[From Bill Powers (951122.0830 MST)]
EAB, PCT, and Newtom
···
---------------------------------
The terms "cause" and "effect" are no longer widely used in
science. They have been associated with so many theories of the
structure and operation of the universe that they mean more than
scientists want to say. The terms which replace them, however,
refer to the same factual core. A "cause" becomes a "change in an
independent variable" and an effect a "change in a dependent
variable." The old "cause-and-effect connection" becomes a
"functional relationship." The new terms do not suggest _how_ a
cause causes its effect; they merely assert that different events
tend to occur together in a certain order. This is important, but
it is not crucial. There is no particular danger in using "cause"
and "effect" in an informal discussion if we are always ready to
substitute their more exact counterparts.
We are concerned, then, with the causes of human behavior. We want
to know why men behave as they do. Any condition or event which can
be shown to have an effect upon behavior must be taken into
account. By discovering and analyzing these causes we can predict
behavior; to the extent that we can manipulate them, we can control
behavior. [Skinner, B.F.(1953) , _Science and Human Behavior_. Ch.
3, Why Organisms Behave, p. 23].
----------------------------------
Let's consider the behavior of a falling object. What causes this
behavior? Before Newton, two answers had been offered. (1) The earth
attracts all objects to itself, as if sucking or pulling them toward the
ground. (2) The objects themselves have an affinity to earth, and seek
the greatest proximity to it that is possible. So in the first case, the
earth's pull causes the object to fall, and in the second, the object's
own efforts to reach the earth cause it to fall.
After Newton, neither answer was accepted any more. Objects fall because
they fall. More precisely, the acceleration of one object toward another
is expressed as a function of the distance between the objects and the
mass of the other object. This acceleration is not a consequence of a
prior event; it is not an event that depends on a prior condition or
event. Instead, the equations express an ongoing relationship among
accelerations, masses, and distances in which neither causes nor effects
can be uniquely determined. It is a present-time relationship.
Even in the simplest of Newton's equations, F = MA, there is no
dependent or independent variable, no cause or effect. Or, to put that
differently, if any of these variables is manipulated and another of
them is fixed, the remaining one becomes a dependent variable. With
fixed mass, the usual situation, you can create a specific force and
measure the resulting acceleration, or produce a specific acceleration
and measure the resulting force. In a rocket ship, you can vary mass
with acceleration held constant and measure the resulting force, or vary
mass with force held constant and measure the resulting acceleration.
There are no fixed roles that make any variable inherently an
independent variable or a dependent variable, a cause or an effect.
Suppose we set up a test object and arrange to apply a force to it as an
independent variable. We assume a constant mass, and measure the
acceleration as a dependent variable. Let's specify that the applied
force will start at zero, with the object (having a mass of 1 kilogram)
at rest relative to our inertial frame. The force will be increased
abruptly to 1 newton for a period of 0.01 sec and then reduced abruptly
to zero again.
What we have done is to create an "event." An event is a pattern of
values of a variable that has a beginning, a middle, and an end. Here it
is a force that is zero before the event, rises very quickly to 1
newton, remains at 1 newton for one one-hundredth of a second, and then
falls very quickly back to zero again. We can call this event a force-
event and measure it in units of newton-seconds (x newtons of force
sustained for y seconds).
After the event, we find that the object is now moving relative to the
frame of reference at 1 cm per second. So, in the old style of thinking,
we can say that creating an impulse-event causes the object's velocity
to change. The cause was the force-impulse, the effect the change in
velocity, a velocity-event.
If we go back to the basic relationships, however, we will see something
that remained exactly the same throughout the event: the relationship F
= MA, or solving for the acceleration, A = F/M. At every instant, the
acceleration of the object was exactly proportional to the applied
force. This relationship never changed at any time before, during, or
after the event. The final velocity is explained by the fact that a 1-
newton force was sustained for 0.01 sec. At every infinitestimal instant
during that hundredth of a second, the velocity changed by F/M times the
duration of the infinitesimal instant. The sum of all these velocity
changes was the final velocity after the force had been completely
removed. It was not the "event" that caused the change in velocity, but
the sum of all the accelerations that existed during the event.
Cause and effect are often defined in terms of "events." First there is
the causal event, which comes first in time. Then there is the effect-
event, which is the later result of the cause. When such a sequence is
observed, it is common to treat the mere _fact_ of the causal event as
having an influence on the mere _fact_ of the effect. This leads to
pseudo-explanations of effects that say "The cause belongs to a class of
events which generally produces a class of other events, and the effect
belongs to the latter class." In other words, causes belonging to the
one class have been observed to produce effects belonging to the other
class. This in turn leads to an attempt to find something that the
events of each class have in common. So an event in the class called
"stressful" causes a reaction in the class called "anxiety." Or an event
of the class called "reinforcers" causes an event of the class called
"increase in probability of a response."
When events are treated in this way, they are quantified by merely
counting them or counting the rate of their occurrances. In logical
terms, an event either occurs or it does not occur; we say IF event A
has occurred, THEN event B occurs. This is equivalent to saying that
event A implies event B: it is not the case that A occurs and B does not
occur. This is the logic of saying that event A causes event B. So a
"force-event" "causes" a "velocity-change event."
Let's go back to F = MA. This says that if we create a certain
acceleration of a fixed mass, a force will result. Suppose we arrange
for a carriage that can be moved to accelerate the object at any rate,
and measure the force existing between the carriage and the object. The
acceleration starts at zero, rises abruptly to 1 meter per second per
second, remains at that value for 0.01 sec., and then falls abruptly to
zero. The velocity will necessarily change from 0 to 1 cm per second. So
we have created a "velocity event" defined as a change in velocity. We
discover that this results in a force that rises abruptly to 1 newton,
remains at that level for 0.01 second, and falls abruptly to zero. So we
can now say that a "velocity-change event" "causes" a "force-event". We
have, simply by interchanging our definitions of independent and
dependent variables, reversed the direction of causation.
But in both cases, whichever way we see causation running, a single
relationship remains true at every instant, before, during, and after
the events. Moreover it is the SAME relationship in either case: F = MA,
or A = F/M which is the identical relationship.
Events are an artifact of human perception. We create them when we
manipulate variables in some simple fixed pattern through time, with a
beginning, a middle, and an end. Events by themselves -- the patterns
that we see --have no effect on anything but our perceptions; the real
effects are the direct effects of the variables which are being changed
to create the appearance of an event. Which variable is seen as cause
and which as effect depends on which variable we manipulate, which we
hold constant, and which we observe. The true law of causation, the true
explanation of the result, is to be found in the continuing
relationships among variables that remain in effect no matter which
variables are being manipulated and which are being observed. Whether we
manipulate force, mass, or acceleration, the true explanation of what
happens is in the relationship F = MA, in whatever equivalent form it is
presented.
-------------------------------
Now, consider the concept of an independent variable, a "cause." There
are, in fact, no independent variables -- that is, variables which
assume an arbitrary value with no relationship to anything else. When we
say "let the force rise to 1 newton, remain there for 0.01 sec, and
return to zero thereafter," we are postulating something that can happen
in only one way: some other variables must have changed in such a way as
to make the "independent" variable follow that pattern of values.
When independent variables are manipulated for experimental purposes,
either they change naturally as other environmental variables change, or
the variables are deliberately made to change according to a
predetermined pattern. The latter is almost always the case. What this
means is that some human being must have conceived the pattern in
advance, and then either produced it by his own efforts or built a
machine that was adjusted until its output pattern matched the
predetermined pattern.
The very concept of an experimentally-manipulated independent variable,
therefore, implies a control system in the background, a system which
can act to make a perceived pattern of change match a reference pattern.
This makes the experimenter an integral part of the experiment. It is
the experimenter who picks which variable in a situation is to be
manipulated in a predetermined way and thus made into an independent
variable, and which other variable is to be observed and thus made into
a dependent variable. The experimenter determines what is to be cause
and what is to be effect. Since many different choices can be made, no
variable is inherently a cause and no other variable is inherently an
effect. Instead, the real system under study consists of a set of on-
going present-time relationships which, when properly described, remain
exactly the same no matter what manipulations are carried out.
-----------------------------------
The difference between reinforcement theory and either PCT or Newton's
laws is that reinforcement theory looks for cause-effect relationships,
while both PCT and Newton look for underlying relationships which remain
exactly the same no matter what events are occurring and no matter which
variables are chosen for manipulation. Skinner, in the passage quoted
above, clearly did not understand why cause and effect are not employed
in physics; he clearly takes the pre-Newtonian view of casuality, and
thinks that the only alternative to cause and effect is "independent-
variable, dependent variable" -- essentially the same thing.
-----------------------------------------------------------------------
Best,
Bill P.