From Tom Bourbon [930813.1112]
I received the following article in the mail on 12 August 1993:
Todd A. Nelson (1993). The hierarchical organization of
behavior: A useful feedback model of self-regulation. Current
Directions in Psychological Research, 2(4): 121-126.
(CDPR, published by the American Psychological Society, is "...
the bimonthly journal that provides a timely source of
information spanning the entire spectrum of scientific psychology
and its applications. CD publishes brief [2000-2500 words],
scholarly reviews that focus on emerging trends, controversies,
and issues of enduring importance to the science of psychology."
I will share some expanded versions of the comments and questions
I wrote in the margins of the journal while I read the article.
If you have no interest in such things, hit "zap" now.
Initially, I believed Nelson might be a good messenger who accurately
reports the bad news he bears: He accurately summarizes the work
of Carver and Scheier, as I understand that work. From what he
says in the article, I do not know if he recognizes the points on
which they are wrong, or if he agrees with them. However, after I composed
this review I learned (from a highly placed reliable unnamed source) that
Nelson knew long ago that there are problems with the article. Therefore,
whenever you come to a section where I expressed uncertainty on that point,
know that I retract that expression. The author, Nelson, must share the
heat with the people he cites. I sent him a private comment at the address
published with his article, but I have learned he might not be there. If
anyone knows his present address, ssend it to me, or send him a copy of this
review. The published address is:
Todd A. Nelson, Dept. of Psychology, Michigan State University,
22817mgr@msu.bitnet.
···
+++++++++++++++++++++++++++++++++++++++++++++++++++++++
The article begins with a brief statement that, for much of the
history of the field, psychologists rejected goals as
explanations of behavior. But more recently, "... an increasing
number of psychologists have constructed theoretical models in
which goal pursuit plays a central role. These models have
attempted to account for the mechanisms whereby goals are
selected, pursued, and attained or abandoned. Such mechanisms
are now widely recognized to be central to the understanding of
the origins of voluntary behavior" (p. 121). (Here the author
cites material on goal-directedness, by L. A. Previn, and by C.
S. Carver and M. S. Scheier.)
COMMENT: The remarks about the history of psychology are
accurate, as is the comment about recent ideas on "goal pursuit."
The first hint that there might be problems with the presentation
in the article comes in the phrase "goal pursuit." Theories of
goal pursuit often differ in significant ways from perceptual
control theory (PCT), which is a theory and model for goal-
maintaining; for keeping perceptions at their reference levels.
To determine whether that potential problem materializes, we must
read the remainder of the article.
++++++++++++++++++++++++++++++++++++++++++++
"In the literature, these models have typically been discussed in
terms of self-regulation. Essentially self-regulation is a term
for a self-correcting system that works to keep an individual 'on
track' toward a particular goal" (p. 121).
COMMENT: Self-regulation = self-correction = a system that keeps
an individual on track to a goal. Is the individual the system
that keeps the individual on track; or is the self-correcting
system different from the individual? This might be a small
point, in one sense, or it might reveal some fuzziness in the
model that will be presented in the article. Will the individual
be portrayed as a living control system, perhaps one renamed a
"self-regulating/correcting system," or will the intrinsic,
defining nature of life-as-control be missed? Is the system
regulating itself, as the name suggests, or is this the name for
a system in which a self regulates other things?
+++++++++++++++++++++++++++++++++++++++++++++
The author says most recent models of this process have limited
focus but that, "One recent model of self-regulation, however,
developed by Carver and Scheier, has been shown in a number of
studies to have broad applicability not only for researchers
investigating self-regulation processes, but also for behavior in
general" (p. 121.).
COMMENT: The hint of possible problems seems to have been real.
The author draws a distinction between self-regulation (a system
that keeps an individual on track toward a goal) and "behavior in
general." This carries the strong implication that self-
regulation is a system or process apart from the central or
defining features of living control systems.
Also, will the author cite new research, different from the
studies Carver and Scheier published over the years? Those are
studies in what Phil Runkel identifies as the tradition of
inappropriately applied net-casting, with statistically
significant results described as though the conclusions apply to
all people. (Those are my personal interpretations of their
research. Anyone who wishes to assess the fairness or accuracy
of my views can read their extensive list of publications.)
++++++++++++++++++++++++++++++++++++++++++
"Two decades ago, Powers postulated what he called a 'control
theory' of human behavior" (p. 121).
The citations here are both from 1973 -- BCP and the Science
article. There follows a brief and generally accurate description
of the workings of an elemental control system.
After dismissing S-R theories of behavior, the author says, "...
many researchers have shown the utility of conceiving of human
behavior as a cause of subsequent behavior in and of itself, in a
closed-loop system" (p. 121).
The citations here are to Wayne Hershberger's book on volitional
action, to D.M. MacKay (1987), to Powers's 1973 publications and to
his Living Control Systems, vol 1.
COMMENT: It was good to see citations of some of the more
recent material on PCT, but discouraging to see it cited as supporting
the idea that behavior causes subsequent behavior, an idea which
is not different, at the core, from the strict S-R concept that
behavior becomes a stimulus for subsequent behavior. That is not
what a closed-loop system is about. If that is literally what
the author meant, then he was talking about a sequential linear
system. Nelson's intention is not clear.
In the paper, there are many accurate statements about
differences between control systems and S-R systems, but they are
interspersed with erroneous, or at least fuzzy, statements that
eventually leave me thinking the author has some misconceptions
about what a control system is and about how one works. More
accurately, I have no doubt that Carver and Scheier hold those
misconceptions; perhaps Nelson is only being the good messenger.
+++++++++++++++++++++++++++++++++
When Nelson compares cause-effect and control-system models, he
aptly identifies a major problem for C-E models: "In other words, for a
continuous behavior, such as driving a car on the freeway, it is
difficult (though not impossible) to pinpoint the cause- effect
relationships of environmental disturbances on behavior, because
the behavior is not discrete. ... The notion of feedback in
control theory overcomes this difficulty quite nicely " (p. 122).
COMMENT: This statement might reveal solid insight into a major
problem for C-E models and with how CT can overcome that problem; but
I am not certain what he means by that parenthetical phrase, "though
not impossible." I agree fully if he means we can apply disturbances
to assumed controlled variables and continue doing that until the
person's opposition to our disturbance lets us know we have found a
potential controlled variable, and if he means that task can be difficult
but not necessarily impossible.
+++++++++++++++++++++++++++++++++++++++++++++
A section headed: "A feedback model of self-regulation"
"Carver and Scheier adopted many of the principles of control
theory as conceptual heuristics in testing and formulating their
own model of self-regulation" (p. 122). (The citations are to
Carver and Scheier, 1981, 1982, 1988.)
COMMENT: They "adopted" principles as "conceptual heuristics" in
developing and testing "their own model of self-regulation."
This is a crucial stage in the article. Will what follows show
that, indeed, C&S started from Powers's ideas, then shaped them
into something new and different, to explain something different
from what is explained in PCT? Or did they not "adopt" some
principles as heuristics, but instead "appropriate" the original
model, taking possession of it, in essentially the original form,
in their own names? That is a common practice in psychology. I
do not prejudge C&S on this point; I must read more before
reaching a conclusion. The job is made easier when reading
claims of originality in control theory; PCT is more like a
theory or model in the "hard" sciences, where new and improved
means, containing the original but accounting for more and doing
it better. In physics, for example, people can't get away with
renaming Newton's or Einstein's or Joe and Josephine Smith's
ideas as their own, then rushing into print. Similarly, it is
often easy to see if a "new and original" version of control
theory is instead a flagrant case of appropriation.
+++++++++++++++++++++++++++++++++++++++++++
The paragraph in which the author describes the new model
proposed by C&S continues: "Essentially, they view behavior as
*self-regulating* [italics in the original] in that humans
actively engage in self-correcting behaviors to maintain a
perception of the environment in accordance with a reference
value (or standard). Carver and Scheier conceptualize behavior
as being goal-directed, and sensed deviations from a state of
goal-directedness bring about behavior aimed at once again
attaining a perceived goal-directed state" (p. 122).
COMMENT: Now I am confused. Earlier the author implied that C&S
say, in essence,
self-regulation = self-correction = a system that keeps an
individual on track to a goal.
Here, he says in one sentence that C&S say,
self-regulation = self-correction = maintaining a perception of
the environment in accordance with a
reference value (or standard),
then in the next sentence I believe he says that they say,
self-regulation = self-correction = a system that maintains a
person's perception of being in
a state of goal-directedness.
The latter statement sounds like a description of a program-level
control loop from the original versions of hierarchical PCT:
reference signal = "doing this";
perceptual signal = state of "doing."
EXAMPLE:
ref.sig = "search for your glasses -- dummy. If ___, then
___, else _____, until ___";
perceptual sig. = perceived state of what dummy is doing.
If this is what C&S say, then they merely describe actions at the
program level, as it was originally presented by Powers. That
would make "self-regulation" synonymous with "control at the
program level." Would this qualify as an "adoption" by them of
Powers's "principles" that they then used as "conceptual
heuristics" that resulted in them creating "their own model of
self-regulation?" No.
+++++++++++++++++++++++++++++++++++++++
"It is important to keep in mind that behavior, according to the
model, is not the end product of self-regulation. Rather,
behavior is the process by which one self-regulates through
changing one's perceptions so that they match a standard" (p.
122).
COMMENT: Now the definition of "self-regulation" is back to,
"acting to make one's perceptions match a standard," but not
necessarily at the program level. My interpretation is in line
with the author's examples, some of which seem not to be at the
program level, although I am not certain about that.
+++++++++++++++++++++++++++++++++++++++++
Subsection: "Feedback and Action"
This starts with the basic diagram and description -- not bad,
for the most part. But there is this: "Recall that self-
regulation refers to self-corrective, conscious action designed
to restore perceived continuity in goal-directed behavior. In
other words, people seek to maintain low discrepancies between
their perceptions of their behavior and their standards for that
behavior, and this is accomplished through a negative-feedback
loop (see Fig. 1)." (pp. 122-123).
COMMENT: The suggestion that readers should "recall" that self-
regulation is "conscious" was actually the first mention of
"conscious." But that is a small point. Here is a clear and
direct statement that, by self-regulation, Carver and Scheier
mean control of one's own behavior, unless of course by
"behavior" they mean "doing a particular program."
I can no longer feign innocence. In most of their writings, since at
least the early 1980s, Carver and Scheier have said we seek to
control our own behavior. Whether they comprehend the
implications of those words, so far as they might be tested in a
working model of a control system, I cannot say for a fact, but I
have an opinion. I have always thought they showed little
interest in the lower levels of the hierarchy, where the model
meets the world.
++++++++++++++++++++++++++++++++++++++++++
I will comment on the next paragraph, from page 123, one
sentence at a time.
"When perceptions match a person's standard, then no correcting
behavior is initiated, and the negative-feedback loop is not
engaged after the comparator function."
COMMENT: This claim reflects a serious miscomprehension of how a
control system works. It is one of the ideas, sometimes
erroneous, sometimes deliberately false, that people use to
refute or reject PCT. It is true that when (r - p = 0), then
(k*0 = 0), and (delta qo = 0) -- when perception matches
reference, output does not change -- but the idea that the loop
is no longer engaged is false. Dag Forssell tells of how Bill
Powers convinced him this idea is false. While Dag was driving
Bill to the airport in Los Angeles, Dag said something about not
controlling when error signals are zero. At the time, Dag was
not turning the steering wheel, but was letting it remain where
it was. Bill reached over and gave the wheel a quick jerk. Sic
transit Dag's misconception.
Also, this interpretation of control leaves out the never-ending
occurrence of disturbances, both in the world outside the skin of
the individual, and inside the machinery of control -- muscles, joints,
neurons and the like.
"Inherent in this model is the assumption that self-regulation is
a continuous, dynamic process that never ends."
COMMENT: No comment, other than, "yes." Well, that depends on
the intended meaning of "self-regulation," and I am not really
certain that I know it.
"Attainment of a goal causes a shift in the whole system, and new
goals are set, as are new reference values."
COMMENT: Sighs, moans and groans. What began as an article for
which I held out hopes, and on which I reserved judgment until I
had read further, just vaporized in front of me. It is in fact
about goals as something different from reference signals.
Exactly what they are in the model, if not reference signals, is
never stated. And the attainment of a goal is said to cause a
shift in the whole system. (When Dag sees the car and road
aligned the way he wants, he lets go of the wheel and starts
writing the script for his next video. Come to think of it, he
does drive in L.A. ... .)
If this is supposed to be a general control-system model of
behavior, then it fails at the most elemental level. On the
other hand, if "Carver and Scheier's model" is supposed to be
about program-level control, perhaps in a TOTE-system-like
program, then it is true that, for some programs, satisfying the
logical contingencies of the program ends the program. At least
that loop in the program ends and perhaps another begins. But if
that is all they (Carver and Scheier) are saying then it is at
least inaccurate, if not dishonest, for someone to believe they
created a "new model" of control.
++++++++++++++++++++++++++++++++++++++++
Section titled: "The hierarchical construction of behavior"
"Carver and Scheier's model adopts Powers's proposal that
behavior is organized hierarchically, in a cascading-loop
structure" (p. 123).
[TB: There follows an accurate description
of three levels: system concepts, principles, and programs.]
"Only the top three levels of Powers's nine-level hierarchy are
presented here, because the focus of Carver and Scheier's model
is on these three levels" (p. 123).
COMMENT: This is true to most of their material I have read, which
is not everything but is still a lot. C & S decided to focus on
the top three levels, then to run a loop from the program level
out into the world and back. There is nothing inherently wrong
with doing that; the practice certainly makes the diagrams
simpler in their articles. But there are dangers. One of the
dangers is that, if we take as a hypothetical that there are
authors who do that and who are not up to snuff on their
understanding of what the model really is -- a working, behaving
model in which changes are made only if they make the model work
better, rather than another psychological "perspective" or
description that can be changed and appropriated as one's own --
if they are not up to snuff on that, then those hypothetical
authors can in blissful ignorance suggest word changes that
vitiate the real model, leaving it nonfunctional. That is not
legitimate, especially if, in the process, the hypothetical
authors assert that now the model is theirs. (Of course, once
they destroy it, they are welcome to it!)
Carver and Scheier begin nearly all of their publications with
credit to Powers and with a solid description of an elemetnal
control loop. Sometimes they include "their" three-loop, high-
level hierarchy, which is accurately described, up to a point.
After that, the appropriation begins and the descriptions of
control as a phenomenon, and of control theory as a model for
control, degenerate. I do not question their general knowledge
of psychological matters, or their skill and integrity as
researchers, or their motives as scientists and authors, but I
have believed, since the early 80s, that they do not know how a
perceptual control model works -- how and why it behaves. And I
have believed it was that specific lack of insight that led them,
time and again, to slide back and forth between a good beginning
to an article or chapter, and statements about control and about
control systems that are demonstrably false. I have also thought
it was unfortunate that they did not avail themselves of the
experimental literature on modeling and simulating perceptual
control systems. Their work addresses important topics in
psychology and in related fields; how much better if, as widely
recognized authorities on control theory, they presented the very
best information -- accurate and informed. Instead, they present
an easy target for critics who mistakenly believe they (Carver
and Scheier) speak as authorities on perceptual control.
++++++++++++++++++++++++++++++++++++++++++++++
"Empirical tests of the notion of negative feedback made it clear
that there were some gaps in Powers's ideas, and that, in order
for the model to be complete, it had to take into account the
effects of attention and outcome expectancy" (p. 124).
COMMENT: I agree that there were gaps in the original presentations
of the model and that there are still gaps, some of them are huge.
But the empirical tests alluded to are studies Phil Runkel could have
used in his book (Casting Nets and Testing Specimens: Two Grand
Methods of Psychology) as examples of misuses and abuses of net-
casting. Many of them rely on correlations that are very high on
the index of uselessness -- the coefficient of alienation. In the
remaining sections of the article, Nelson cites numerous experimental
results from C & S in which, "when X happens, people do Y," or more
often, "when Q happens, people TEND TO do Z." There is no evidence
that the experimenters ever performed direct tests to identify controlled
variables or that they realized the inadequacy of their data as
evidence concerning the phenomenon of control.
++++++++++++++++++++++++++++++++++++++++++++++
The article continues for a few more pages, but I will not
comment on more details. The remainder includes examples of the
experimental "evidence" for "Carver and Scheier's model."
The research tradition that is home for Carver and Scheier is at
the opposite pole from PCT research. The criteria by which they
decide a construct is supported are the inappropriately-applied
tests of statistical hypotheses used in mainstream behavioral
science. The criteria by which they decide a restatement of a
theory deserves designation as a new theory -- their new theory --
are foreign to PCT, or at least to the Control Systems Group.
But "Carver and Scheier's model" is widely published and widely
known. And here we are.
Until later,
Tom Bourbon
Department of Neurosurgry
University of Texas Medical School-Houston Phone: 713-792-5760
6431 Fannin, Suite 7.138 Fax: 713-794-5084
Houston, TX 77030 USA tbourbon@heart.med.uth.tmc.edu