Shouten, Hebbian learning, and let's all take a breath

[From Dick Robertson, 2009.06.05.1422CDT]

Do I remember, and am I completely delusional, that the entry into this thread was Martin’s claim that some non-PCT research could still contain useful information, despite being based on a false paradigm?

I have been trying to follow the discussion, and trying to keep focusing my perception of the discussion on that original assertion. I understand that refining the interpretation of Shouten’s article involves achieving consensus (if possible) about whether the whole thing can be considered an example of closed loop behavior, or whether some components can be, or none of it.

The discussion has become quite complex and I’m starting to feel like the kid who’se trying to jump on the merry go round when it’s going too fast. But I still would hope that original assertion gets addressed (assuming I recall it correctly) as the struggle for consensus on the analysis proceeds.

Best,

Dick R

···

----- Original Message -----

[From Rick Marken (2009.06.05.0830)]

i[From Rick Marken (2009.06.05.1420)]

Dick Robertson (2009.06.05.1422CDT)--

Thanks for the heads up on the mindreadings domain. My registration
did run out and the page that's up there is just a filler (according
to the domain vendor); it wasn't taken over by someone else. I've now
paid for 5 more years and I was told that mindreadings.com will point
to my files again in about 24 hours.

While I'm on the subject I should note that I have been working on the
multiple regression (MR) improvement to the "Test for the Controlled
Variable" demo. I've rewritten the java program so that it recomputes
an MR on every iteration of the program and you can't tell at all in
terms of the smoothness of the tracking; the matrix inversion code is
actually quite concise. It turned out to be a pretty easy upgrade. I
haven't put the new version up yet because I'm not sure it really does
a better job than the current version, which doesn't take the between
disturbance correlations into account (like the MR version does). I'm
currently involved in doing some tests to see what's going on.

Do I remember, and am I completely delusional, that the entry into this
thread was Martin's claim that some non-PCT research could still contain
useful information, despite being based on a false paradigm?

That's exactly right. Martin said he liked my "Revolution" paper
(which said just what you say: you can't get useful information from
research based on the wrong paradigm) but then he went on to say that
it didn't apply to certain research (like the Schouten experiment)
because... and that's where we get into what Martin sees as "support
artillery", what I see as "friendly fire" and what you see as
complexity.

I have been trying to follow the discussion, and trying to keep focusing my
perception of the discussion on that original assertion.

Yes, and I'm trying to do that too. It's not easy. My current approach
is to ask Martin to explain the difference between a conventional
experiment, like Schouten's, that provides useful results, and one
like, say, the Shepard/Metzler experiment I describe in my
"Revolution" paper, that doesn't.

I understand that
refining the interpretation of Shouten's article involves achieving
consensus (if possible) about whether the whole thing can be considered an
example of closed loop behavior, or whether some components can be, or none
of it.

Martin seems to be saying that in the Schouten experiment, unlike in
the conventional psychology experiments I discuss in the "Revolution"
paper, the causal path from IV (light presentation) to DV (button
press) is open loop because the DV has no effect on the IV. And
because of this the observed relationship between IV and DV gives
useful information about the open-loop function relating stimulus
input to perception. I am saying that the path from IV to DV is open
loop in all experiments; it's the connection between controlled
(stimulus) input (qi, which is affected by both IV and DV) and output
that is closed loop, which is what makes it impossible to get useful
information about the path from stimulus input (qi) to perception in
any experiment where you don't test to determine what variable is
under control (disturbed by the IV and influenced by the DV).

The discussion has become quite complex and I'm starting to feel like the
kid who'se trying to jump on the merry go round when it's going too fast.

I think this is one of the unfortunate consequences of Martin's
"supporting artillery". The methodological implications of PCT are
quite complex, as you can see if you have ever tried to make it
through Bill's masterpiece on the topic: Powers, W. T. (1978)
Quantitative analysis of purposive systems: Some spadework at the
foundations of scientific psychology. Psychological Review, 85,
417-435). It's hard enough to make this stuff clear and easy when all
the "teachers" are on the same page. But when one of the teachers
keeps contradicting the others things get really chaotic.

I would suggest waiting until Martin presents an example of a
conventional experiment that gives useless results ("useless' in the
sense described in my "Revolution" paper) and explains the difference
between that experiment and the Schouten experiment, which, according
to Martin, gives useful results. I think if Martin can explain the
difference in a nice, simple, concrete way we both may be better to
understand Martin's point.

Best regards

Rick

···

--
Richard S. Marken PhD
rsmarken@gmail.com