[From Bruce Abbott (970317.1720 EST)]
Well, I'm back from Spring Break -- had a great vacation, only to find
better than 110 CSGnet messages awaiting me. Rather than chime in on any of
the current topics, I have chosen to return to an earlier one, having given
it some thought in the interim. I don't expect to win any arguments, just
state my view as I see it.
The Dutch astronomer Tycho Brahe collected some of the most accurate data on
the apparent motions of the planets then available, prior to the
introduction of the telescope; these were used by Johannes Kepler, who was
able to account for these data by assuming that the planets move about the
sun in elliptical orbits, such that each planet sweeps out equal areas in
equal periods, etc. Although Kepler's laws of planetary motion were
reasonably accurate descriptions of planetary motion, they failed to explain
_why_ the planets moved according to those laws rather than others. That
explanation emerged from the work of Isaac Newton, who posited a system of
mechanics in which concepts such as mass, gravitational acceleration, and
inertia were related by more fundamental laws, whose application to the
specific case of a planet orbiting the sun gave rise to a predicted
planetary motion that followed Kepler's laws. Newton's mechanics also
revealed previously unsuspected deep relationships among apparently
unrelated phenomena such as the motions of planets and pendulums, and in
circumstances where assumptions held reasonably well, allowed an
unparalleled accuracy in quantitative prediction.
So what is the basic difference between Newton's conception and Kepler's? I
would offer that it is the difference between describing a system of
interacting parts whose behavior emerges naturally from the interaction
(assuming given starting conditions) and describing the behavior of the
system, as seen by the external observer, under whatever conditions exist at
the time of observation.
Obviously the Newtonian approach is to be preferred to the Keplerian for its
power and breadth. However, even today there are conditions too complex for
Newtonian mechanics to handle; real pendulums have frictional losses, real
planets are not point masses and are gravitationally attracted not only to
the sun but to each other, and so on; so for some complex cases getting a
reasonably accurate prediction may demand reversion to an empricial
approach. That is why, for example, there are still wind tunnels.
Psychology is still mostly operating within the Keplerian mode, developing
limited models that describe behavior from the external viewpoint as
determined under restricted experimental conditions. PCT offers a Newtonian
approach but as thus far tested can offer only restricted models of limited
components of the behavioral system, such as those intermediate-to-low-level
components involved in tracking. In those areas of application to which it
has been successfully applied, it is clearly superior to those other,
descriptive (Keplerian) approaches. Yet there are many areas of research
for which reasonably reliable descriptive laws have been discovered, for
which as yet there is no alternative in the form of a PCT model that would
explain why these regularities hold (when they do), or under what conditions
one should expect them to break down.
I agree with Bruce Gregory that too much research in psychology is more like
alchemy than chemistry, and that the findings of such research are likely to
retain as much relevance to a future psychology as ancient alchemal formulae
do today. However, it has been and continues to be my position that
psychology has also produced its share of Keplerian descriptions, which in a
future psychology will be explained by more fundamental principles but not
displaced by them.
These two ways at looking at a system -- from the outside and from the
inside -- do not generate opposing theories of behavior but instead offer
different modes of description. Newton's mechanics did not invalidate
Galileo's conclusion that heavy projectiles follow a parabolic path
(approximately) through the atmosphere, nor will PCT invalidate Herrnstein's
conclusion (for example) that pigeons match (approximately) relative rate of
responding to relative rate of reinforcement. In both cases, the former, if
they are correct and properly applied to the given specific case, should
explain the later.
At the present stage of PCT development, we have a basic mechanism and a
proposal of how that mechanism may be arranged to form the human behavioral
system (HPCT). (This is equivalent to Newtonian mechanics versus a proposed
application to explain the motions of, say, the hands of a clock.) However,
until a more detailed knowledge of the actual structure of the human system
is worked out, it may prove impossible to apply control-system principles in
such a way as to adequately account for many of the known regularities of
behavior which have been formalized to date within current psychology; until
then those formalizations may remain the best we can do, and when properly
applied within those conditions in which these "laws of behavior" hold,
provide a reasonable guide to informed action.
As an illustration, consider the oft-mentioned rubberband demo, touted as a
clear example of the "behavioral illusion." It is an illusion only to the
extent that one believes it to reveal a direct, unidirectional causal
linkage (Aristotle's "efficient" cause) between the tug at the end of one
rubberband and the compensatory movement at the opposite end of the other
rubberband. However, there is no illusion in the statement that varying the
position of the far end of one rubberband produces a more-or-less
mirror-image movement of the far end of the other rubberband under the
conditions in which this behavioral illusion is observed, the axis of
reflection being located approximately at the position of the coin. That
description is accurate so far as it goes. It does not suggest a mechanism
but only describes what leads to what. Similarly, the pattern of behavior
revealed by the cumulative record demonstrates clear regularities associated
with particular schedules of reinforcement (e.g., the pause-and-run
staircase pattern charactistic of FR schedule performance); this description
may be elaborated by showing how variables such as amount of food delivered
at schedule completion or level of deprivation relate to the average rate of
responding and pause-length. But these are simply descriptions of observed
regularities, not competing mechanistic models of behavior that would be
swept away by some Kuhnian paradigm shift. Belief in an underlying
unidirectional causal model to explain these regularities would have to go,
but not the regularities themselves. Newton's mechanics, not Kepler's laws,
would have been discarded if Newton's famous equations had predicted oval
planetary orbits, or that the planets sweep out equal square roots of areas
in equal times. This is why I continue to insist that all of psychological
science is not about to go out the window the day it is generally accepted
that behavior is the control of perception.
It doesn't really matter whether my view on this or the opposing one is
correct, insofar as doing the research in concerned. Once you have adopted
the Newtonian mode, you organize your research accordingly. Where it does
matter, I believe, is in how those who adher to the PCT interpetation of
behavior will be regarded by the established scientific community. PCT will
have a better chance of being accepted in the near future if it is offered
as a theory that promises to organize and explain the many existing
empirical "laws" of behavior as opposed to being offered as a view that
promises to dump all existing psychological knowledge and ongoing research
into the trash. The fact that so many of the proponents of PCT wish to
pursue the latter strategy is, I fear, not only misguided but a genuine tragedy.
Regards,
Bruce