Response to Locke and Latham

[From Bill Powers (930526.0830 MDT)]

I suppose I have to do me dooty:


RE: Locke and Latham quotes provided by Greg Williams.

"As the influence of behaviorism has declined, a neo-behaviorist

theory is emerging to take its place. It is called control

Control theory was first formally applied to behavior in 1948;
the present version called PCT started in 1953. This was at the
height of the influence of behaviorism. What is emerging is a
belated and dim awareness by some psychologists that control
theory as an approach to behavior has been in existence for over
40 years.

" ... and can be viewed as a combination or integration of

Control theory shows that organisms are organized to control
their own inputs, not their outputs (their behavior). Behaviorism
has always maintained that the environment is ultimately in
control of behavior. Control theory shows that behavior is
produced by organisms to satisfy their own internal requirements,
and that influences from the environment are regularly rejected
and opposed if they interfere with meeting these internal

machine- computer theory (cybernetics),...

Control theory is based on the theory of analogue control
systems, which were invented by engineers trying to build
machines capable of controling variables outside them, which
before then had been the almost exclusive province of living
organisms. It is not a computer theory.

"goal setting theory [championed by Locke and Latham], ..."

Since control theory has been around far longer than the concepts
called goal-setting theory, it is hard to see how it could have
been derived from them. In fact control theory explains what
goals are and how they work, a subject about which the present
authors appear to care little and know nothing.

" and, by implication, drive-reduction theory."

Control theory does not employ the concept of drive, except in
the form of an error signal.

"It is derived directly from Miller, Galanter, and Pribram's TOTE
model (1960)."

PCT was initially developed in the years from 1953 to 1960; it
was first published in 1960, the same year the TOTE unit
appeared. The TOTE model is not a control-system model, but an
erroneous attempt to represent control as a program loop in a
digital computer. PCT was developed directly from the principles
of negative feedback control as found in engineering control
theory, and from neuroanatomy and physiology.

"The major concepts of control theory have been presented by
Campion and Lord (1982), Carver and Scheier (1982), Hyland
(1988), Lord and Hanges (1987), Powers (1973), and others."

All these authors but Powers acknowledge Powers as a main source
of their understanding of control theory, which is otherwise
sketchy. These other authors have said little in print to show
that they have any acquaintance with the major principles of
control theory. They treat control theory as if it is the theory
that organisms behave like control systems. They do not
understand what a control system is, what its properties are,
what distinguishes it from other types of systems, or what makes
it a particularly appropriate model of the behavior of organisms.
They make no use of the methods by which control systems can be
analyzed and modeled.

"In brief, the theory asserts that there is INPUT (a stimulus),
which is detected by a SENSOR. If there is a deviation (also
called a 'disturbance'), a SIGNAL is sent to an EFFECTOR, which
generates modified OUTPUT (a response)."

This is stimulus-response or stimulus-organism-response theory,
not control theory. A disturbance is not an input, but an
environmental event that would alter the input if it were not for
the actions of the organism that affect the same input at the
same time.

"This output becomes input for the next cycle."

Control systems do not operate in "cycles." All parts of the
control system and its local environment are active at the same
time. An analysis of a control system that treats each component
in sequence, one at a time, will make false predictions about the
behavior of the system and will be unable to show its actual

"In goal theory language, the input is feedback from previous
performance, the reference signal is the goal, the comparator is
the individual's conscious judgment, and the effector or response
is his or her subsequent action which works to reduce the
discrepancy between goal and performance."

If it is understood that all these processes are continuous
rather than sequential, this is a rough statement of control
theory. Control theory considerably predates "goal theory" as
proposed by Locke and Latham. If this statement is intended as a
contrast with control theory, it shows that the authors have no
understanding of control theory. They are reinventing the wheel.

"While control theory acknowledges the importance of goal
setting, there are serious, if not irredeemable, flaws in the
model. First, observe that the major 'motive' for action under
control theory is to remove disturbances or discrepancies between
the goal and the input (feedback). The natural state of the
organism is seen to be one of motionlessness or rest."

This is false. If the goal, for example, is to move ahead of a
competitor in the last 100 meters of a race, the control system
involved will adjust running speed to bring the perceived speed
to a match with the goal specification, and maintain it there
during this race. This can hardly be done through attaining a
state of motionlessness or rest.

"This is true of machines, but not of living organisms which are
naturally active."

It is not true of machines any more than it is true of organisms.
The control systems that maintain the proper acceleration,
velocity, and position in space of a departing space shuttle
craft are continuously active and maintain the sensed state of
affairs in a continually-changing condition, matching a
continually-changing goal specification. All control systems are
"naturally active," whatever that is supposed to mean.

"It is, in fact, a mechanistic version of the long-discredited
drive-reduction theory (Cofer & Appley, 1967)."

It does not use the concept of drive, nor is it based in any way
on the psychological theory of drive reduction.

Erlbaum, Hillsdale, New Jersey]) has observed that in this
aspect, control theory fundamentally misstates the actual source
of motivation: 'The behavioral process... does not begin with a
"test" of the discrepancy between the standard and the actual
states of affairs. Instead, it begins with a preliminary and
fundamental operation, namely the construction of the standard
itself, which, as a goal, is at the origin of the action and
directs its further course' (p. 145)."

The extended control-theoretic model known as hierarchical
perceptual control theory (HPCT) explains goal-setting as the
output act of a higher-level system: goals are set in order to
achieve higher-level purposes. From the standpoint of the higher
system, action consists of adjusting goals for lower systems.

From the standpoint of the lower systems, the adjustment of the

goal is one of the events that calls for action.

However, goals do not determine action. They specify inputs, not
outputs. The outputs, because of the negative feedback control
process, become what they must in order to counteract
unpredictable disturbances originating in the environment, as
well as maintaining the inputs near the specified (constant or
changing) goal state.

The setting of a goal seems "fundamental" only if one does not
ask why one goal is set rather than another, why a given goal is
set low or high on the scale of the variables involved, or why a
goal may be altered.

"Similarly, Bandura (in press [A. Bandura, "Reflections on
Nonability Determinants of Competence," in J. Kolligan & R.

Control theory agrees: if an existing goal is changed by a
higher-level system, the immediate result is to create a
discrepancy between the current perception and the newly-
specified goal-state. This discrepancy is then erased by changes
in action that alter the relevant perceptions toward the new

However, goal adjustments (a more inclusive term than goal-
setting) are not the only possible causes of discrepancies.
Independent environmental disturbances can alter perceptions,
creating a discrepancy between the perception and the already-
existing goal state. This, too, results in action that forces the
perception toward a match with the goal-state.

Another source of discrepancies is reorganization: perceptions
can alter, creating a changed impression of a constant
environment. This results in actions that bring the perception to
the goal state, but in doing so they bring the environment to a
new state rather than stabilizing it in the old state.
Reorganization can occur at any level; it can, for example, occur
at a level higher than the control system under consideration,
resulting in a change in the goal given the lower system and thus
producing a discrepancy -- an error signal -- in the lower

"Control theory begins in the middle rather than at the beginning
of the motivational sequence. To quote Bandura (in press):

        Human self-motivation relies on both DISCREPANCY
  control as well as FEEDBACK control. People initially motivate
  themselves through feedforward control by setting themselves
  valued challenging standards that create a state of
  disequilibrium and then mobilizing their their effort on the
  basis of anticipatory estimation of what it would take to
  reach them...."

The movitational sequence is not a sequence, but a closed loop.

Bandura tacitly defines a higher-level control system in saying
what he means by feedforward. Where does the "value" come from in
a "valued challenging standard?" The implication is that the new
goal state is not simply arbitrary, but is set as a means of
achieving something perceived as valuable to the organism: in
other words, to satisfy a higher-level goal. If the goal were
achieved (by the subsystem to which it is given), the result
would be perception of something that the organism desires at a
higher level: a challenge, or some other valued situation. So
"feedforward" is simply the action of a higher-level _feedback_

Setting a new goal, for any reason, creates an error signal; this
is already part of control theory, and has been since the 1950s.

  "Figure 1-3 [not reproduced here] shows how little of the
motivational process control theory, in its 'core' version,

The "core" version is undoubtedly a single isolated control
system, not the model of behavior contained in HPCT.

"The above is important because if discrepancy reduction is the
major motive, as implied by control theory, then the most logical
thing for an individual to do would simply be to adapt his or her
goal to the input."

This can happen if the higher system finds that a given
perception is difficult to control. Other perceptions will be
adjusted, by altering goals for them, to make up the difference.
However, discrepancy reduction is not a goal of a control system.
It is part of the basic mode of operation of the control system,
a side-effect. The aim of the control system is defined by the
reference input, the input that specifies the goal setting. It
brings its perception to that level by actions based on the
discrepancy. The preceding statement implies that there is only
one goal and only one level of organization. Control theory
proposes that many systems are active at many levels.

"Machines, of course, cannot do this because the standard has
been fixed by people at a certain level (as in setting a
thermostat). But people can and do change standards that diverge
from present performance."

Machines, of course, CAN do this quite easily, and commonly do
it. Multi-levelled control systems act just as HPCT proposes that
organisms act: one level of control adjusting reference signals
for lower levels. The attitude-control systems in a spacecraft
are given new attitude goals as a way of adjusting the thrust
vector, which is given new thrust goals as a way of adjusting the
orbital path -- all without human intervention.

The distinction being made is between a primitive one-level
machine and a complex multileveled human being. Naturally, the
simple machine loses. All this proves is that the authors are
ignorant of what machines can accomplish.

"If the individual's major motive were to remove disturbances,
people would never do this."

Multilevelled control systems operate just as much by creating
discrepancies as by correcting them. "Removing disturbances" is a
straw man, as well as showing that the authors don't understand
what the term "disturbance" means in control theory.

"Control theorists argue that lower-level goals are actually
caused by goals at a higher level in the individual's goal
hierarchy (Carver & Scheier, 1982). But this only pushes the
problem back a step. Why should people set higher- level goals if
they only want to reduce tension?"

Why, indeed? Control theory does not propose that people "only
want to reduce tension." "Tension" is a gratuitous term with no
relevance here. In the course of reducing discrepancies at a
higher level (for example, making a task more challenging in
order to reduce the discrepancy between a task perceived as easy
and a goal of perceiving that one is meeting a challenge), one
necessarily alters lower-level goals and thus creates
discrepancies for the lower-order systems, which they they
proceed to correct, thus bringing about the situation that the
higher system desires to perceive.

And higher goals don't "cause" lower ones. That is an
excruciatingly simplistic rendition of the relationships in HPCT.

"But in reality, people do set goals and then act to attain them;
they do not focus primarily on eliminating disturbances."

True. That is what control theory says, too.

"Removal of discrepancies and any associated tension is a
CORRELATE of goal-directed action, not its cause."

True, as mentioned above, except that goal-direction does not
determine action. It specifies perceptions. Actions are
determined in part by the goal but in equal and often greater
part by independent disturbances tending to make the perception
deviate from the goal. And discrepancies are not removed because
they produce some sort of tension. They are removed because
action is based on them, directly.

"The causal sequence begins with setting the goal, not with
removing deviations from it."

Yes, it can begin with a change in the goal (which includes, of
course, setting a new goal from scratch), but it can also begin
with a change in external disturbances that tend to cause a
perception to deviate from a goal state already being maintained.
Removing deviations, of course, occurs only when such deviations

"At a fundamental level, discrepancy reduction theories such as
control theory are inadequate because if people consistently
acted in accordance with them by trying to eliminate all
disturbances, they would all commit suicide -- because it would
be the only way to totally eliminate tension."

This is a sophomoric argument. If a new goal is set, thus
creating a discrepancy, what happens next? Is the discrepancy
simply left as it is, and the goal left unachieved? Control
theory is not a "discrepancy reduction" theory. Discrepancy
reduction, as the authors themselves say, is simply a correlate
of the control process. The discrepancies play a role in creating
the behavior that tends to eliminate them. But the disrepancies
do not produce themselves. They are produced by changes in
reference signals -- goals -- and by independent external

"If people chose instead to stay alive but set no goals, they
would soon die anyway."

Choosing to stay alive is a goal. The means of achieving it,
normally, is through setting other more specific goals.

" By the time they were forced into action by desperate,
unremitting hunger pangs, it would be too late to grow and
process the food they would need to survive."

This is degenerating into silliness.

  "In their major work, Carver and Scheier (1981) denied that
discrepancy reduction is motivated by a desire to reduce a drive
or state of tension. But their own explanation as to why people
at to reduce discrepancies is quite puzzling. 'The shift [of
action in the direction of the goal or standard] is a natural
consequence of the engagement of a discrepancy-reducing feedback
loop' (p. 145). This statement, of course, explains nothing. Why
is discrepancy reduction a 'natural consequence'?"

What a clear confession of ignorance! The authors obviously don't
understand how a control process works, why discrepancy reduction
follows from the organization of the system and its relationship
to its environment. They don't understand how a system can be
analyzed into functions, and how the operation of the functions,
given their interconnections, can be deduced from their
properties to predict the behavior of the whole system. What
seems to them a non-sequitur is simply a sign of a deficiency in
their educations. A person who knows nothing of physical
mechanics might ask, in the same vein and with the same air of
superior skepticism, what conservation of angular momentum has to
do with the fact that a spinning top precesses while resisting
falling over. The problem there is that the skeptic hasn't even
done enough homework to understand the answer.

"According to goal theory, BOTH discrepancy creation AND
discrepancy reduction occur for the same reason: because people
need and desire to attain goals."

This is not a "because" but a reiteration of the same observation
in different words. What is a "need" or a "desire"? It is a goal.
Control theory explains what goals are and how control systems
receiving goals work to create the sort of goal-seeking behavior
that we observe and experience. The explanation is not simply the
same observation repeated in different words, but an analysis of
how various functions work together to create the emergent

"Such actions are required for their survival, happiness, and

That's a true statement, but how is that requirement met? Why is
it a requirement? I require one million dollars for my well-
being. That does not tell me how to get it.

"A second problem with control theory is its very use of a
machine as a metaphor. The problem with such a metaphor is that
it cannot be taken too literally or it becomes highly misleading
(e.g., see Saundelands, Glynn, & Larson, 1988 [L.E. Sandelands,
M.A. Glynn, & J.R. Larson, "Task Performance and the 'Control' of
Feedback," Columbia University, unpublished manuscript])."

Control theory is not a machine metaphor. People are not "like"
control systems any more than the heart is "like" a pump. The
heart IS a pump; people ARE control systems. It is the
servomechanism that is a metaphor for the real behavior of living
control systems. Only those who know nothing of control theory
could possibly think it is "misleading."

And how the hell is a person to read an unpublished manuscript?

"For example, people do not operate within the deterministic,
closed-loop system that control theory suggests. In response to
negative feedback, for example, people can try harder or less

"Negative feedback" meaning, of course, a discouraging response
to one's performance. This is the pop-psychology version of
control theory: give me some feedback, I gave him some positive
feedback, and so forth. In control theory, the sign of the
feedback has nothing to do with its encouraging or discouraging
content. Feedback is the effect of a variable on itself through a
closed causal loop. The feedback is negative if the result of a
perturbation is to produce an opposite effect at the place where
the perturbation was introduced. The authors are talking about a
Psychology Today usage of the term "negative feedback" that is
irrelevant to the technical meaning of the term. Their source of
information about control theory clearly comes from a point as
far as possible from the horse's mouth.

"They can focus on the cause and perhaps change their strategy.
They can also lower the goal to match their performance; in some
cases they may raise their goal."

All of which is irrelevant; what they do depends on what will
bring their perceptions closer to the states they desire at all
levels of importance. When you say "they" can lower or raise
"their" goals, you are talking about two different "they"s: a
higher-level system sets goals for a lower-level system, not for

"Furthermore, they can reinterpret the discrepancy as unimportant
and ignore it or can even totally deny it."

Yes, a higher-level system can reduce the gain of a lower-level
system to zero, or set the reference signal to zero which is in
some cases equivalent. There are many things people can do. None
of them is inconsistent with control theory.

[Skipping more examples]

"In short, they can do any number of things other than respond in
machinelike fashion."

What is a "machine-like fashion?" It is the fashion of a machine
as conceived within the very limited knowledge of the authors. If
a machine is designed to work like a human being, however, it
becomes hard to tell just by observing the behavior whether the
responsible party is natural or artificial. The authors are
simply revealing how little they know about machines. They are
exhibiting, in their terms, a machine-like reaction to the sound
of the word "machine."

"Furthermore, people can feel varying degrees of satisfaction and
dissatisfaction, develop varying degrees of commitment to goals,
and assess their confidence in being able to reach them (Bandura,
1986). These emotions, decisions, and estimates affect what new
goals they will set and how they will respond to feedback
indicative of deviations from the goal (Bandura, 1988). Control
theory, insofar as it stresses a mechanistic model, simply has no
place for these alternatives, ..."

This is complete nonsense. All of these subjects have been
discussed in HPCT terms. Emotions and estimates are perceptions.
Decisions convert the difference between what is perceived and
what is desired to be perceived into selected plans. And so

"... which basically means that it has no place for

There is a non-sequitur par excellence. What do they mean,
"basically?" Do they have any notion of the nature of
consciousness? Of course not. Nobody has.

"Insofar as this is the case, the theory must fail for the same
reason behaviorism failed. Without studying and measuring
psychological processes, one cannot explain human action."

Ah, the bellowing owner of the gored ox. What is there about a
psychological explanation other than it uses undefined terms in a
vague way to purport to explain poorly-observed phenomena?
Control theory is precisely an attempt to improve on the
fuzziness and poor predictivity of psychological explanations.
Just producing explanations that sound familiarly psychological
is no indication that they actually explain anything.

"One might ask why control theory could not be expanded so as to
accommodate the ideas and processes noted above. Attempts have
been made to do this, but when it is done, the machine language
may still be retained. Hyland (1988), for example, described the
effects of goal importance or commitment in terms of 'error
sensitivity,' which is represented diagrammatically by a box
called an 'amplifier.' "

So it's the language that is bothersome. What about the functions
to which these oily clanking terms refer? If a critic is put off
by the sounds of words, there is little chance of finding out
what is meant by them.

"Expectations and memory are represented as 'symbolic control
loops.' Decision making is done not by a person but by a
'selector.' What is the benefit of translating relatively clear
and well-accepted concepts that apply to human beings into
computer language that is virtually incomprehensible when used to
describe human cognition? The greater the number of concepts
referring to states or actions of consciousness that are
relabeled in terms of machine language, the more implausible and
incomprehensible the whole enterprise becomes. Nuttin (1984, p.
148) wrote on this: 'When behavioral phenomena are translated
into cybernetic and computer language, their motivational aspect
is lost in the process. This occurs because motivation is foreign
to all machines.'"

This simply underscores the allergy to technical terms having
clear meanings. Anyone who uses terms like "implausible" and
"incomprehensible" is simply confessing ignorance, or laziness.
What is "relatively clear" about well-accepted concepts in
psychological language? The whole problem with unexamined
psychological terms is their almost total lack of clarity and
specificity. This entire paragraph is nothing but a complaint
against the unfamiliar and a declaration of loyalty to well-
established custom, no matter how little has been accomplished by
that custom.

"On the other hand, if additional concepts are brought into
control theory and not all relabeled in machine language (e.g.,
Lord & Hanges, 1987), then control theory loses its distinctive
character as a machine metaphor and becomes superfluous -- that
is, a conglomeration of ideas borrowed from OTHER theories."

Control theory is not just a collection of new terms. It is a
body of concepts concerning relationships and processes. It does
not borrow anything from other theories of human behavior: its
roots are in first principles. One can describe a control system
correctly and exactly without using a single machine-like term.
The interpretation of psychological terms in control theory
language is an attempt to give some clear meanings to those
psychological terms, meanings that are presently imprecise or
totally lacking. But to a person who hears only the language, and
is ignorant of the underlying relationships and processes, the
flavor of control theory is all in the sounds of the words.

"And if control theory does not make the needed changes and
expansions, it is inadequate to account for human action. Control
theory, therefore, seems to be caught in a triple bind from which
there is no escape. If it stays strictly mechanistic, it does not
work. If it uses mechanistic language to relabel concepts
referring to consciousness, it is incomprehensible. And if it
uses nonmechanistic concepts, it is unoriginal."

Control theory works quite well whether expressed in mechanistic
terms or not. Control theory is not strictly mechanistic. It does
not relabel concepts referring to consciousness, because nobody
knows how to refer to consciousness anyway, not even the authors.
It is incomprehensible only to people who have been too busy
fending it off to bother with learning anything about it. And
when it uses nonmechanistic concepts, it still reveals phenomena
and relationships among them that no psychologist has noticed --
and predicts them, where progress has permitted, with a precision
that is unknown in the field of psychology (particularly the
authors' branch of it).

"It has been argued that control theory is useful because it
provides a general model into which numerous other theories can
be integrated (Hyland, 1988). However, a general model that is
inadequate in itself cannot successfully provide an account of
the phenomena of other theories."

The authors have already rejected in principle any control theory
account of the phenomena of other theories; now they berate it
for being inadequate to the task. The line of this argument is
veering dangerously close to sleaze.

[There follows a long section finding fault with Carver and
Scheier's work in personality theory. As Carver and Scheier know
no more about the fundamentals of control theory than Locke and
Latham do, this is a squabble between two sides in an argument on
some other subject having nothing (but a few words) to do with
control theory. I therefore skip this section.]

"Finally, some have argued that control theory is original
because it deals with the issue of goal change (e.g., Campion &
Lord, 1982)."

This is not what makes control theory original. Control theory
explains what a goal is, and with HPCT offers a comprehensive
model explaining what it is that can change a goal, and what
effects are to be expected at several levels when this
arrangement exists. But what is really original about control
theory is its proof that behavior is not the production of
organized outputs, but the control of perceptual inputs in
relation to goal-specifications for those perceptions. Everyone
has known since the dawn of history that people have and pursue
goals. "Goal-setting theory" simply repeats what common sense has
always known. But before control theory, nobody could explain
what they meant by a goal, or by setting a goal.

"In sum, the present authors do not see what control theory has
added to our understanding of the process of goal setting; all it
has done is to restate a very limited aspect of goal theory in
another language, just as was done by behavior mod advocates."

It is very clear that the authors do not see. Their nothing-but
statement of what control theory has done is based on what other
researchers with only slightly less ignorance of control theory
have said and done. They do not recognize that the very concept
of a goal calls for a deeper explanation of the nature of a goal;
they are content to stop at the obvious phenomenon and reject all
attempts to find a deeper level of explanation. Control theory is
being evaluated strictly at the level of words and word
associations, with its real message being lost because of this
superficial treatment.

But what else can people do with control theory, when they know
no other level at which to understand anything?

Bill P.