[From Rick Marken (940929.1145)]
Paul George (940926 10:30) posts about "An interesting simulation that might
be recastable in formal PCT terms." I think that what would be needed to
"recast" this simulation in PCT terms could help clarify the difference
between a model in the PCT sense and the "model" described in this article.
The "Resistance to Change" model by Peter von Stackelberg makes the following
assumptions:
1. Humans build mental maps which are, in effect, their reality.
2. Changing a mental map causes a negative feeling best described as
discomfort or pain.
3. The greater the pain, the greater the resistance to making mental map
changes.
4. The greater the resistance, the fewer the number of mental map changes.
5. Humans experience dissonance when their mental maps are not synchronized
with their external environment.
6. Changes in the external environment increase the dissonance.
7. Increasing dissonance reduces resistance to changing the mental map.
8. Changing a mental map decreases dissonance as the internal and external
realities are brought into closer synchronization.
Besides the fact that these assumptions are very vague (in 2, for example,
what "changes" a mental map?) they are not tied to observation in any obvious
way. There is no description of the variables that consitute the phenomenon
that is explained by the model nor is there any description of what
constitutes the behavior of the model itself. The only assumptions that
seem to relate the model to even potentially observable phenomena are (5) and
(6) which relate the model to the environment. But there is nothing about
what this environment is or how it is related to model constructs (how does
one model "synchronization" between environment and mental map, for
example; I can think of many ways but these assumptions sure don't limit
the possibilities).
I could see many different ways in which these assumptions could be
implemented as a computer simulation; and all these simulations would
produce all kinds of behavior that one might see as vaguely consistent with
all kinds of things that people do. But these is no clear description of how
the model relates to its external environment and, even more importantly,
there is no clear description of what data -- quantitative data -- the model
is designed to explain.
This kind of modelling seems to me to be the high tech equivalent of reading
Tarot cards. It is about as far from science as one can get. Becuase it can
be implemented as a dynamic simulation it might seem very "scientific" but,
as it sits right now, it is only a little less worthless than just plain
words. In order to be of any value, the people building this model must show
that the behavior of the model has a very high correlation (.95 minimum) with
some aspect of the behavior of the systems being modelled (people) over a
significant period of time. They must also show how their model of people
relates to existing physical models of the environment. One nice things about
PCT is that it is a model that is consistent with very successful models of
other aspects of our experience besides purposeful behavior.
Best
Rick