[From Bruce Abbott (951214.1635 EST)]
Bill Powers (951213.0805 MST) --
Bruce Abbott (951213.2020 EST)
I have had no trouble following your analysis and do not disagree
with it, but want to try out a few alternative assumptions and see
where they lead. Hints: (1) There are other reasons why
performance might be rate-limited even if the underlying
relationship is one of positive feedback. (2) satiation has nothing
to do with it. [I keep educating you about this, Bill, but you keep
backsliding. (;->]
Apparently I should not have telegraphed my intentions; you have used the
advance notice to launch a pre-emptive strike designed to nullify virtually
_any_ suggestion I might offer even before I have made it. You say:
(1) Any explanation designed to make up for the fact that a basic theory
predicts incorrectly is ugly. Sometimes it is necessary, but it should
always be a last resort and should immediately lead to a search for
_independent_ experimental confirmation of the added assumptions. Of
course, given that the basic mechanism for reinforcement leads to
positive feedback and runaway, anyone would realize that infinite
behavior never happens.
Nothing to disagree with so far . . .
But to look for a "rate-limiting" process is
simply to say "Even if the basic model predicts incorrectly, I choose to
defend it." If all you want is to save the theory, there are countless
ad-hoc assumptions you can make that will do the trick.
Well Bill, how about if I present a control-system model in which the system
is capable of generating infinite output, and then demonstrate that it
incorrectly predicts the behavior of the real system when a large error is
present. Should you choose to offer a modification to the model that will
make it behave correctly, I will then say that "to look for a
"rate-limiting" process is simply to say "Even if the basic model predicts
incorrectly, I choose to defend it. If all you want is to save the theory,
there are countless ad-hoc assumptions you can make that will do the trick."
Now, if you offer such changes to the model _as I presented it_, that will
only serve to prove you guilty of ad-hoc theory-patching.
It's a wonderful debate trick called "poisoning the well." After the well
has been poisoned, any attempt to drink from it (e.g., offer changes to the
model as _you_ defined it) instantly kills the argument. Well done.
All you have shown, of course, is that the simplest possible version of
reinforcement theory would lead to instability for all values except maximum
output and zero output. This version includes the highly unlikely
assumption that change in behavior rate and reinforcement rate are linearly
coupled. Why should I be forced to accept YOUR model of the reinforcement
process, especially when it has been known since DAY 1 that such a model
would not accord with the data? By offering this ridiculous model and then
asserting that any attempts to change it on my part will convict me of ad
hoc-ery, you poison the well.
Apparently it was not enough to make this claim, you then went on to
anticipate proposals I might offer, and in case the point might be lost on
the reader, clearly labeled each one as an "ad-hoc assumption," invented to
save the theory. So:
(2) I can propose a meaning for satiation and a set of ad-hoc rules
about satiation that will explain why the behavior reaches an asymptote
even though the basic theory says it should run away. When the behavior
produces enough reinforcer that the animal starts to get full, the
reinforcing value begins to decrease and can actually become an aversive
value. If the rate at which this occurs is greater than the rate at
which the exponential runaway tendency grows with the increase in the
contingent consequence, the aversion will overcome the runaway tendency
and a limiting rate will be seen.
Satiation is not even involved in the process we are considering, but just
in case I might want to appeal to it, you've poisoned it. But there's more
. . .
Another ad-hoc assumption would be that the effect of C on dB/dt is
nonlinear, and above a certain level simply fails to produce any more
increments in dB/dt. A third assumption is that the physical limit of
performance is approached, giving rise to fatigue and strain, which
creates aversive effects because of the increasing cost of behaving,
nullifying the reinforcing effects of C. Another assumption is that ...
All these proposals have been given the pejorative "ad hoc" stamp of
disapproval even before they have been offered (if indeed they were to BE
offered). One might even miss the fact that the application of the "ad hoc"
label amounts to an unsupported assertion that there is no independent
evidence for these assumptions. No matter, they have been poisoned by the
accusation; before they can even be offered, anyone who would do so must
fight an uphill battle just to establish that the ad-hoc label is wrong.
The well has been poisoned.
A good liar is always able to find an explanation for why his lie
contradicts the facts.
Now we add more poison to the well. If one attempts to offer an alternative
to the simple model you constructed, he or she is a liar attempting to
explain his lie. The other possibility is not considered. Perhaps the
simple model said to be "the" reinforcement model does not represent any
model ever held by reinforcement theorists. Perhaps the implied assertion
that it does capture the essentials of reinforcement theory is the lie.
This sort of embarrassing string of excuses is best avoided by starting
with a theory that gives the right result to begin with. The theory that
can do that, I am sure, is most likely to be nearest to the correct one,
while the theory that requires all kinds of made-up excuses for its
failures is likely to be wrong. That, I think, is how the wise money
would bet.
Who would argue otherwise? What is in dispute is whether any change to the
simple model of the reinforcement process that you invented would constitute
"an embarrassing excuse." If we're going to play the game this way, then
please give me a few minutes to whip up a simple PCT model that does not
include any limits and, perhaps, doesn't even identify the right controlled
variable. After I have shown that this model contradicts the observations,
I will offer for your amusement a series of what I will term "ad-hoc
patches" to the PCT model that will "save the theory." I will state that
anyone offering such patches is only a good liar trying to explain his lie.
At least then we will again be on a level playing field, with both our wells
poisoned.
From your reply to Sam Saunders:
Very nice answer, Sam. I like your notion of "looking backward in
time" to determine whether every feeder-operation was associated
with a contact closure: it simplifies assessment of the role of
contact-closure per se in the contingency.
Hey, you guys are really complicating this problem. What's the big
mystery about the contingency? You built the apparatus and programmed
it. You can take it apart any time; you can check out the program to see
that it's still behaving as designed. The contingency is just the
operation of a physical device. Nobody said you have to _deduce_ the
rules by observing inputs and outputs. Even if you didn't design the
apparatus yourself, it wouldn't take you long to determine exactly how
it works. Everything you need to know about the contingency is readily
accessible to physical investigation.
I'm with Sam on this one. Your challenge was to demonstrate _empirically_
that the contingency built into the apparatus was really there. I think we
both took this to mean that we should specify an empirical procedure for
proving that the apparatus works as designed.
This is not, unfortunately, true of the organism, but you can't have
everything.
In natural settings, contingencies exist that were not programmed by an
experimenter. The procedures Sam and I outlined work just as well to
identify and characterize these "natural" contingencies as they do for
"artificial" ones. That's important if one is to avoid offering ad-hoc
explanations for behavior when "natural" contingencies may be present, as
they often are, even in the environment of the operant chamber.
···
---------------------------------------------------------------------------
Getting back to the modeling of the reinforcement _process_, is there some
way we can collaborate on this? I thought this was the direction we were
taking, but all I had to do was to suggest that I might have some
alternative ideas to try and you launched a full-scale offensive to stop me
in my tracks. Obviously a very high-gain system defending SOMETHING, but what?
Some problems for the proposed PCT model:
1. The model proposes that reinforcement rate is controlled, and that the
reference level is under normal circumstances within reach of the system.
This needs to be demonstrated. Alternative model: the rat is controlling
for food presence, and against too high a level of effort.
2. It does not account for acquisition and extinction. Even the flawed
bistable reinforcement model at least did this much, without requiring
(guess what) an _ad-hoc_ assumption about the "resetting" of the ref.
level for the CV.
It was my intention to work on those problems, but you evidently think that
it was my intention to demonstrate that reinforcement theory can be patched
up so as to compete with PCT. That would explain the strong response to a
small disturbance.
Regards,
Bruce