Challenge

CSGnet Archive CSG1997
1

[From Bruce Abbott (970724.10:00 EST)]

i.kurtzer (072397)

on the thread of learning Bruce A. cast "function" as a contigent relation
to be experimentally determined and clearly suggested the noted
contingencies would better our understanding of that subject matter. I
feel that the logic of contingency without qualifiers is severely
open-ended, giving all facts equal priority, and even leading to
questionable deductions. For that I gave and still give the example of a
bird's wing having the 1/16 function of entering a snake's mouth--an
example clearly ludicrous but acceptable by a logic without principles.
Instead i ask anyone to suggest a principle that might delineate our
studies from the infinate cataloguing of contingencies, so that some
contingencies are more worthy/significant/truth-revealing than others.
Any takers?

Isaac, the criterion by which any proposed function should be judged is
whether the function can be shown experimentally to contribute to its
owner's survival and reproductive fitness. By this criterion, your proposed
"function" of the bird's wing entering a snake's mouth (and other equally
ridiculous proposals) can be excluded.

Regards,

Bruce

2

[From Bill Powers (970725.0506 MDT)]

Bruce Abbott (970724.10:00 EST)--

Isaac, the criterion by which any proposed function should be judged >is

whether the function can be shown experimentally to contribute to >its
owner's survival and reproductive fitness.

I guess what bothers me about this way of "verifying" the definition of a
function is that it allows you to skip over all the explanatory steps that
I consider essential. It's like explaining goal-seeking behavior in terms
of reinforcement: the organism achieves the goal state because it is
reinforcing do so. This way of talking allows you to "explain" goal-seeking
without having any idea of how goal-seeking works.

In the case of Aplysia, you can show that if you eliminate the ability to
withdraw the gill, the animal becomes vulnerable to injury. Unfortunately,
it is possible to trace out one causal path through which a touch can
elicit this withdrawal under certain conditions, which seems to show that
this simple pathway exists because it has an evolutionary advantage.

However, what is observed is not just gill withdrawal -- it is a rather
complex pattern of habituation that produces gill withdrawal at first, but
then ceases to produce it, sometimes even to the point where the
connections cease to exist. If gill withdrawal promotes survival, then
clearly habituation works directly against survival. A clever predator
would bump against Aplysia repeatedly until it ceased to withdraw the gill,
then feast on the exposed gill at leisure. You could probably show
experimentally that if the animal faced predators while in a condition of
complete habituation, its fitness would be greatly reduced.

Faced with this problem, one then has to explain the function of the
pattern of habituation. Of course this can be done, with a little
imagination: if Aplysia withdrew its gill at _every_ touch, it would be
expending a lot of energy uselessly, and that would clearly (i.e.,
logically) work against survival. Obviously, there are two "competing"
effects on survival here, one acting against the other. So whether Aplysia
habituates or withdraws its gill, the overall function of this arrangement
is still survival. "Fitness" now explains both withdrawal and failure to
withdraw. Even the autoasphyxiation effect of gill withdrawal can help
explain habituation; the habituation prevents Aplysia from suffocating
itself, and that, too, is good for fitness.

Then there is the problem that a touch elsewhere can suddenly restore the
habituated pattern to operation. Once again, however, one can say that the
function of this restorative reflex is to promote survival by rescuing the
animal from the adverse consequences of habituation. And when it turns out
that the basic reflex can be turned on and off by other neural connections,
once again functional analysis triumphs, by showing that the other
connections also promote survival, by cleverly allowing the reflex to occur
only when doing so is the most advantageous thing to do, survival-wise, and
otherwise allowing the musculature and nerves to carry out other
survival-enhancing functions.

By the time you have carried this functional analysis to its ultimate form,
you will have explained every detail of Aplysia's behavior, and will have
shown why every connection is just as it is, without understanding anything
at all about how Aplysia works. By explaining _everything_ in terms of a
balance among opposing effects on fitness to reproduce, you will have
explained nothing at all.

I think that behaviorists have fastened onto functional analysis because it
appears to offer an explanatory principle to fill in the void left by a
purely descriptive approach to behavior. What a wonderful gift! Without
having to study neurology or systems analysis or biochemistry or control
theory, and indeed without having to know very much at all, one can cut
through all the fussy details to the ultimate explanation and leave all
those plodding, obsessive-compulsive nit-pickers in the dust! And of course
all this is achieved without having to give up the underlying faith that
the environment is the ultimate determinant of all behavior, a faith that
is so self-evidently correct that it needs (and gets) no defense.

Best,

Bill P.

3

[From Bruce Gregory (970725.1010 EDT)]

Bill Powers (970725.0506 MDT)

By the time you have carried this functional analysis to its ultimate form,
you will have explained every detail of Aplysia's behavior, and will have
shown why every connection is just as it is, without understanding anything
at all about how Aplysia works. By explaining _everything_ in terms of a
balance among opposing effects on fitness to reproduce, you will have
explained nothing at all.

Hear, hear! There seems to be a new cottage industry carrying
out these "Darwinian analyses". I think your analysis of their
appeal is also right on target.

Bruce

4

isaac kurtzer (970727)

rehash:

I
feel that the logic of contingency without qualifiers is severely
open-ended, giving all facts equal priority, and even leading to
questionable deductions.
Instead i ask anyone to suggest a principle that might delineate >>our
studies from the infinate cataloguing of contingencies, so that >>some
contingencies are more worthy/significant/truth-revealing than >>others.

[From Bruce Abbott (970724.10:00 EST)]

Isaac, the criterion by which any proposed function should be >judged is
whether the function can be shown experimentally to contribute >to its
owner's survival and reproductive fitness. By this criterion, >your

proposed

"function" of the bird's wing entering a snake's mouth (and other
equally
ridiculous proposals) can be excluded.

Your evolutionary criteria would not exclude my admittedly ridiculous
example. There are many concievable scenarios which the bird's
reproductive success increase by its wing being snake-chow, like by
allowing it littl'uns to sucessfully flee and not get gobbled--one me for
six quasi-me's. This situational specificity of your criteria is more of
a convenient phantom that cannot delienate any class as a function, but
only self-inconsistent particulars. This suggests it is fairly useless.

I am still waiting for any offers for a principle of delineation.
Please someone offer one, as we might learn something and you might win my
beer-prize, Zima excluded.

i.
i.

5

[From Bruce Abbott (970727.1035 EST)]

isaac kurtzer (970727)

[From Bruce Abbott (970724.10:00 EST)]

Isaac, the criterion by which any proposed function should be >judged is
whether the function can be shown experimentally to contribute >to its
owner's survival and reproductive fitness. By this criterion, >your

proposed

"function" of the bird's wing entering a snake's mouth (and other
equally
ridiculous proposals) can be excluded.

Your evolutionary criteria would not exclude my admittedly ridiculous
example. There are many concievable scenarios which the bird's
reproductive success increase by its wing being snake-chow, like by
allowing it littl'uns to sucessfully flee and not get gobbled--one me for
six quasi-me's. This situational specificity of your criteria is more of
a convenient phantom that cannot delienate any class as a function, but
only self-inconsistent particulars. This suggests it is fairly useless.

I am still waiting for any offers for a principle of delineation.
Please someone offer one, as we might learn something and you might win my
beer-prize, Zima excluded.

I would like to have heard your reply _before_ Bill Powers jumped in and
showed you the politically correct answer. (And the misplaced carots in my
reply as quoted show that you copied it from Bill's post and not from mine.)

However, from the tone of the challenge itself, it was clear that your mind
was already made up, and since you get to decide who "wins" your prize,
there was never any chance I was going to do so regardless of what answer I
provided.

If the bird could be shown to facilitate the survival of its own genes
through the sacrifice of its wing to a predator, it would by my criterion
qualify as one function of the wing in the life of the bird. Grasshoppers
do something similar; when caught by the leg they constrict a special set of
muscles that severs the leg from the body, allowing the grasshopper to
excape while the predator is detained feasting on the abandoned leg. But of
course, the function of this mechanism must remain an eternal mystery to
PCTers like you, although any ten year old can see that it helps the
grasshopper to preserve its life.

Regards,

Bruce

6

i.kurtzer (970728)

bruce, your ad hominems are irrevevant for me and i assmue those sincerely
interested in the discussion.--determining the value of principles to
delieate significant behavioral contingencies.
Please, anyone here can offer some principle and win my prize,
but only if you suggest a good one.
I will again reiterate:
Since there are concievable and actual instances of bird's wings entering
a snake's mouth is this a significant contingency for the life sciences to
note and proffer explanations for? Since, the functional principle can
only delineate this situation according to each situation--sometimes yes,
sometimes no--I feel this would be a poor principle as it is tailor fit
afterwards and, therefore, possesses no heuristic force.

For example

Bruce Abbott (970727.1035)

If the bird could be shown to facilitate the survival of its own genes
through the sacrifice of its wing to a predator, it would by my criterion
qualify as one function of the wing in the life of the bird.

So it is a fraction of the function of the wing after all?!
I specifically brought up this as ridiculous, AND IT WAS AGREED AS AND
EVEN RIDICULED AS RIDICULOUS.
Now there are appeals to examples of this across species!
Come on what type of priciple is that?

I now up the ante to two foreign (non-American) or a six pack of Shiner
Bock!! I'm straight up serious, just offer a heuristic principle.

i.

7

[From Bruce Abbott (970729.0915 EST)]

i.kurtzer (970728)

bruce, your ad hominems are irrevevant for me and i assmue those sincerely
interested in the discussion.

Isaac, an ad hominem argument says that your opponent shouldn't be believed,
because of who he or she is. For example, I might have "argued" against
your position by saying that the reader should reject it because you are
"only a student" or "incapable of rational thought."

This is of course an invalid and low form of argument, and not one I have used.

Bruce Abbott (970727.1035)

If the bird could be shown to facilitate the survival of its own genes
through the sacrifice of its wing to a predator, it would by my criterion
qualify as one function of the wing in the life of the bird.

So it is a fraction of the function of the wing after all?!
I specifically brought up this as ridiculous, AND IT WAS AGREED AS AND
EVEN RIDICULED AS RIDICULOUS.

You overlook the word "if." You have not shown that any bird actually uses
its wing in this way, and the loss of a wing would surely be quickly
followed by the loss of the bird. As young birds require a lot of
parenting, this would almost surely result in the deaths of the offspring as
well. In addition, nothing was said in your original scenario about the
wing being offered as a decoy to distract a preditor from the young. The
offering of a wing to a predator in the absence of such of a contingency
would have no advantages for reproductive success (on the contrary, it would
have extremely serious negative consequences). Furthermore, you stated that
the function of the wing would be to fit into the mouth of a snake, if I
recall correctly. That is the image I judged to be ridiculous. That it
might have the function of distracting the snake from an attack on the
bird's young is plausible, although that hypothesis needs to be tested (thus
the "if").

Look, Issac, it is plain that you have no interest in considering my
position; this whole "debate" has been a farce. You are only looking for
reasons to reject. You can keep offering your empty "challenge," but it's
pointless to continue arguing with you.

Regards,

Bruce

8

i.kurtzer (970730)

i again emphasize that this forum is open to anyone, particularly if one
offers some principle other than "functional analysis" .
Again this is a staightforward query: what principle is there to delineate
behavioral contingencies as significant? Functional analysis fails as it
is loosy goosy enough to allow the most ridiculous examples, is
capriciously convenient after the fact, invites major philosophical
problems--such as how to fractionalize functionality, and is really no
principle at all but a awkward redescription cast as an explanation. Come
on, win my beer..hell i'll buy you a whole 12 of our Texas sipp'n tea if i
see you in Durango. It's really not hard.

i.

9

[From Rick Marken (950917.1250)]

Bill Powers (950915.2100 MDT) --

If you run the program with k = 1.0, you will get an RMS error of 6 or
8, which is very small. Setting k = 0.3, the RMS error becomes about
600.

The performance of my implementation of Hans' model did not deteriorate
with k<1 becuase I did not include k in the model's computation of
the disturbance estimate. That is, I still used dnew := x - u to estimate
the current disturbance effect (dnew) rather than dnew := x-k*u. I did
this under the assumption that the system could not possibly know k. But,
of course, Hans' system is a model based controller so we can assume that
there is an algorihm that can estimate (model) k perfectly in real time.
Your result suggests that such a model actually would make things worse
for the controller.

I realized today that the part of Hans' control algorithm that makes
his model perform better than the standard control model is the following:

dpre := 2.0 * dnew - dold

dpre is a better estimate of the next value of the disturbance than
dnew alone. The stored value of dold contributes to the increased accuracy
of the prediction. I had a feeling that the reason this was true was because
the disturbances we are using have nice, smooth changes over time. I tested
this hypothesis by running Hans' model against the control model when
the disturbance was a square wave rather than a relatively narrow band
random disturbance. When I did this, the control model did much better
than Hans' model. Indeed, the rms error for the control model (with
square wave disturbance) is about half that for Hans' model.

So the results of the challenge are disturbance dependent; Hans
model based controller does better than the PCT model with smooth
disturbancews; the PCT model does better than Hans' model with
square wave disturbances.

The difference in the way the PCT and Hans' model delt with square wave
disturbances suggested a way to test the two models; just compare
the performance of the two models with that of a human working against a
square wave disturbance (NB. Jeff Vancouver: here is a way to try to
disconfirm PCT. PCT makes a very different prediction than Hans model-
based control model regarding the behavior of humans trying to control
against square wave disturbances).

I did a quick test, visually comparing the graphs of the behavior of
a human against those of the PCT and Hans'model based control model. The
human curves were nearly identical to the PCT curves and nothing at all like
the model-based control curves. This was true even when I put an environmental
integration between the output and the input of the two models (to mimic the
environmental integration between the force applied to the mouse and the
resulting position of the mouse). Even then, the behavior of Hans' model
based controller was nothing like human behavior.

I think further comparisions of the behavior of Hans' model based controller
to human behavior would be very useful. I think we can learn more about Hans'
model this way -- and also learn more about where model based control might
fit into our understanding of human behavior.

Best

Rick