[From Bruce Abbott (2000.09.17.1100 EST)]
Bill Powers (2000.09.17.0215 MDT) --
Bruce Abbott (2000.09.16.2225 EST)
Chewing out Domjan isn't quite what I had in mind. I had hoped that we (on
CSGnet) could go over his paper have a balanced discussion of the proposal,
con _and_ pro (yes, I think it offers some ideas worth thinking about with
an open mind).
What we have to do, in that case, is discuss "classical conditioning,"
because that is really what Domjan's article is about.
Yes.
Nothing he says
about control systems is of any interest, because while he asserts that his
models could accomplish control, he doesn't demonstrate that they can. I
don't believe that any of his proposals about control would actually work.
Of course a demonstration would be worth a thousand "don't believe's_"
In the interest of fairness, we need to keep in mind that Domjan et al. use
the term "control" in a wider sense than we do. When folks in Domjan's
field state that some variable "controls" some other variable, they mean
that manipulating the former results in some predictable, systematic change
in the latter, under certain specified conditions. An example of such usage
in ordinary discourse is the statement that the trigger-setting on a
variable-speed drill controls the speed of the drill. Defined in this way,
the term "control" comprises both open-loop and closed-loop systems.
I, too, have serious concerns about the workability of Domjan et al.'s
proposal, but the basic idea seems simple enough: change the input-output
relation of an S-R mechanism until the response to the stimulus is one that
is most beneficial to the organism (according to some criterion). This is
equivalent to clamping a rifle to a stand in a position that aims it toward
a target, firing the rifle, and then adjusting the rifle's aim on the basis
of the difference between target center and bullet hole. After a few
iterations of this, the rifle is accurately aimed and the bullet will hit
dead center every time, without need for further adjustment, so long as
conditions remain constant. Such a scheme might work in a biological system
where disturbances are rarely a significant factor: see steak, squirt so
much saliva into mouth prior to delivery of steak to mouth. Or if you are a
juvenile wolf, see big, bad alpha male, emit submissive gestures. (To avoid
misunderstanding: I would assume that those gestures would be performed by
closed-loop, negative feedback control systems. The S-R mechanism would
produce R by varying the appropriate reference for performing the behaviors
in the presence of S, the alpha male wolf.)
The problem is that Domjan seems to have believed in classical conditioning
through all of his professional life, so to him it is an established
reality not to be questioned as an interpretation of behavior.
I "believe in" classical conditioning too -- as an empirically demonstrated
phenomenon, it is as well established as any scientific observation, as well
establsihed as, for example, the attraction of iron filings to a magnet.
What is in question is not the phenomenon, but its explanation in terms of
mechanism. Domjan et al. have proposed a mechanism, although as we both
know it is rather vaguely stated (e.g., how does this "cost/benefit ratio"
computer work? Magic?).
The idea of
explaining it in terms of a more basic model seems not to have occurred to
him. Is such an explanation what you have in mind, or did you just want to
go over Domjan's modeling effort with respect to control theory?
Well, we've already done some work together on a "more basic model"; we
certainly could try to elaborate such a model further. Tell you what, when
I get a bit more free time I'll go over Domjan's paper line by line and
offer comments where I think appropriate. Perhaps that would be a good way
to begin.
Bruce Gregory (2000.0917.0737)]
I think that anyone who proposes a complex system of dubious workability
owes it to his readers to demonstrate that it not only works, but is
superior to simpler systems that undeniably work.
"Dubious workability" is a judgment for which you have offered no
justification. Opinions are fine, but you really need to explain their
basis in fact or reasoning.
I don't know of any simpler systems that "undeniably work" to account for
all the phenomena of classical conditioning. If you have one to offer, I'm
all ears.
This is my reaction to any
proposal that requires elaborate modeling including "cost-benefit" analysis.
Perhaps it can be made to work, but why bother? The chance that natural
systems are based on elaborate modeling seems to require a dramatic
demonstration rather than a simple unlikely claim.
That's one of my concerns, too, but the "why bother" bothers me. It may not
be well thought out, but at least it provides a starting point for thinking
about what mechanisms might be at work. Do you have a better proposal? Or
do you assume that the problem has already been solved? (Belive me, it hasn't.)
Bruce A.