[From Kent McClelland (95128.1500 CST)]
Replying (finally) to Bill Powers (951221.1315 MST) and Gary Cziko
Sorry if my inept choice of arguments made it sound like I was advocating a
relativistic view of scientific truth in stressing the social dimensions of
the scientific enterprise. I wouldn't want to be pigeon-holed as a
supporter of naive relativism or "truth by vote", any more than I'd want to
be seen as a naive realist. I agree fully with Bill that just because all
the big scientists say some supposed fact is true, does not make it so.
I guess I look at it this way: Our perceptions of "scientific" reality are
never going to match perfectly the "boss" reality on which those
perceptions are based. If we play the science game by the rules,
however--the rules Bill outlined in his post (951221.1315 MST) which call
for public demonstration by empirical observations and mathematical
reasoning--we have a much better chance than by any alternative method of
minimizing the discrepancy between our perceptions and the reality we hope
they reflect. It would help, of course, if everyone else who called
himself a (social) scientist were playing by those same rules--not always
the case for psychology and the social sciences.
I intended my earlier comments to remind folks that science is a public
game, not one which can be successfully pursued entirely on one's own (a
point which might better have gone without saying!), and of course in now
calling science a "game" I wouldn't want to imply that science isn't
serious. . .
A random thought, which may or may not be related:
I've seen a lot of PCT models with random disturbances, and some with
constant disturbances or ramped disturbances. In all these cases, the
disturbance is some function of time.
I wonder if any PCT modeler has tried constructing disturbances which are
structured or patterned in other ways, for example as some function of the
controlled variable, which would be like having ramps or bumps or dips in
the environmental space in which the control was taking place--some regions
in which control was relatively easy and other places in which control was
harder or more effortful or more likely to result in substantial errors.
Perhaps a model might contain both random disturbances and these structured
Presumably, structures in the physical environment constrain people to
reoranize until their lower-order control loops are pretty much invariant
from person to person. But how does one model these structures in the
environment? The super-simplified one- or two-dimensional environment I've
been using for my models seems to be completely structureless, a space in
which the control system can choose any arbitrary reference value and have
the same chance of successfully controlling at that level as at any other.
I suppose Bill Powers has tackled such questions of environmental structure
in his arm model by accounting for gravity, inertia, etc. Do you do this
by giving a structure to the disturbances or in some other way?
It seems to me that models of learning by random reorganization might be
more interesting if they operated in an environment that had some structure
to it. Would the reorganization, then, eventually reflect the
environmental structure in some way? Perhaps some work along this line
might lead eventually to "pattern recognition" models. Most of our
modelling of perceptual control has focused on motor actions rather than
"perception" in the way other people see it.
I think, too, that a tracking experiment in which people encountered a
structured disturbance (of course invisible on the monitor) might also be
interesting to try. How would people perceive these invisible bumps or
chutes or barriers?
I don't intend for my suggestion to raise once again the old controversy
about whether a control system somehow uses information from the
disturbance, or gains information about it, or whatever it was that Bill
and Martin once argued about on the net. I assume that argument somehow
got resolved. Maybe my question is related to it, however, as well as to
the more abstract question of how boss reality disciplines our perceptions,
scientific and otherwise.