[Martin Taylor 2018.07.24.16.27]
There seems to be some uncertainty on CSGnet, let alone in the wider
world, about what constitutes a behavioural illusion and what
constitutes a side-effect. Here’s my take on the question.
The earliest reference to a behavioural illusion that I can find
occurs in Bill Powers’s response to comments on his 1973 Science
paper (p 84 in LCS I). Here’s the relevant paragraph, in which the
reference is to the Science paper. (I did not find the word
“illusion” in the Science paper itself, but maybe it is there and I
just missed it). The emphasis on it in the quote is mine.
-------quote------
In the section on controlled quantities in my article, there appears
an approximation, g(d) ≈ -h(o) , which says that the
cause-effect relationships that can be observed between stimulus
events and consequences of nervous system outputs—responsesâ—are
expressible wholly in terms of the physics of the local environment,
containing almost no information about the behaving system at all. I
see no way in which behaviorism can survive a full understanding of
the derivation and significance of this harmless expression. If
control-system theory does indeed describe correctly the
relationship between organisms and their environments, behaviourism
has been in the grip of a powerful illusion since its
conceptual bases were laid.
-----end quote------
In this quote, Powers argues that the behavioural illusion is a
false interpretation by behaviourists of their observations of the
relationship between a “stimulus” and a “response”. The illusion is
not inherent in the experimental situation, nor in the observation
of the stimulus and the response.
A simple illustration of what Powers was saying (and many times
repeated and expanded upon in the following years) is provided by a
simple lever whose angle relative to the horizontal is controlled by
a subject pressing on one end while a disturbance is applied to the
other (the difference between the figures is only that in the lower
figure the subject presses on a balloon attached to the lever rather
than directly on the lever).
In both figures, the person is controlling for keeping the lever arm
horizontal. To do this the force applied by the person depends not
only on the stimulus force S, but also on the ratio L between the
lengths of the lever arms either side of the fulcrum. The response
force R required to keep the lever just level is R = S*L. From a
simple-minded S-R analysis, the subject’s internal processes must
have been trained to produce a force R when a stimulus S is
provided.
From the viewpoint of PCT, the fact that the subject produces
exactly R rather than some other force is due entirely to the
physics of the lever, and provides no information about how the
subject’s internal processes result in exactly that output force.
All that is needed is that those processes allow the subject to
control the lever angle. The behavioural illusion is in the mind of
the analyst who observes S and R, and deduces from those values
something about the internal processes of the subject. It is not in
the lever or the relationship between the two forces. They are
observations, not illusions.
In the lower figure, the subject presses on a balloon rather than on
the lever directly. Newton’s laws ensure that the force applied to
the lever by the subject is exactly the same whether or not the
balloon intervenes between the subject’s hand and the lever. But the
observer now can see something new happening. The balloon loses
height and bulges in width when the subject presses down on it. That
is a side-effect of the subject controlling a perception of the
angle of the level with respect to the horizontal.
If the observer could not see the lever, might it not appear as
though the subject were controlling a perception of the width of the
balloon? That would be an illusion of a different kind, not a
behavioural illusion. The width of the balloon would be responsive
to the disturbing force if the subject ceased controlling the lever
angle while keeping a constant downward force on the balloon, and
the force the subject applied to the balloon when controlling would
vary exactly as it does when the subject controls the lever angle
rather than the balloon width. Even if the observer could see the
lever and the balloon, how could the observer tell which is
controlled variable and which is side-effect? The
observer/experimenter has no way of knowing, but they cannot both be
controlled independently. (Of course, the observer as an
experimenter could use the TCV component in which the subject’s
ability to perceive one or other of the candidate controlled
variables, but that is a different issue.)
Again, the physical structure determines the relationships among the
observations, provided that the subject controls one or other of the
variables – balloon width or lever angle. One is a side-effect, but
there is no behavioural illusion in the observations. The
behavioural illusion, if any, would be in the mind of someone who
sees a “stimulus” of applied force to produce by way of analyzable
processing in the subject a “response” of a certain balloon width.
Side-effects are unintended effects on the environment that are
consequences of actions performed to control some perception;
behavioural illusions are in the minds of those who interpret as
consequences of internal processing relationships that actually
depend on properties of the environment.
Martin