Objective modeling vs direct experience

Unedited posts from archives of CSG-L (see INTROCSG.NET):

Date: Mon Mar 07, 1994 11:26 am PST

Subject: Knowing

[From Bill Powers (940307.0820)] Bill Leach (940307.0021)

Glad you've broken the ice with BCP. When you've finished it, you'll findtwo chapters that my editor cut out of it in LCS II: the one on emotion and theone on the method of levels.

Bill Leach (940306.1903 EST) --

> I sort of accept that a "control loop" can not "know" anything aboutthe disturbance. I'm not so sure that I fully accept the idea that the entirecontrol system can not have knowledge about the disturbance.

RE "knowing" (anything):

This is a difficult subject because it straddles two worlds: the world ofobjective modeling in which we try to "reverse-engineer"the nervous system, and the world of direct experience, where we experience theoperation of the same system from the viewpoint of an occupant of thebrain.

In the PCT diagrams of control systems that you see in BCP and elsewhere,you will never find anything labelled "awareness" or "consciousness." From themodeling standpoint, all that is required for us to say that a perceptionexists is that a perceptual signal be present in the appropriate pathway.Perception of a cubical shape requires that signals representing the sensoryattributes of the cube (sensations) enter an input function capable ofgenerating a perceptual signal whose magnitude indicates the degree to which acube is perceived to be present. There is no requirement that the person as awhole have any consciousness of experiencing a cube.

The reason for this odd-soundingconcept is that if control depended on _conscious awareness_ of perceptualsignals, then only those control systems containing perceptual signals of whichwe are consciously aware could work. Conversely, we would have to be consciousof all perceptual signals in all control systems that are actually working at agive time. Neither premise fits experience. When I am typing this stuff, I amaware of the words I choose appearing on the screen and of a sort of scrabblingof fingers over the keyboard, but I am not aware of the joint angle controlsystems, the velocity control systems, or the force control systems that areconverting my desire that a given letter or word appear on the screen into thespecific reference signals being sent to the lower-levelsystems, and the resulting states of the controlled perceptual signals. Yet ifI wanted to, and shifted my focus of awareness, I _could_ be conscious of atleast a great part of those lower-levelperceptual signals. I assume that's true of everyone.

It's the hierarchical structure that causes the biggest problem. I can beaware of control processes at a certain level, but the scope of awareness, mineat any rate, is limited. If I'm "concentrating" on how to get a program toprint something out into a text file, my consciousness is almost totallypreoccupied with the logical and procedural goals and the difference betweenwhat the program IS doing as opposed to what it SHOULD BE doing (the Germanword for reference level is SOLLWERT --should-be).But while I'm thus preoccupied, the shifting higher-levelerrors are being continually translated into more specific lower-levelgoals, and so down the hierarchy all the way to the systems that are doing thetyping for me, keeping me from falling out of the chair, and so forth. Clearly,in order to control the perceptions I am conscious of, there must be countlessother perceptions at lower levels that are also being controlled. In fact, evenfor the higher levels of perception in consciousness to _exist_, the lower-levelperceptions of which they are functions must also exist. But I'm notsimultaneously aware of those lower-levelperceptions, for the most part.

This problem exists in the other direction, too. While I am working out theproblem of printing the data to a file, I know that my goal is to get the dataprinted to the file, but WHY I am trying to reach that goal is not in directawareness. When I finally solve such a problem, I sometimes have a moment ofdisorientation: now just why was it that I wanted to print that data to a file?Then there's a sort of "Oh, yes!" feeling as I return my awareness to the higher-levelprocess in which solving this particular problem was only a means to anend.

If the higher-levelcontrol system hadn't been working, there would have been nothing to supply thegoal to the system where my awareness was. So clearly, the higher-levelcontrol system can go right on working when my awareness of it is absent. Thismeans it must have had an intact input function, perceptual signal, errorsignal, and output function, all humming quietly away, working totallyautomatically. Again, perceptual signals exist and MUST exist without being inawareness. Only now we are talking about perceptual signals at levels _above_the level of awareness.

So when we speak of what a control system "knows", we have to keep thequestion of consciousness separate. The knowledge in a control system consistsentirely of its perceptual signal. We must also, however, remember that this isa multi-ordinalmodel with perhaps 11 levels in it and many systems at each level. Theknowledge contained in one perceptual signal at one level is put together withknowledge in other control systems at the same level to create knowledge at ahigher level in the form of higher-levelperceptions. From the same body of lower-levelperceptions there may be many different ways of extracting knowledge at ahigher level, in many different control systems operating in parallel. A cube-signal,put together with other configuration signals, can lead to signals representingrate of spin, bouncing off other objects, spatial relationships with otherobjects, symbols representing the cube and its relationships, temporalfunctions, logical functions, principles, and system concepts --all at the same time although at different levels.

With awareness out of the picture, we have a system that contains signalsat many levels representing various aspects of a world at various levels ofabstraction. It carries out all functions of a living human system, includingthoughts, feelings, actions, goal-seeking, whatever. It's just a big analogue-digitalcomputer, with no more awareness of its own internal processes than my 80486has.

When you put awareness into this system, what you get (according to myhypothesis) is the effect of connecting a bunch of perceptual signals to somesort of receiver. This receiver needs no cognitive functions, no computingcapacities, no capability for action: all it does is receive. When it doesreceive, we get a conscious world composed of some subset of all the perceptualsignals in the hierarchy. The whole hierarchy continues to function as usual;the only difference is that we become aware of some part of its functioning.Then we feel that we are consciously _doing_ the things that the hierarchywould be doing anyway. When control systems in the hierarchy experience anerror and produce actions, we sense the error-basedoutput, through the imagination connection, as what we are doing to achieve thegoal --we the conscious observers, not just the automatic machinery. But it is thelearned control systems that are actually doing the doing, the thinking, thecognizing --even when the doing is something as intellectual as making a choice or adecision, or formulating a plan, or making a judgment. The observing systemmakes no judgments. It is simply aware.

The other side of awareness is volition: producing a change in a referencesignal in the hierarchy for a reason unconnected to anything that is going onin the hierarchy. And I can make a case that awareness and volition areassociated (vaguely) with the reorganizing system, so that awareness can serveto focus the process of reorganization. You'll be getting to that in BCPeventually.

All of this is a rather extreme position, saying that awareness orconsciousness carries out NO functions in the hierarchy; that the hierarchydoes every last thing that we call either physical or mental, and withoutawareness being required. Being stated so firmly, this hypothesis can easily bedisproven, by finding effects on control processes that result from shiftingawareness. I am quite sure that such effects can be found --but until they ARE found, and experimentally verified, the hypothesis as it isnow stands.

Out of this hypothesis, the method of levels grew. This is a method ofpsychotherapy in which a person is encouraged and helped to move the locus ofawareness up one level at a time, each shift bringing the focus ofreorganization to bear on a new level of organization, and presenting toconsciousness a world organized in a new way. When carried as far as possible,as Kirk Sattley and I did experimentally about 40 years ago, this procedureleaves a person in a state of what seems to be pure awareness, with many of theoperations of the hierarchy being laid out to view but with no identificationof awareness with them, no participation in them. One is then what DavidGoldstein and I have come to call the Observer. It was through working thismethod with a particular patient that David was able to help a woman withmultiple personalities to begin reintegrating. The woman came to understandcompletely what was meant by "the Observer," and from that time on, anypersonality could be reached through the Observer.

For all I know, the Observer is another level in the hierarchy that Ihaven't been able to identify. Maybe there are levels beyond that (one mysticaland somewhat nutty friend once said, "Oh, there are THOUSANDS of levels!").That doesn't much concern me. Working out this whole scheme at the levels wesort of understand is enough of a project for one lifetime.

Knowledge and aware experience are different things. Knowledge is just oneperception as a function of other perceptions. It covers the whole range, fromsensations to system concepts, and it requires no awareness for it to exist.It's just how we become organized to perceive and act on the world. Or so itsez as of March, 1994.

Best, Bill P.