My Objective is to Help You Achieve Yours
|
Book Review Doing Research on PurposeA Control Theory Approach to Experimental Psychology by Richard S. Marken Reviewed by Bruce Nevin © Bruce Nevin 2015
|
|
It took a peculiar courage for Galileo to defend and
elaborate the Copernican theory that the earth was just one of the planets
orbiting around the sun when the consensus of science was that the sun
moves along a celestial sphere that spins about the earth. Such is the
courage demonstrated in this third volume of papers by Richard Marken in
experimental psychology. The title points to the problem. Just as the sun is
at the center of the solar system, simple introspection tells us purpose
is at the center of behavior. Our actions are not purposeless. But purpose
and introspection both were banished a century ago in a drive to make
psychology more scientific - and more serviceable to moneyed patrons of
science. The mathematical basis for understanding purpose was
worked out by engineers in the 1930s and has been put to work in
everything from cruise control to cruise missiles, but behaviorism held
the inertial weight of methodologies, money, and professional reputations.
Stimuli cause an organism to 'emit' responses according to conditioning
established by rewards and punishments. This view was attractive to managers of a
discontented workforce and commanders of a volunteer army. Control the
rewards and punishments that motivate people, the promise goes, and the
direction of their behavior is in your grasp. Note the plural. The
individual disappeared into a Gaussian distribution. The promise to predict and control behavior has been
inherited by cognitive psychology, which arose with the invention of the
programmable digital computer. ENIAC was announced in 1946 as the first
'electronic brain'. The metaphor of the digital computer has been
irresistible. Cognitive psychology differs from behaviorism by interposing
an information processing device between stimulus and response. Marken
lays out the inadequacies of this view with devastating effect in paper 9,
"You say you had a revolution: Methodological foundations of closed-loop
psychology" (pp. 151-175). Hindsight again and again demonstrates that the
secrets of nature are hidden in plain view. The causes of behavior are not
out among the stimuli of the environment, they are the purposes harbored
within each individual. What counts as a 'stimulus' depends upon what
matters to the individual. How do you identify and study these interior
purposes in a scientific way? The key insight is that we do not control our
behavior. Rather, behavior is variable in just the manner and extent
necessary to make our experience be the way we want it to be. The title of
the locus classicus of this science of psychology is Behavior:
The control of perception, published in 1973 by William T. Powers. "The feeling among ... psychologists seems to be that
simply being aware of the purposeful nature of behavior is a sufficient
basis for saying that one is taking purpose into account in one's
research" (p. 1). But mere hand-waving does not a science make. In these
papers, Marken demonstrates the methodology of Perceptual Control Theory
(PCT) and its ample explanatory fruits. In "Looking at behavior through Control Theory
glasses", the first essay in this collection, Marken reinterprets a number
of phenomena that have previously been given stimulus-response
'explanations', and he does so with reference to online computer
simulations so that the reader can directly experience how the given
behavior results from negative-feedback control. What is a purpose, and how do you identify and
specify one? The fundamental step in PCT research methodology identifies a
variable in the environment whose perceived state matters to the given
subject. This step is called the Test for the Controlled Variable. The
preferred state of a controlled variable, its 'reference condition',
specifies the purpose of the subject with respect to that variable as
perceived in the environment. The reference (or setpoint) is inferred to
be the condition to which the subject restores the controlled variable
when an environmental disturbance affects that condition. Crucially, this cannot be done without taking the
point of view of the subject. What perception is the greylag goose
controlling by her 'egg-rolling' behavior? What perceptions are the
individual birds controlling, such that we outside observers perceive
'flocking behavior'? What perceptions does a baseball player control in
order to catch a fly ball? These and more are explicated in the papers in
the first section, "Looking for the purpose of behavior". The papers in the second section, "Illusions and
confusions", explain how and why well-meaning scientists have continued to
misinterpret behavior for so long. When experimenters expect independent
variables to cause dependent variables in a linear way, they control that
perception as well as they can by averaging results for many instances of
behavior and many behaving individuals. Any statistical results better
than a coin-toss are deemed significant and worthy of publication. The
actual data of individual behavior are discarded after the statistical
analysis. When properly perceived, these data for individuals demonstrate
the stabilization of selected variables by variable behavioral means
resisting environmental disturbances. The major disturbance in a
conditioning experiment is kept virtually out of sight as the
'establishing condition' for the experiment, e.g. starving an animal to
85% of its body weight so that it will do whatever it takes to get some
food. Essay 8, "Control theory for whom?", is a review of a
textbook, Control Theory for Humans,
in which two control systems engineers aim to explain control theory to
behavioral scientists. Although the technical presentation is excellent,
the authors fail to address the perceptions that an experimental
psychologist (or sociologist, linguist, etc.) must recognize and control.
This is because control systems engineers do not understand control
systems from the inside out. What I mean by this is that they naturally
assume the point of view of an engineer operating a system and analyzing
its performance. The engineer knows in advance what variables are to be
controlled, and the engineer reaches into the system and adjusts the
reference levels for those variables. An experimental psychologist can do
neither. The controlled variables and their reference levels must be
experimentally inferred from the Test for Controlled Variables. This methodological revolution is the subject of the
last major section. Paper 9 was mentioned above. Paper 10, "Methods,
models and revolutions", alludes to a shift in the behavioral sciences
from statistical methods to the testing of models. However, the models
being constructed and tested are still derived from the same old IV-DV
methodology. The building and testing of models is fundamental to PCT.
This paper succinctly delineates the methodological revolution that is
necessary for the construction of adequate models of behavior. Until a
working, generative computer model can replicate the measured behavior of
an individual with greater than 95% fidelity (preferably greater than
99%), with the deviations from perfect control that are seen in the
individual's actual performance, it is not ready for publication. This is
how to raise the so-called 'soft' sciences above standards of acceptance
that would be laughable in the 'hard' sciences, up to a level on a par with
physics and chemistry. As the older generation, deeply committed to the
illusions of IV-DV methodologies, retire and die, science will progress
and supplant currently received opinion, just as heliocentric astronomy eventually
replaced the mathematically sophisticated epicycles of Ptolemy. The final
chapter of this excellent third collection of Marken's publications is an
imagined 50-year retrospective from the year 2053. Reading this book could
be part of your participation in that progression from illusion to
explanation.
|
|
This page last updated on August 2, 2019 |
