Re: PCT Popularity; and Why 99%?
From Tom Bourbon (921022.13:45)
I second Rick Marken's claim that we should not neuter PCT in an
attempt to gain a wider audience. I will describe some of the
pressures I have encountered to do that. Then I will identify
several people who have published extensively, claiming that
control theory is compatible with everything else in behavioral-
social-cognitive science. Then I will describe a clear example of
the disappointing consequences that follow from their having
published distorted, simpatico versions of control theory.
[Rick Marken (921021.1000)]
Greg Williams says:
Bill, who am I to tell you that you should perceive a problem
with PCT being a tough sell? That I believe there IS a problem
with it is MY ideology. There is no perceived problem for you if
you believe that you are right about what everyone SHOULD believe
to be important, and that virtually everyone else is wrong, even
if you believe PCT ideas deserve wider attention... until you
begin to perceive people shrugging and walking away when you tell
them what they should and shouldn't believe to be important. And
then I won't have to tell you that you should perceive a problem;
you'll be perceiving it.
Rick Marken replies:
···
I think this gets at the heart of Greg's complaint; he would
really like to see PCT ideas get wider attention. I think he sees
a lack of interest on the part of some PCTers (like myself) in
finding common ground with psychologists, roboticists, biologists,
AIers, ALifers, etc etc -- ie. with others in the community of
life scientists who might profit from an examination of PCT ideas.
#############
TB:
Rick describes a frequent interpretation offered by reviewers,
editors, and others who see or hear manuscripts and presentations
on PCT, especially when the presentations or manuscripts come from
"hardcore" PCTers. I have a collection of reviews in which the
writers say I (and my co-authors if there are any) went out of my
way to make PCT unpalatable, or that I want readers to reject PCT.
Most of them continue with remarks that I should point out how PCT
"is like --- ;" or "is similar to --- ;" or "is just another way of
saying --- ." Or THEY say I should say that "we (they) already know
--- ." A few say that most strong assertions by PCT writers (eg.,
behavior controls perception) are "merely ideological." etc.,
etc., etc.. When we decline to mend our evil ways (by going along
with what the reviewers say) our refusal is often taken as proof
that we do not want people to read about or to understand PCT. Of
course, their comments are self-fulfilling: after they tell the
editors not to publish our manuscripts, no one reads about PCT.
I believe it is essential that we avoid presenting PCT in a
watered-down version and that we resist all suggestions that it
"offers another perspective" on the same old things, or that it is
"a convenient framework" for "viewing" and unifying things "we
already know." Below, I will present a clear example of why I
think we must hew close to the basics.
#######################
RM:
Those of us who do not seem to "compromise", "see commonalities",
etc believe that we are just presenting the PCT model -- we don't
feel that there is an agenda to alienate potential friends; but
apparently it seems like this is true to some people (like Greg,
I think).
##################
TB:
Rick's point is well taken. You have not lived life to the fullest
until you labor to submit a manuscript on PCT (perhaps for the
second or third or ... time) then read a reviewer's smug accusation
that you (the one who wrote the manuscript) are "determined to
elicit rejection." It calms the spirit and soothes the nerves.
####################
RM:
I think this is worth a discussion. I take the view that the way
to promulgate PCT is to present the model and the research
honestly, doing what we can to relate this to existing relevant
concerns, but not shying away from explaining the true
implications of the model. I believe that attempts to "find common
ground" produce the Carver/Scheier approach to PCT -- which ends
up using the terminology of PCT but misses the basic point (and
succumbs to the causal view of behavior in the end). I realize
that the approach I advocate is not a good way to drum up a PCT
following -- most people do shrug and walk away simply because
they don't see the problem that PCT solves -- or don't get PCT
even if they do see the problem. But I don't think it is worth it
to compromise the model to try to get recruits -- PCT is neither
a religion nor a political party. And every so often someone DOES
get stoked on it (like I did). I think we are gaining PCTers who
really get it (ie. model-ers) at the rate of about 1 a year now.
That's plenty for me. I just don't think you can MAKE people be
interested in PCT. When the light does go on in a person, PCT
sells itself. It's not worth distorting the model to try to get
people interested -- you just end up with people who are really
interested in the version of PCT that you made up for their sake.
But, I'd like to hear your point of view Greg.
#####################
TB:
Agreed, on practically every count. It is a mistake to distort,
water down, neutralize, or defang PCT. We should not go out of our
way to build bridges, identify communalities, find common ground,
etc, when the other side of the river is quicksand. Control BY an
organism is different from control OF an organism (even if "control
of" is camouflaged in the contemporary jargon of "behavioral
analysis" or "cognitive science"). Period. Crisp predictions by
a generative model of control behavior are not the same as
statistical mush, in which significant differences between mean
scores from groups, and low but "significant" correlations, are
offered as evidence supporting one or another "theory" of behavior.
Period. Usually the behavioral phenomena or the cognitive-
emotional-social processes alleged to exist on the other side of
the river are phantoms.
In such cases, there is nothing to which we can build a bridge --
there is no communality. But that does not prevent some writers
from publishing extensively with claims that PCT can be all things
to all people. Rick mentioned Carver and Scheier. They are part
of the list of "villains" that rolls from my tongue or fingertips
as one entity:
Carver-Scheier-Hyland-Lord-Hollenbeck.
That would be C.S. Carver, M.F. Scheier, M. Hyland, R.G. Lord, and
J.R. Hollenbeck. There are others, but this group deserves special
attention. All offer CT as a "framework" or "perspective" or
"conceptualization" or "view" for virtually everything. All "build
bridges," "promote unity," "integrate," and all of the other things
PCTers are so often urged to do. Collectively, the members of my
rogues' gallery have published tens of times more material on what
they call control theory than have any participants on CSG-L. For
them, publication is easy -- say that control theory is not a
threat, that it affords another compatible perspective on
everything, and no one is bothered. Isn't that the best way to
spread the word about PCT?
No!
Their presentations of a nonfunctional and eviscerated control
theory have done far more harm than good. On this opinion, I
cannot be moved. Many people have formed their "understanding" of
control theory by reading the numerous publications of that group.
Their collective writings are so extensive that an innocent reader
could easily believe they form an authoritative literature. That
is not true. Time for my major case in point.
Bandura, Albert (1989). Human agency in social cognitive theory.
American Psychologist, 44, 1175-1184.
As part of his presentation, Bandura raised and knocked down a
pathetic description of a "negative-feedback" system. That
exercise occupies much of pages 1179-1191. To any PCTer, it was
obvious that he did not understand negative feedback, but Bandura
is an authority and now the article is cited widely and favorably.
Bill Powers wrote a "comment:"
Powers, William T. (1991). Commentary on Bandura's "Human
agency." American Psychologist, 46, 151-153. I recommend it.
A few others also submitted comments, some dripping with praise.
Of course, Bandura wrote a reply:
Bandura, Albert (1991). Human agency: The rhetoric and the
reality. American Psychologist, 46, 157-162.
Bandura missed the point of Bill's comments and continued to
discuss control theory in thoroughly negative tones. While I was
reading his reply, the awareness dawned that he was writing about
the control theory presented by the unholy alliance of Carver-
Scheier-Hyland-Lord-Hollenbeck, not about PCT. My insight was
confirmed when I read:
"Locke (in press) has argued that much of control theory involves
translation of the principles and knowledge of goal theory into
stilted machine language without providing a new perspective or
predictive benefits. He further showed that adherents of control
theory have now grafted so many ideas from other theories on the
negative feedback loop as (sic) to remedy its prediction problems
that control theory has lost its distinctiveness" (1991, p. 158).
Sadly, those are exactly my own conclusions when I assess the
literature of the devil's alliance. I knew Locke would cite the
popularizers and bridge builders.
He did:
Locke, Edwin A. (1991). Goal theory vs. control theory:
Contrasting approaches to understanding work motivation.
Motivation and Emotion, 15, 9-28.
Locke accurately summarized much of the material produced by the
"nice guys" -- the sweetness and light brigade -- of control
theory. EVERYONE who believes we should go out of our way to
mollify people from every other camp in cognitive-social-behavioral
science should read that article.
Weak-kneed presentations of PCT, in which core concepts are
abandoned or verbally "modified" every time some established critic
squeaks, do no good. They are misleading and harmful.
I am not suggesting that we go to the other extreme and bash
everyone who does not catch on to PCT as quickly as we might like.
But anyone who believes we will spread awareness of PCT more
quickly by deliberately softening its implications should read the
references I have cited, in chronological order.
Why 99%?
Earlier this year, a former graduate student and I submitted a
manuscript describing our modeling of cooperation by pairs of
people. Our results include numerous +.997 correlations between
predictions of moment-by-moment actions made by two interacting PCT
models and the actions of two people. In the manuscript, we cited
the exchange between Bandura and Powers, in American Psychologist,
and tried to build a bridge -- a real one, not a string of b***
s***. We suggested that Bandura's misunderstanding of control
theory and of negative feedback came from his familiarity with
faulty sources -- the nasty five. We also suggested that our
results provided a modest example of the predictive power of a
legitimate negative feedback model, namely, the model from PCT. We
were rejected. Among the many fascinating reasons, one reviewer
"assured the authors" that "Bandura would not be impressed." (Who
knows, maybe that review was by Bandura himself!)
Assuming the reviewer was right, what WOULD impress Bandura? He
gives a strong hint in his reply in American. Psychologist:
"As shown in Table 1, perceived self-efficacy accounts for a
substantial amount of variance in phobic behavior when anticipated
anxiety is partialed out, whereas the relationship between
anticipated anxiety and phobic behavior essentially disappears when
perceived self-efficacy is partialed out" (1991, p. 160).
Forget about the problems of defining terms and constructs --
quicksand and phantoms, all! Table 1 is on page 161. In it is a
summary of correlations from several studies on the aforementioned
phantoms. They range from -.22 to +.77. They are accompanied by
a cloud of asterisks: * ** ***. How silly I was not to see why
Bandura would be unimpressed by correlations of +.997! Our
correlations "accounted for" 99.4% of the variance; he must only be
impressed if you can "account for" a paltry 4.8% to 59%. Our
coefficient of alienation (aka, probability of failure in a
prediction) would be < 0.1; if his representative, the reviewer, is
correct, Bandura must want probabilities of failure ranging from >
0.98 to about 0.66. Now I understand.
(The "coefficient of alienation," or "coefficient of failure," was
a big topic on CSG-L long ago. For a good source -- an old one --
see:
Guilford, J.P. (1956). Fundamental statistics in psychology and
education. NY: McGraw-Hill.)
Why strive for 99% of the variance "accounted for?" It is simple:
To avoid the mistake of believing phantoms are facts; to avoid the
sham and scam of saying you have a science, when all you have is
statistical mush.
Why insist on hewing as close to the core of PCT as possible,
rather than making PCT an easy pill for all to swallow? Figure it
out.
Best wishes,
Tom Bourbon e-mail in care of:
MEG Laboratory PAPANICOLAOU@UTMBEACH.BITNET
1528 Postoffice Street PAPANICOLAOU@BEACH.UTMB.EDU
Galveston, TX 77550 PHONE (409) 763-6325
USA FAX (409) 762-9961