Martin Taylor 951128 17:30
Peter J. Burke (11/28/95 13:37)
At some level, the pigeon is controlling for
the presence/absence of the girlfriend in the picture.
I would doubt that very much. This statement translates as "the pigeon wants
{to see/not to see} a picture of Herrnstein's girlfriend." To test this,
the pigeon would have to be pecking at something that made the particular
woman's picture appear after a disturbance had removed it from the pigeon's
view. I don't think that was what was tested. The pigeon had no influence
on whether the next picture to be presented was of that woman or some other.
ยทยทยท
----------------
I am not sure (philosophically,
even) whether it is possible to discover "the" controlled variable
(perception), since it is likely (as in this case) to be a gestalt which is
made of some weighted combination of stimuli from the external world (and
possibly from internal states as well).
That really isn't a big deal. Remember that in HPCT, each controlled
perception is a scalar value, the result of a function of several variables.
A function z = Ax + By can achieve a value of z=1 through an infinite
number of combinations of x and y. That doesn't stop the experimenter
from seeing if the manipulation of z (which the experimenter has intuited
might be the controlled variable) actually is countered by the subject's
output.
Now it is true that if A=1.01, B=2.1, and the experimenter intuited A=1
and B=2, the experimenter hasn't guessed right but will nevertheless
discover very good control. So in that sense you are correct. But all the
same, the experimenter's guess is a close approximation to a description
of the actual gestalt used by the subject, and so one could legitimately
say that the controlled perception has been (nearly) found.
It's more difficult when one is trying to decide between two functional
forms that have a range in which they run fairly parallel, as do log and
square-root, and yet more difficult to make the first intuition. If the
pigeon were in fact controlling for seeing Herrnstein's girlfriend, what
disturbances would be resisted? Fun-house mirrors? Showing the girlfriend's
twin sister? Showing another woman of similar build and age?...
The experimenter verbalizes or symbolizes the class that the perceptual
input function of the control system determines. The experimenter probes
different areas of the high-dimensional subspace in which the function is
presumed to lie, and in principle can determine the function even without
being able to give it an anlytical form. But a prior intuition makes it
a lot easier, especially if the field covered by the subspace can't be
predetermined.
Martin
[From Bruce Abbott (951129.1040 EST)]
Bill Powers (951129.0600 MST) --
I would like to add a question to that discussion: How did Herrnstein
discover that the controlled variable involved the presence of
Herrnstein's girlfriend in the picture? The pigeon appeared to be
controlling for "(girl present and YES key) or (girl not present and NO
key)". But perhaps the pigeon saw something else in common among the
pictures in which the girl was present, other than recognizing that a
particular human being was there (you brought up the same point).
Herrnstein evidently assumed that because HE saw the same person in the
YES pictures, the pigeon also saw the same person. This explanation is
_sufficient_ to explain the results, but is it _necessary_? A
considerably more careful experiment would have to be done to answer
that question, at least as thorough as the experiments that exposed the
Clever Hans hoax.
No single experiment can eliminate all the sensible alternative
explanations. Herrnstein didn't claim that he had proven his conclusion,
only that the experimental results strongly supported it. But remember how
this study was conducted: Throughout training and testing, the pigeon never
saw the same slide twice. The pigeon had to learn to identify _something_
that was consistently present in the slides containing Herrnstein's
girlfriend and consistently absent in the slides that didn't. Whatever it
was, when the pigeons made mistakes, they tended to be the same ones made by
the human participants, suggesting that both species used either the same
features, or that each used features highly correlated with those used by
the other. This was not so much the case when a different set of slides was
used in which the relevant feature was the presence of water, suggesting
that pigeon and human were then using somewhat different criteria.
In focusing on this supposed defect of the study, you have managed to
deflect attention from the issue this example was intended to address. It
was claimed that all the observations from operant conditioning studies are
useless because they are merely a product of the "behavioral illusion." I
offered this study as a counterexample, and so far the only replies I have
received are (a) Rick's assertion (without supporting argument) that his
claim is still true and (b) quibbles about Herrnstein's interpretation.
There's an old saying that "the exception proves the rule." As currently
interpreted by the lay public it is idiotic (how can an exception show that
the rule is correct when in fact it undermines the rule?). However, when it
was originally coined the word "prove" had a different meaning, and with
this meaning the saying makes sense. The word "prove" meant "test." Thus,
the exception _tests_ the rule.
With Herrnestein's experiment, I've offered the exception that tests the
rule. Unless you can demonstrate that the exception does not apply, the
rule is hereby disproven.
Regards,
Bruce
[From Chris Cherpas (951129.0912 PT)]
[re: > Bill Powers (951129.0600 MST)]
[re: >>Martin Taylor 951128 17:30]
BP:
Excellent reply to Chris Cherpas on Herrnstein's girlfriend.
cc:
Not that it matters much, but I had not posted anything on
Herrnstein's girlfriend (my wife forbids me to do anything
on Herrnstein's girlfriend =8^@).
By while we're on the subject, I have a question. Don't you
think it's an interesting fact that a pigeon will reliably
peck one key in the presence of slides that Herrnstein
perceived his girlfriend was in and peck on another in the
presence of slides in which Herrnstein did not perceive
his girlfriend to be? By the way, a variation was done
by Will Vaughan in which, I believe, the slides contained
either artifacts (like buildings) or didn't contain artifacts
and pigeons acquired the differential pattern described
above; then, when the experimenter switched the contingencies
so that artifact slides were "no" and non-artifact slides
were "yes," the pigeons reversed their pattern after only a few
instances -- in other words, they partitioned a set into
two equivalence classes. This "reversal" kind of experiment
is one of the ways comparative psychologists have claimed to
study intelligence.
My main point: so far in the PCT experimental literature I've
encountered, there seems to be a lot of tracking going on.
Fine. Is there anything else actually interesting that PCTers
can do an experiment on? For years I've heard the argument
from cognitivists that behaviorist research may be OK for
simple phenomena but that the cog sci people were tackling
the "interesting" problems (e.g., the nature of expertise)
-- and, in many cases, I'd have to agree, although I don't
know how useful some of that cog sci stuff will prove to be
for a unified science.
What, for example, has PCT experimental work contributed to
our understanding of "concepts?" In short, is there anything
beyond "thermostat research" here?
Sincerely,
cc