[From Bruce Abbott (960406.1255 EST)]
Bill Powers (960405.1045 MST) --
This is it. People are looking for principles like conservation of
energy and mass, so they keep proposing ideas about what might "really"
be going on. What's wrong with this is looking for the generalization
first, instead of being led to it by an understanding of the more
detailed processes.
To understand why researchers began to ask questions like whether pigeons
optimize or meliorate (the latter is close to Simon's "satisfice"), you have
to know something about the sort of data they were confronted with, and also
about the linkage of these ideas to those from other fields (economics in
particular, and theories of human choice [e.g., Luce]). One area in which
the data were raising questions was in avoidance conditioning. Research by
Sidman, by Dinsmoor, and by Herrnstein and Hineline strongly suggested that
rats could somehow integrate events over time and respond on the basis of
average shock frequency reduction, even when some responses were followed
immediately by shock. Data from concurrent and concurrent chains schedules
indicated that the proportion of responding emitted on one of two keys
tended to match the relative rate of reinforcement associated with the
schedule programmed on that key (for VI schedules, at least), again
suggesting that animals (pigeons, in this case) were sensitive to overall
rates of reinforcement and perhaps responding on the basis of this more
global characteristic rather than being driven directly by the individual
reinforcer deliveries. Other theorists suggested ways in which the apparent
influence of rates could be explained in terms of the individual events and
their temporal characteristics, leading to a still-continuing debate
concerning whether behavior on these schedules is best described in terms of
a "molecular" (moment-by-moment) or "molar" (long-term averages) conception,
and this debate led to questions about what the animals might be doing or
perceiving, moment by moment, and whether the result could be described in
such terms as "momentary maximization," a view which could account for such
things global outcomes as matching in terms of the animal's "decisions" as
determined by its current view of the options and their liklihoods of
success. Such thinking made explicit a possible linkage between what was
being observed in the operant chamber and theoretical formulations produced
by psychologists concerned with human judgment and decision-making, who in
turn had patterned many of their ideas on economic theories involving such
concepts as utility and probability of success.
But this principle doesn't come out of more basic observations; it's
proposed in place of basic observations. It's truly a "perspective," a
way of looking at something. Unfortunately, once you accept the
principle, you have to start supplying facts to back it up, and if the
facts aren't available you have to invent them.
I really must disagree with this way of characterizing how investigators
have behaved. Although I believe you don't really intend it this way, it
sounds as though you are accusing researchers of making up their data, in
order to support the theory. What I think you mean is that, having adopted
a particular interpretation, researchers have tended to ascribe properties
to things (e.g., reinforcers) because these properties are required by the
theory, rather than empirically determining what the facts may actually be.
It this close?
The notion that _rates_ of events like reinforcer delivery might be directly
perceived and affect behavior is an almost shocking departure from
reinforcement theory as it stood until the early 1960s; it is a view that is
more congenial to the PCT way of thinking (i.e., continuous variables) than
the old contiguity-based notion. The theoretical speculation about
matching, melioration, and the debate between molar and molecular theories
arose in the context of this advance. It may not have solved the problem,
but in my view it was a step in the right direction, toward dynamics and,
perhaps, preparing the way for the ultimate acceptance of PCT.
Regards,
Bruce