Conventional vs. PCT research: Summary

[From Bruce Abbott (951022.1150 EST)]

It seems to me that the debate over the value of conventional research
findings to PCT researchers has gone on long enough (or more likely, too
long); in my judgment it is unlikely that further debate on the issue will
be productive. Whether it has changed any opinions I do not know, but I
believe that the debate itself was valuable nevertheless. I don't recall
who said it, but there is a famous quote that states, "it is better to
debate an issue without settling it than to settle an issue without debating
it." At least now we've all had the chance to hear the arguments on both
sides and reconsider our opinions in their light.

Before closing this discussion, I'd like to summarize what I perceive to be
the basic position on each side. If someone else has a different
perception, feel free to offer your correction.

My view has been that _some_ (certainly not all!) previous and current
research using what has been termed here "conventional methodology" has
yielded observations and conclusions about human and animal behavior, and
the brain structures subserving those behaviors, that provide a secure
foundation on which to consruct future research programs designed to uncover
the detailed functional and structural organizations of the brain's
perceptual control systems. In my view the best source of information of
this sort comes from ethological studies, which have revealed highly
organized, species-typical patterns of behavior, perceptual and hormonal
inputs to those systems through which these patterns emerge, and the
functions those patterns of behavior serve in the lives of individual
members of the species. It seems to me but a small step from those
observations to PCT research designed to identify the specific control
systems underlying those patterns. In addition, I hold that work
investigating which brain systems participate in and organize those patterns
will also be of continuing value to those who are trying to identify the
physical mechanisms within the brain that realize the functional
control-systems organization uncovered by PCT research.

Science is a cumulative process. The great value to science of theories
such as PCT is their ability to organize and explain the observed phenomena
within their domain. A theory's _ultimate_ acceptance by the scientific
community depends on this ability. To claim at the outset that observations
and findings obtained in previous research undertaken from a different
theoretical perspective are irrelevant seems to me to be fundamentally
anti-scientific and highly likely to impede rather than foster research
aimed at testing and developing perceptual control theory. Thus, by
ignoring previous work _in areas of relevance to PCT and its predicitons_,
we not only waste valuable time "reinventing the wheel," but also isolate
ourselves from the broader scientific community. In my view, this is bad
strategy.

On the other side of this debate are, as I understand them, some excellent
points to be made. As Bill Powers has argued, PCT is still in its infancy,
as very little PCT-style research has yet been conducted. Rather than
attempting to tackle phenomena identified by previous researchers, it would
be better at this stage to begin by carefully and thoroughly examining the
lowest-level control systems. Once these are well-characterized, one would
then move up to the next level and do the same there. Eventually, one will
get to the point where the system's organization is fairly well understood.
The now highly developed model can then be applied to explain any previously
known behavioral phenomena, many of which will then be shown to be mere
irrelevant side-effects of control and thus perhaps not worthy of further
attention.

Another excellent point (mainly advanced by Rick Marken) is that much
current research employs methodology that is inappropriate to the study of
control, particularly group-based methods using inferential statistical
analysis. The problem with such studies is that they can only establish
facts about average behavior, and only to a given level of statistical
certainty, and say little or nothing about the factors at work in a given
control system. They are essentially useless for determining the perceptual
input functions of given individuals, the particular inputs being controlled
(which may differ across individuals), individual system gains, and so on.
Such facts cannot be used to construct a precise, generative model of an
individual control system, nor do they provide the kind of data against
which the performance of such a model can be checked.

I see merit in the arguments on both sides of this debate; our differences,
it seems to me, lie mainly in where we fall along a continuum. Rick Marken
believes that no previous research is of value; I argree concerning some
research findings but still see great value in others. Bill Powers believes
that PCT should build up a solid foundation on its own terms, independent of
previous work, on the grounds that it will all have to be done over again
anyway, using PCT methods, and that once the control structures have been
identified, previous findings will then stand or fall in the light of PCT.
I agree that PCT research should be done on its own terms, but believe that
much wasted effort can be avoided by building on relevant previous work and
that studying phenomena previously identified in other research, by
demonstrating PCT's superior ability to provide a coherent account, will
speed acceptance of PCT by the greater scientific community. I am well
aware that Bill has already tried this approach in several areas and has
encountered only hostility and rejection in so doing, but against this
contrary evidence I remain foolishly optimistic that this approach will
eventually prevail.

Regards,

Bruce