[From Bill Powers (2010.02.07.0340 MST)]
With my brain still reorganizing after the last encounter with Bruce G.,
I’m reminded of similar, though less upsetting, encounters with Bruce A.,
David G., Martin T., and others over many years and up to the present.
The same question keeps coming up: are we going to have a revolution or
aren’t we?
PCT is the present state of a process that began with learning about,
accepting, and starting to work out the implications of a revolutionary
scientic concept, the idea of negative feedback control. The revolution
started in the mind of H. S. Black on the morning of August 27th, 1927 as
he was on the Lackawanna Ferry going to work at the Bell Labs, and spread
rapidly over the next 20 years. When it started to leak out of the
engineering world into the life sciences, however, it ran into
resistance. The resistance arose because all of the life sciences with
only a few minor exceptions had been developing for many decades in total
ignorance of this new concept, and had created a huge network of
concepts, terminology, and classifications based on other – and
completely spurious – ideas of what makes behavior work. So not only did
the revolution have to spread into the life sciences, it had to displace
the ideas that were already there. And that aroused fierce
defenses.
Arthur C. Clark gave us Clark’s Theorem: the products of any highly
advanced civilization will appear to us to work by magic. To this I want
to add Powers’ Corrolary: to the inhabitants of any sufficiently retarded
civilization, everything will appear to work by magic.
Civilizations begin in ignorance and strive toward knowledge; they move
from magic to science,
Magic is causation without mechanism. The mere fact that event A occurred
is enough to cause event B to occur, with no intermediate processes to
explain how A was transformed into B. The mere wave of a wand at Hogwarts
causes someone on the other side of the quadrangle to fall flat on his
back. What science does is to provide connections from A to B in the form
of smaller magics. These smaller magics are called mechanisms, and while
they still involve causation without mechanism, they also provide
stepping stones from A to B that are useful and in fact are the source of
immense increases in understanding. Having seen these new mechanisms, we
can now see how combining them differently can lead not only from A to B,
from A to C, D, E, and so on.
The structure of the behavioral sciences has been mostly magical, which
is to say, empirical. I attended a seminar at the University of Colorado
at Boulder, or CU, last Tuesday, in which the subject was “Aiming
Accuracy: What have we learned since Woodworth (1899)?” What we have
learned, apparently, is how moving one hand or two hands to a target or
to a target and back again, slowly or rapidly, with the same or different
distances to the target, alternating hands or repeating with one hand,
with pauses between the trials or no pauses, and with spaced or
continuous learning sessions, affects the accuracy of pointing. Some
conjectures were offered about what the subjects were thinking by way of
strategy, but nothing organized or systematic. So this was pure magic:
these changes of conditions affected accuracy just because they existed,
not because of any intervening processes. Afterward I said to the
presenter, “This is good old-fashioned experimental psychology,
isn’t it?” He was quite pleased that I put it that way. He would
not, I presume, have been so pleased if I had said he was studying magic.
But he was. All science begins with studying magic and formulating
beliefs. But after 111 years of studying, you’d think it would have gone
a little way toward knowledge, wouldn’t you?
Fortunately, I was told later that the presenter was quite excited at the
idea that I might provide a model of pointing behavior to use in his
work. Maybe he can start getting some science into his empiricism. My
estimate of his abilities went up when I heard that. And of course
whatever he decides, one has to admire his skill, persistence, and
patience to have spent 20 years meticulously studying pointing
behavior.
So the question is, are we going to have a revolution or not? I think
there is only one way to do that. Scrap everything and start over. If you
don’t go all the way, if you aren’t willing to give up everything you
think you know about behavior, it will simply be too hard to make the
transition. You won’t be free to explore any part of the new approach any
way you please; you’ll always have to be careful not to upset any of your
favorite applecarts. That will inhibit your thinking and generate blind
spots, like continuing to believe that the way to create repeatable
results is to create repeatable behaviors.
Maybe – in fact quite likely, though we shouldn’t start out by thinking
this way – we may discover some things about behaving organisms that the
old-time psychologists also discovered, even though they had the wrong
explanations for them. Even after Lavoisier put an end to 150 years of
phlogiston, it remained true that if you put mice into dephlogisticated
air, they will die. Only now we know that there never was any such thing
as phlogiston; the oxygen had merely combined with carbon and become
unbreathable. Lavoisier had the role of H. S. Black, and the result of
his finding the role of oxygen in combustion was a scientific revolution
that ended up replacing alchemy with chemistry. So PCT is the start of a
revolution that will replace psychology and many other allied disciplines
with something entirely new. As Kuhn observed, the new science will not
be built on the old science; it will replace the old science.
In LSC3, chapter three, a “Live Block Diagram” is discussed;
the program comes with the book and can run on a Windows-based PC or an
Intel-based Mac with a suitable virtual-machine program in it. In this
diagram you will find all the basic features of the revolutionary idea
behind PCT. You will see that despite time-delays in the control loop,
the loop gain is high and the control is highly accurate, and the
control system is not unstable as so many behavioral scientists seem
to believe it must be. The time-constant of the output function, out of
the box, is 30 seconds (that is, after a step-change in the error signal
to a new constant value, the output will take 30 seconds to change 2/3 of
the way to its final new value, 30 more seconds to change 2/3 of the
remaining way, and so on). Despite that very sluggish response, the time
constant of the overall control process is 0.3 seconds. The gain of the
output function is 100: that is, the output is 100 times the magnitude of
the error signal, after it comes to equilibrium. Reducing the output gain
to 50 – cutting it in half – reduces the output by 2%.
In other words, a negative feedback control system doesn’t behave in
accord with ordinary causal logic or common sense. Our common sense has
been trained to fit a different model, the cause-effect model that
underlies all conventional theories of behavior. If you want to be part
of the PCT revolution, you have to retrain your common sense, which is
exactly why you must simply give up every previous thing you learned
about behavior that was based on the old common sense – that is, you
have to give it all up. It is entirely wrong at its foundations.
Study the Live Block Diagram. Experiment with it any way you can think
of, until it begins to make sense to you, until it starts to be part of
your common sense about behavior, about control systems, about organisms.
Behind it is a running model of a real control system, the same model
that’s used in Chapter 4 to match your own behavior in a real tracking
experiment. There’s nothing hypothetical about it any more; it really
fits actual human behavior very closely. The more sense this block
diagram makes to you, the less sense any other psychological theory will
make. Do that enough and you will become part of the revolution whether
you like it or not. You can’t un-understand PCT once you have understood
it.
Best,
Bill P.