Costs; truth by vote

[From Bill Powers (951221.1315 MST)]

Martin Taylor 951221 00:45 --

     But I tempered that by saying essentially what you said above, that
     the exact form of the function didn't matter, so long as it had a
     sufficient increasing derivative, and that everyday experience with
     resource limitations suggest that our _perceived_ cost functions do
     have that characteristic.

The _perceived_ cost function doesn't enter unless the organism is
controlling for costs. This is why I said your proposal is excessively
cognitive for a rat model. All that is really required is that the
effectiveness of reinforcers fall off with increasing behavior rate (or
reinforcement rate). There is no need for any perception of the cost,
and indeed if you propose such a perception then you have to propose an
auxiliary system for perceiving it, comparing it with the desired cost,
and acting on the food-getting system as appropriate. This takes us
pretty far from the original subject, which was the addition of ad-hoc
nonlinearities to make the reinforcement model match what is observed.

···

-----------------------------------------------------------------------
Kent McClelland (951220.1500 CST) --

     In terms of the argument I was making, something could be a fact
     whether you agree or not, as long as the preponderance of those who
     are socially designated as competent to make the judgment agree
     that it is. ... In saying something is a "collectively controlled
     perception," I am emphatically not arguing that the complete
     agreement of everyone involved is necessary for the perception to
     be controlled, so your comment that it's "self-negating" misses the
     mark.

I think you missed my point. I was discussing what you propose as a
sociological fact, namely:

     A scientific fact is deemed true if competent judges agree that it
     is true.

Is the indented statement above a fact? It is a fact if a preponderance
of those who are socially designated as competent to make the judgment
agree that it is true, but otherwise it is not a fact. So all it would
take would be for a preponderance of competent judges to reject the
indented statement, and it would no longer be true that the statement is
a fact if enough competent judges say it is true. And from then on it
would not be possible to make the statement true again, because once the
statement is rejected, it cannot be made true simply by a sufficient
number of competent judges saying it is true. Once denied, the statement
is false forever.

One of the great battles of science was the battle against argument by
authority. "All the best (most revered, holiest, most ancient, most
intelligent, most powerful) people agree that X is true" does not make X
true. All the best people can be completely wrong if X is a statement
about nature (rather than a matter of definition). The procedures of
science are supposed to begin with low-level observations (as you
pointed out) on which there is essentially no disagreement. The theories
of science are then related to those observations BY METHODS OF
REASONING THAT ARE PUBLIC. Most generally, this means by mathematical
methods. No number of authorities can make a mathematically-incorrect
derivation correct.

The invention of new theories is not the question here; theories are
generated by a creative act which is far from public. But the methods of
_testing_ a theory, of deriving the implications of the theory and
making predictions from it, are (or are supposed to be) public. What
this means is that anyone given a statement of a theory should be able
to arrive at exactly the same predictions from it as anyone else who
knows the same method of reasoning. And these predictions, on which the
truth of the theory rests, are to be judged by comparison with low-level
observations, not merely by agreement that they are correct.

Any suggestion, therefore, that a scientific fact can be based on mere
agreement -- "truth by vote" -- is contrary to this view of science. I
can agree with you that there are social organizations in which the
people do judge truth by a preponderance of authoritative statements,
and that these people often claim that they are doing science. But my
definition of science, which is shared by many, rejects this claim.

We're talking about system concepts here. My concept of science differs
from the one you appear to be supporting. Simply as a matter of
pragmatics, I claim that adopting my system concept will lead to greater
scientific progress than adopting the other one will. When scientific
facts are determined by majority vote, all kinds of pressures come to
bear -- the desire for approval, self-doubt, the desire for power, fame,
and fortune, religious or philosophical beliefs, political maneuvering,
prejudices, rhetoric, and so on down the line. There are many ways to
win agreement, most of them being irrelevant to the determination of
facts.

Under the system concept I have adopted (from others), there is only one
kind of pressure on the scientist: the pressure to make predictions
match observations. Ten thousand of your colleagues may say your theory
is wrong, but if your theory predicts correctly, they are wrong and you
are right.
-----------------------------------------------------------------------
Best to all,

Bill P.