business, Rick. In two sentences you solved a
problem that bedevilled computer science researchers for five decades.
Marvellous. I am humbled.
It's easy once you learn to look at control from a PCT perspective;-)
However, under the assumption that you may have waxing facet1ous, I will
point out that I was responding to the notion that the problem being
solved with the bird in the bomb was the design of a "navigation system".
Bruce Abbott had suggested that Skinner was capitalizing on the pigeon's
navigational capabilities when he used the pigeon to guide a bomb to a
target. I think of "navigational capabilites" as the ability to determine
one's current position, select a course to a target position and "steer
the course" towards that target position. So from my point of view the
pigeon in the bomb was not navigating -- or using any special navigational
capabilities.
What I thought the pigeon was doing, at first, was pecking to control
a two dimensional representation of the position of the bomb (or target).
No navigation would be involved here; just the kind of control that
Skinner was studying (and not noticing) all along. This would have meant
that the pigeon could have been easily replaced by a simple control system
(just as a human can be easily replaced by a simple control system in
our compensatory tracking tasks). It turned out, however, that my concept
of what the bird was doing was wrong; it was not pecking to control an
image of the bomb (or target); it was pecking to control a relationship
between its beak and the target. The consequence of these pecks was used as
a representation of the position of the target in an electro/mechanical
control system. This perceptual representaton of target position was the
variable controlled by the control system.
So the problem that was solved by the pigeon was not a "navigational"
problem; nor was it (as I had thought) a far simpler control problem. The
problem solved by the pigeon was a _perceptual_ problem; the problem of
representing the controlled variable (target position) to the control
system. This is the same problem that exists for "smart" (I'd prefer
to call them "incompetent" in honor of Asimov's inspiring observation)
bombs today; the problem is not navigational; it's perceptual. The bomb
has to perceive the location of what should constitute the "target" in
the input sensory field. If the bomb could reliably perceive target
location, the solution to the navigation problem would be a simple
control system.
Hans Blom (960729c) --
What I _do_ know is that MCT "explains" (has similarities to)
real world phenomena that PCT cannot model.
Martin Taylor (960816 15:10) --
I'm sorry, Hans, but do you have any examples?
As you can see, Martin, it has been some time since Hans made this
claim. He even proposed an experiment to test the relative merits of
MCT and PCT . I'm ready to run it as soon as Hans approves the method.
I'm sure Hans is just taking his time because he knows how devastating
it will be for me to find out that MCT is better than PCT;-)
Phil Runkel (16 Aug 96) --
I am confident that the next great breakthrough will be, in effect, the
discovery of Powers's 1973 book
Well put, Phil. Like the "breakthrough" of Mendelian genetics, the
breakthrough of Powersean behavioristics will be a case of "Back
to the Future".
Best
Rick