# Standards; Plasticity of top-level Reference

**URL:** <http://discourse.iapct.org/t/standards-plasticity-of-top-level-reference/10756>\
**Category:** CSG1992\
**Created:** [May 22, 1992, 8:23pm UTC](http://discourse.iapct.org/t/standards-plasticity-of-top-level-reference/10756 "1992-05-22T20:23:27Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Jeffrey\_E\_Dooley](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/j/f07891/32.png) [@Jeffrey\_E\_Dooley](http://discourse.iapct.org/u/Jeffrey_E_Dooley)\
**Post date:** [May 22, 1992, 8:23pm UTC](http://discourse.iapct.org/t/standards-plasticity-of-top-level-reference/10756/1 "1992-05-22T20:23:27Z")

</div>

[Jeff Dooley 920522.1000]

(Bill Powers 920514b)

> The only reference signals (and perceptions) that can't be  
> changed freely as required by higher levels are systems  
> concepts.

As you point out, output of higher-order control systems  
enables the plasticity of the lower-order system reference  
values. But then there is an apparent problem, illustrated in  
your comment above, which stems from our apparent inability to  
go up a level so that systems concepts can have adaptable  
reference signals too. If a world-view can be considered a  
systems concept (in the sense that it can organize  
principles), then how can we explain the possibility that a  
"world-view" can change (Ptolemy vs. copernicus, etc.).

Difficulties like these have surfaced in philosophy of science  
recently surrounding attempts to provide a normative  
epistemology within the scope of a purely scientific  
(naturalistic) study of how we come to know anything. This  
problem boils down to how we can derive norms or prescriptive  
standards from an apparently descriptive account of  
epistemology.

I think that some of the work of Larry Laudan may help clarify  
these questions as they may pertain to system concept reference  
adjustment.

In Laudan's \_Science and Values\_, (U. Cal. Press, 1984) he  
proposes a circular causal loop in which AIMS, THEORIES, and  
METHODS are all recursively interdependent and plastic as a  
function of incremental changes among one another over time  
and practice. An important point in Laudan's model (he calls  
it "the reticulated model of scientific rationality") is that  
the thing is grounded in practice. One way this grounding is  
achieved is through the actual realizability of aims  
through methods. Methods are in turn constrained by  
theories while the theories are justified by the methods.  
The thing is entirely circular and intended to  
self-correct over time as each vertex of the triangle impinges  
upon each other one. New perceptions are introduced into the  
system as practice continues and are either harmonious or  
anomalous. These perceptions occasion either assimilation of  
the system concept (world view) in case they are harmonious  
or accommodation of the world view if they are anomalous.  
(They may also be irrelevant and have no impact). Though  
Laudan doesn't invoke any explicitly control-theory language,  
it seems that his model could be described as an n-order  
control system for the derivation of scientific norms. It  
also is an explanation for the rationality of scientific  
change and the rational role of norms and values fueling such  
change.

My question here is simply: can we let systems concepts have  
plastic reference values as a function of all the other-order  
reference changes at any point on the analog continuum of  
experience? This sort of ties the reference hierarchy into a  
closed loop with all the other reference levels at the same  
time.

jeff dooley dooley@well.sf.ca.us

---

<div class="post-metadata">

**Author:** ![Human\_Factors\_Lab](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/h/b3f665/32.png) [@Human\_Factors\_Lab](http://discourse.iapct.org/u/Human_Factors_Lab)\
**Post date:** [May 22, 1992, 9:54pm UTC](http://discourse.iapct.org/t/standards-plasticity-of-top-level-reference/10756/2 "1992-05-22T21:54:20Z")

</div>

[From Rick Marken (920522 14:30)]

Jeff Dooley (920522.1000) says:

> (Bill Powers 920514b)

> > The only reference signals (and perceptions) that can't be  
> > changed freely as required by higher levels are systems  
> > concepts.

> If a world-view can be considered a  
> systems concept (in the sense that it can organize  
> principles), then how can we explain the possibility that a  
> "world-view" can change (Ptolemy vs. copernicus, etc.).

Two possibilities: 1) ptolomy vs copernicus is not a system concept  
level difference. They could just be two different models (programs?) that  
are possible components of a "science" system concept. If one of the  
principles of this concept is "best fit to data" and "simplest" then you  
adopt copernicus. If it's "best bit to data" using existing assumptions"  
(epicycles) then its ptolemy. 2) ptolemy vs copernicus are system  
concepts. Whichever you believe (have a reference for) is then a result of  
previous reorganizations (random selection of system level references  
triggered by intrinsic error). So people would only adopt one if  
doing so happens to produce the least intrinsic error. Intrinsic  
error might result from observed discrepencies between model  
predictions and observations (if the person cared about such things) or  
something else (desire to differ from the church?).

> I think that some of the work of Larry Laudan may help clarify  
> these questions as they may pertain to system concept reference  
> adjustment.

> My question here is simply: can we let systems concepts have  
> plastic reference values as a function of all the other-order  
> reference changes at any point on the analog continuum of  
> experience? This sort of ties the reference hierarchy into a  
> closed loop with all the other reference levels at the same  
> time.

If this is what Laudan is suggesting then I think it might help  
if he hooked up these ideas in a working model. My guess is that  
he could quickly eliminate this hypothesis; if lower level outputs  
influenced the reference signals of systems that determine the reference  
signals of the system that determines the output then it seems like there  
could be problems; like positive feedback type effects. But maybe  
it could work? It just doesn't seem to make much sense in terms of  
the architecture of the PCT model as it currently stands. Perhaps  
if you gave a quantitative analysis or functional diagram of Lauden's  
suggestion it would clarify his point.

By the way, the references for the highest order systems (in the  
PCT model) can be varied -- but not as a means of controlling  
some higher order variable. So changes in the references for the top  
level variabls cannot be SYSTEMATIC (as are the changes in reference for  
the tension in my muscles which varies to maintain the position  
of my outstretched arm even while someone is pressing down on my  
hand with variable force). The top level references are varied  
UNSYSTEMATICALLY by the reorganization system -- and they stop  
changing once intrinsic error disappears. This aspect of the model  
seems to me to give a good account of why people adopt certain system  
concepts rather than others -- its BY CHANCE. People become christians  
because by adopting that reference they have been able to reduce intrinsic  
errors that result from fighting with their family. People become  
dodger fans because, by adopting this reference they reduce the intrinsic  
error that comes from the stares generated by wearing an oakland A's  
jersey at Dodger stadium. People believe in Copernicus because everyone  
else does (it looks to me like the sun goes around the earth -- but I'll  
never tell). I like the fact that system concepts (in control theory)  
are not selected systematically; and this goes for the "control theory  
system concept too (to the extent that it is a system concept). There is,  
really, no good reason to adopt it except that it feels good. Of course,  
part of that good feeling comes from accounting for data to within .001%.  
So it seems like the only scientifically RIGHT thing to do. So why aren't  
social scientists moving to PCT in droves? Because it doesn't feel  
good to them; they are happy with their own system concepts and they  
don't even know that there is a system concept to change (if system  
concepts really are the TOP of the hierarchy and control theory and  
ptolomy and copernicus are system concepts then it seems to me that  
we would not even be able to see that there is a CHOICE. I think a  
case can be made for the fact that "real" system concepts are invisible  
to consciousness -- you can take the point of view of that system concept,  
put you shouldn't be able to see it AS an optional point of view.).

Regards

Rick

> **···**
>
> \*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*
> 
> Richard S. Marken USMail: 10459 Holman Ave  
> The Aerospace Corporation Los Angeles, CA 90024  
> E-mail: marken@aero.org  
> (310) 336-6214 (day)  
> (310) 474-0313 (evening)
> 
> Return-Path: \<@vm.utcs.utoronto.ca:CSG-L@UIUCVMD.BITNET\>  
> Received: from vm.utcs.utoronto.ca by ben.dciem.dnd.ca (4.1/SMI-3.2)  
> &nbsp;&nbsp;id AA09128; Fri, 22 May 92 22:07:21 EDT  
> Message-Id: \<9205230207.AA09128@ben.dciem.dnd.ca\>  
> Received: from UTORONTO.BITNET by vm.utcs.utoronto.ca (IBM VM SMTP V2R1)  
> &nbsp;&nbsp;&nbsp;with BSMTP id 3698; Fri, 22 May 92 22:07:15 EDT  
> Date: Fri, 22 May 1992 19:06:36 PDT  
> Reply-To: "Control Systems Group Network (CSGnet)" \<CSG-L@UIUCVMD.BITNET\>  
> Sender: "Control Systems Group Network (CSGnet)" \<CSG-L@UIUCVMD.BITNET\>  
> From: marken@AERO.ORG  
> Subject: control and consciousness  
> To: Multiple recipients of list CSG-L \<CSG-L@UIUCVMD.BITNET\>  
> Status: RO
> 
> [From Rick Marken (920522 19:00)]
> 
> Well, it looks like the machine is still up so let me  
> make one quick observation. I think there has been some  
> confusion about what control theory says about control  
> and what it says about the processes that might influence  
> control -- ie. consciousness. This confusion becomes  
> particularly acute in the discussion of standards (principles)  
> where there is talk about "setting appropriate standards"  
> and such. When we talk like this, WHO do we imagine to  
> be "setting the appropriate standards"? The hierarchical  
> control model says it is the "higher order systems". The  
> references for principles are AUTOMATICALLY set by the systems  
> controlling system concepts. The references for programs are  
> automatically set by the systems controlling system concepts  
> and/or principles, etc. The point is that all this varying  
> of lower level references to control higher level perceptions  
> is carried out smoothly and automatically by the control  
> hierarchy. I hate to point this out again, but this process is  
> nicely illustrated by my hierarchical spreadsheet and (in a  
> less abstract manner) by Bill's "Little Man". So ordinarily,  
> there is no external "agent" (other than the control hierarchy  
> itself) that sets references -- and varies them -- and this  
> varying happens automatically. This means that principles are  
> varied AUTOMATICALLY to control system concepts; if the system  
> concept control systems need to vary the honesty principle  
> reference to preserve the perception of the system concept, it  
> DOES it; that's YOU doing it -- but there is no choice going  
> on; no conscious decision to be a little more or lest honest in  
> this particular situation. It just happens (just as you muscles  
> tense automatically to control the position of your limb).
> 
> When we talk about a person "setting appropriate standards" I  
> think we are talking about a phenomenon that is external to  
> the PCT hierarchy. Subjectively, when I talk to myself and  
> say "maybe I should try X" it is my consciousness that is  
> doing this. Consciousness is like something that hovers  
> over the hierarchy and tinkers with it occasionally; at least,  
> consciousness is what CAN tinker with the hierarchy.
> 
> I think we know a hell of a lot LESS about how consciousness  
> works (in terms of phenomena and models) then we know about  
> how hierarchical control works. But I do think that consciousness  
> (the feeling of having to choose -- "should I do X or Y"?)  
> only comes up when there is some degree of internal conflict  
> or lack of "output functions" that can be used to control  
> the required perceptions. When you are in control you are  
> rarely conscious of it -- unless you make some effort to  
> notice how well your hierarchy is working. When there is a  
> failure of control (due to conflict, lack of skill or insuperable  
> disturbance) then consciouness is there. As I said in  
> an earlier post -- moving consciousness to systems that don't  
> need attending to can create more problems than it might  
> solve.
> 
> Best regards
> 
> Rick
> 
> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*\*
> 
> Richard S. Marken USMail: 10459 Holman Ave  
> The Aerospace Corporation Los Angeles, CA 90024  
> E-mail: marken@aero.org  
> (310) 336-6214 (day)  
> (310) 474-0313 (evening)