[From Bruce Nevin (2000.04.21 18:16 EDT)]
In a message dated 4/19/2001 11:13:57 AM Eastern Daylight Time,
bnevin@CISCO.COM writes:
<< I believe our difficulties with CSG-net would be eased by more desire to
understand and less desire to persuade or, as Dick says, to be King on the
Mountain. .... >>
I thought the original purpose of the CSGNET was to provide information and
report on current research about PCT and HPCT rather than to criticize people
for not understanding PCT or HPCT. That rarely happened on CSGNET.
Another stated purpose was to publicize PCT and draw more people to it. Providing information has frequently taken the form of correction when things said were inconsistent with PCT. This can be perceived as criticizing people. Given a purpose of providing information and reporting on research, and an influx of people just learning about PCT, the first hurdle for a newcomer is to establish that the information or research they are reporting is within the scope of PCT. For example, there has been some contention about Experimental Analysis of Behavior (EAB), the basic research component of "behavior analysis" (whose clinical application is behavior modification).
Contention is usually not about specific experiments and models, however, but about extrapolations from the theory to any of the many domains where we are (some of us more than others) confident that it applies, but where specific experiments and models of recorded behavior do not yet exist. Why are these unsupported extrapolations so frequently posted, and why are they so contentious?
Some of us take PCT as the basis for a philosophy of life and guide to conduct. This gives debate the existential bite of personal investment. At the same time, it can make it harder to distinguish between informed extrapolation and imaginative speculation. I doubt that there is enough experiment and modelling of social psychology as perceptual control to support a philosophy of life based in PCT; there evidently is not enough demonstrable science to identify what is unreasonable and rule it out so clearly that a reasonable person must say "Yes, given those premisses, this must be so," so that to continue in dispute they must identify specific PCT premisses to challenge. These extrapolations to relationships, politics, etc. are just too damned fuzzy.
We may become deeply invested we may be in speculative PCT interpretations of events and relations in our daily lives. Perhaps they are felt as some kind of test and validation of the theory. But these extrapolations from basic principles of PCT can amount to no more than an accumulation of anecdotal evidence. Anecdotal evidence may seem personally convincing but has no scientific validity beyond suggesting possible lines of research.
Folks with less grounding in the valid science of PCT can be attracted to such discussions and follow along only on this superficial level of making PCT reinterpretations of everyday experience. PCT has been characterized by some detractors as cult-like. We have to acknowledge, I think, that in the characteristics that I have been talking about are the brand-markings, if not the actual makings, of true believers in a cult.
At this level of anecdotal discourse about extrapolations of the theory, absent specifically relevant experiments and models the only remaining criteria for validity are verbal. From this stems a puritannical policing of discourse that starts many a conflict. Unfortunately, this is yet another trait found in cults.
We want to attract people to PCT. Often what attracts people is the suggestion that PCT applies to a field that already interests them. They attempt to extrapolate PCT to their familiar experience in their field. We say "no, no, you have to recreate your field from the ground up." Their extrapolation was no weaker than, say, someone's recent disquisition on Northern Ireland or Milosevic or voters in Florida. It begins to appear to them that you can talk about anything so long as you use the correct vocabulary in the correct ways.
What might be helpful here would be some serious discussion of methodology. Yes, we know that the basic method is
(1) Test to identify controlled variables.
(2) Record CV values, behavioral outputs affecting the CV, and disturbances affecting the CV.
(3) Create a model that maintains the CV at like values in the face of those disturbances by means of very nearly identical behavioral outputs.
Examples of putting this into practice almost all abstract to a lab situation like controlling a cursor on a screen. The benefit of this is that the experimenter can be sure of the CV and can control all disturbances very tightly. The cost is the artificiality of the setup. These highly constrained scenarios prove the fundamental validity of the theory again and again. It is the application of the theory to less artificial situations that is problematic, and that invites all the extrapolations that are so contentious. What we need is more discussion and exemplification of how to actually do this in various realistic domains. Rick's baseball-catching demo is excellent, but (1) and (2) were done by others who are not among us to talk about the practical issues involved. Even a reprint of the Methods section of their report would be helpful, associated with the demo on the web, but that cannot substitute for conversation.
I suppose I feel this acutely because I have no training or significant experience in experimental design. I am not sure this is entirely a bad thing: I haven't ingested any dogmas of behaviorism. But not only would more focus on methods in naturalistic research fill in education gaps for me (and I suspect for others as well), it would direct our discussions in inherently more productive and less disputatious directions. Instead of focusing on principles with verbal purity as the criterion of acceptance we would focus on methods and practices with consistency with principles as the criterion of acceptance. We could set as a ground rule that any extrapolation to naturalistic settings must be in the form of talk about actual research, or at least about how to do PCT research in the topic domain.
This would serve all three charter purposes of CSG-net, providing information, reporting on current research, and proseletyzing. It implicitly incorporates the ground rule suggested by Hank Folsom (2001.04.17.0900), that PCT must only be discussed on its own merits, and not by comparison to any other theory.
Ground rules are a good idea. Hank also suggested (2001.04.20.0700PDT) a ground rule that all posters must spend several hours composing and thinking through each post. To enforce this would be very difficult (many email fora have a moderator; I doubt any of us wants to assume that role but it is possible). But short of that it would be relatively much easier to establish a custom of comment when someone shoots from the hip, noting that the first post and its reply were much less than 24 hours apart. Of course, the relevant baseline is not the timestamp of the first message vs. that of the reply, but the time that the first message was actually read by the replier, but the quality of mercilessness is generally as obvious as a hailstorm.
I realize that this message is not itself an example of what I am recommending (except for the time taken to think it through). It is not about control so much as about controlled perceptions and reference levels for them. I realize too that there's a bunch of "saying" here, and to find out if I am "telling" you anything it would be most helpful if you were to say what you think it is that I am saying. Or of course if you act in ways that I perceive as agreeing with what I have said, that would convince me that I have communicated something useful.
Hank, I haven't forgotten your proposal to apply PCT to CSG-net discourse. Its just that so far there has been little made explicit about mutually sustained feedback loops between autonomous control systems (communication). Maybe this is a start. What do you think?
Bruce Nevin
···
At 20:31 04/19/2001 -0400, Charles W. Tucker wrote:
