Explanation and clinical terminology
Unedited posts from archives of CSG-L (see INTROCSG.NET):
Date: Thu Feb 24, 1994 11:17 am PST
Subject: Depression
[From Bill Powers (940224.0915 MST)]
I've toyed on and off with a "universal error curve" that seems to explaina number of phenomena that are hard to explain otherwise. The basic problemoccurs with "giving up," which is relaxing one's attempt to correct error whenthe error gets too large.
The conventional comparator has an input-outputcurve like this:
* * * * * * * * * *
* | error
* | signal
* |
* |
----------------------------*-----------------------------
actual error | *
| *
| *
| * * * * * * * * * * * *
In other words, as the actual difference between r and p, the actual error,increases in either direction, the error signal also increases in theappropriate direction. This is clearly not physiologically feasible, because itimplies infinite error signal for infinite actual error. In a real system, allsignals are limited by the maximum possible impulse rate. So an error signalshould increase to some limit and then remain constant, as shown above.
But this implies that for a disturbance large enough to cause the errorsignal to reach a limit, and for all larger disturbances, the error signal willsimply remain maximal, and the output will be pegged at the maximum possibleamount. This is not how people behave. When disturbances are clearlyoverwhelming, people don't waste their energy indefinitely in a futile attemptto oppose the hurricane or the tidal wave or the earthquake; they give updirect opposition and try something else.
This phenomenon could be explained in terms of higher-levelcontrol systems, which perceive the futility and turn off the reference signalto the system that has been overwhelmed, sending reference signals instead tosome other system(s) that can still operate. While this may be a correctexplanation in some circumstances, there are others in which it seems lesspersuasive.
For example: if you are an ice-creamfreak, you don't go around all the time striving with maximum effort to getsome ice-cream.In fact when there's no ice-creamaround, you pretty much forget about wanting some. There may be a little bit oferror signal, but not enough to warrant all-outaction. If asked, you would say that you still really, really like ice-cream,but you aren't acting as if you do.
But what happens when you see some available ice-cream?As soon as you see it, the desire for ice-creamis felt; the closer you get to being able to obtain the ice-cream,the more you want it and the harder you try to get it. The smaller the get-some-ice-cream error gets, as you approach zero error, the more effort you put intogetting it.
If you haven't noticed, this is the exact opposite of what control theorywould predict. As the error gets smaller, the effort to correct it gets larger.This effect has been observed many times, however, by experimentalpsychologists; they called it an "approach gradient," and measured it by (forexample) measuring the force exerted by hungry rats on a tether that restrainedthem from approaching a food dish. The force increased as the rats were allowedto approach more closely.
Of course those approach-gradientexperiments weren't done right --the approach distance allowed never reached zero. If it had, the experimenterswould have seen an abrupt reversal in the gradient. Within some small radiusaround the food, the effort must have _decreased_ as the distance to the fooddecreased. If this were not the case, then an unrestrained rat would have beenunable to stop at the food dish to eat: maximum acceleration toward the fooddish would have occurred just as the rat arrived at the food. So clearly thecurve is the way it ought to be for control in the immediate vicinity of thefood --but outside that radius, the effort decreases as the error gets larger.
All this leads to my "universal error curve," shown below:
* | error signal
* * |
* * | zero error
* *| /
*---------------------------*-----------------------------*
actual error |* *
| * *
| * *
| *
In the immediate vicinity of zero error, the curve is the same as in thefirst diagram: a decrease of actual error produces a decrease in error signal(with appropriate sign for negative feedback), and hence a decrease in output.This is the "normal control range". But if disturbances can push the error pastthe peak in the curve, in either direction, further increases in error willlead to a _decrease_ in error signal and output. Depending on the stiffness ofthe connection between the disturbing variable and the controlled variable, theresult will be either a smooth decline in output effort with further increasesin disturbance (stiff connection), or a positive-feedbackeffect in which all resistance will abruptly disappear while the disturbanceforces the error to some very large value (elastic connection). The actualerror will end up large, while the error signal (at the ends of the abovecurve) ends up small. [Electronikers will recognize this as describing theslope of the load line].
Now apply this curve to the getting-ice-creamcontrol system. We assume that the reference level for ice-creamremains at its usual non-zerosetting all of the time. The conventional error curve predicts maximum effortto get ice-creamwhen there is no possibility of getting it --no ice-creamto be had, or no effort that will produce it.
The "universal error curve," however, predicts that when the actual erroris far above some rather small amount, the error signal will be very small, andthus there will be almost no effort to obtain ice-cream.We are far out to the right on the universal error curve. Now, if thedifference between the reference signal and the perceptual signal becomessomewhat smaller --if there is a chance to get some ice-cream--we will move to the left, toward the intersection of axes, and the error signalwill _rise_, causing a greater amount of effort (and felt desire). This effortwill bring the perception even closer to the reference level, further reducingthe actual error and further increasing the error signal. The result againdepends on how rapidly the perceptual signal changes with an increase in effort --either the effort simply increases as approach to the ice-cream gets easier, or there is a positive-feedbackeffect and the effort rises abruptly to a level sufficient to bring theperception within the normal control range.
This has a close relationship to what Martin Taylor proposed as "time-bombs"in the control hierarchy. Some kind of limit or nonlinearity is required tobring it about. It's possible that the universal error curve is apparent only,arising from interactions at several levels of control or between severalsystems at the same level, but clearly it would also work if it were acharacteristic of a single comparator or combination of comparator and outputfunction. In experimenting with a single externally-observedcontrol phenomenon, however, it doesn't matter how this curve arises: theeffect is as if it is a property of a single equivalent control system.
I haven't done any experiments with this, but experiments are certainlypossible. Another task added to the vast backlog of things that PCTexperimenters must get around to some day (a backlog that doesn't decrease anyfaster while we sit around debating systems philosophy).
So, depression. Depression is giving up, isn't it? With the universal errorcurve in mind, we can see how depression might result not from abandoninggoals, but from maintaining goals but experiencing so much error that theeffort to oppose or correct it has dropped almost to zero --the depressed person is way out to the right or left on the universal errorcurve. The person still wants love and understanding, or still wants to conquerthe world and eliminate all the Jews, but the world has put such a largebarrier in the way that the person has ceased to try any more.
Consider one more factoid. My theory of emotion says that affect is aconsequence of error signals in the control hierarchy, a physiologicalpreparation for the action that will be used to correct the error. But if theabove interpretation of depression is correct, the error signals will be small,because the actual error has driven the person to an extreme on the universalerror curve. The result will be not an extreme of emotional arousal, but almostno emotion at all. The depressed person will find the experienced world verydifferent from the desired world, but will not act with appropriate energy tocorrect the error, and will feel emotionally dead. The depressed person can'tget psyched up enough to act.
Curing depression, if it is caused as I imagine above, would probablyrequire some method like the one Hal Pepinski proposes (940223.1000). Somehowthe world that a person perceives must be brought closer to the world that theperson wants to perceive but is unable to act to bring about. Or, the referencelevels must be changed enough to get the perception back into the normalcontrol range, where goals, errors, efforts, and feelings are normally relatedand feasible in the world that exists.
This can be done in some cases through loving interactions with supportiveothers, but not all cases are so easy. Hitler became very depressed because theAllies would not let him conquer the world and get rid of the Jews, and in factthe war increased the difference between what he wanted and what he perceivedfar beyond his capacity to act in opposition. He ranted and raved, and then ashe passed over the peak of the universal error curve he fell into depressionand committed suicide, the ultimate of giving up. Putting Hitler into a socialsituation where his goals became achievable might have cured his depression,but the rest of the world would not have considered this humanitarian approachacceptable. The only way the world would have let Hitler become a normal personwould be to find a way to change his goals. The only alternative was to destroyhim and those who followed him. Shit happens.
Best, Bill P.
Date: Thu Feb 24, 1994 11:28 am PST
Subject: Depression (from Mary)
[from Mary Powers 940224]
Hal P. on depression (940223)
> I diagnose the problem as not having one's own reality and feelingsvalidated by others. Clinical depression is the feeling that one's experienceof the world is isolated and apart from others'.
Hal, I think that you are confusing the problem with the cure. Depressionis a rotten feeling about oneself and the world. It is aggravated, but notcaused, by people saying things like "things aren't that bad, cheer up, look atthe bright side, you'll feel better tomorrow" and similar invalidating remarksthat lead a person to feel isolated (no one understands me).
A major role of therapists (unless they are into curing depression with aprescription for Prozac) is to understand, appreciate, and accept the depressedperson, and communicate this to him -i.e. validate his feelings. This is what Carl Rogers' client-centeredtherapy was all about. PCT suggests that this helps the client to validate hisown reference signal that he himself could not fully accept because itrepresents one side of a conflict. By doing this, the therapist also helps theclient to become aware of the opposing reference signal as well -the part of him that is resisting feeling so bad. Getting both sides of theconflict into awareness is the first step towards resolving the conflict -reorganizing.
What is important about the therapist is that a depressed, conflictedperson is unable to think very well about his problem. As he talks about itwith a person who is not in conflict in that area, he is able to get histhoughts back in a clearer fashion than if he were trying to run the situationthrough his own imagination. The therapist doesn't share the client's referencesignals, but he is uncritical and unjudgemental, and supports the clientthrough a scary process.
This is as far as PCT and a brief career 40 years ago at Rogers' counselingcenter takes me. I'd love to hear from Dick Robertson and David Goldstein onthis topic. And David, as the only PCT clinical psychologist in captivity (i.e.trapped in this Net) how about getting out your DSM or whatever it's called andtelling us what depression "really" is -and what you think it is after 20 (?) years of PCT.
Mary P.
Date: Fri Feb 25, 1994 9:00 pm PST
Subject: Definition of Depression
From David Goldstein 02/25/93
The Diagnostic and Statistical Manual of Mental Disorders-Revised(theDSM-3-R)has a few entries related to the definition of depression.
Under the heading "Mood Disorders" there is:
296.2x Major Depression, single episode
296.3x Major Depression, recurrent
300.40 Dysthmia
311.00 Deressive disorder NOS (Not Otherwise Specified)
Under the subheading "Bipolar Disorders" there is: 296.5x Bipolar disorder,depressed
The x in the fifth digit could be--1=mild,2=moderate, 3=severe, without psychotic features, 4=with psychotic features,5=in partial remission, 6=in full remission, 0=unspecified.
In addition, under the heading "Adjustment Disorder" there is: 309.00Adjustment disorder with depressed mood.
Furthermore, under the heading "Organic Mental Disorders" one finds: 293.83Organic mood disorder
Each of the above categories has its own diagnostic criteria. I will givethe criteria for major depressive episode.
A. 5 symptoms in the past 2 weeks from the list--
depressed mood
diminished interest or pleasure
weight gain or loss
insomnia or hypersomnia
feelings of guilt or worthlessness
motor agitation or retardation
energy level low
concentration, thinking, difficulties
suicide ideas or ideas about death
One of the 5 must be diminished interest or depressed mood. Each of theabove are described in more detail than I have indicated.
B. Organic factors or situational factors not causative.
C. Not preceded by psychotic symptoms
D. Not part of psychotic disorder
Mary, this should give you a flavor of the DSM-3-Rdefinition of depression. There is no attempt to theorize in any way in thisdefinition. It takes most mental health professionals a while before they canuse this definition with any confidence. I will write about some applicationsof an HPCT approach to this topic in a separate post.
For those not familiar with the DSM-3-R,a diagnosis consists of descriptors on 5 "axes":
Axis 1--Presentingsymptoms. Major Depression might appear here.
Axis 2--ChronicPersonality Disorders or Traits.
Axis 3--PhysicalHealth Conditions
Axis 4--Stresslevel/condition
Axis 5--Generallevel of functioning now and best in past year.
There will be a new, DSM-4manual coming out in the next few. months. I have seen the progression from theoriginal DSM and have seen the changes. There is a definite improvement fromedition to edition.
A person with a Major Depressive Disorder on axis 1 will be/behavedifferently depending on the status on the other axes. A person may havemultiple diagnoses on each axis. There is no attempt to assign one diagnosis toa person. The diagnoses are treated as descriptions which apply or not.
Date: Sat Feb 26, 1994 10:40 am PST
Subject: DSM-3-R
[From Bill Powers (940226.1115)] David Goldstein (940225.2134)
> Each of the above categories has its own diagnostic criteria. I willgive the criteria for major depressive episode.
> A. 5 symptoms in the past 2 weeks from the list--
depressed mood
diminished interest or pleasure
weight gain or loss
insomnia or hypersomnia
feelings of guilt or worthlessness
motor agitation or retardation
energy level low
concentration, thinking, difficulties
suicide ideas or ideas about death
One of the 5 must be diminished interest or depressed mood.
So to diagnose depression, you must have observed either diminishedinterest or depressed mood. Since you can't diagnose depressed mood without adiagnosis for depression, that means you have to observe diminished interest.Strike the first item.
The rest of it looks loose enough to allow the diagnostician to call almostanyone "depressed." What it boils down to is that if the diagnostician decidesthat this guy looks depressed to him, he or she can, if asked to justify thisimpression, point to the items above. There isn't the slightest indication ofwhat's wrong with a person who shows these symptoms, is there?
This sort of one-size-fits-alldiagnosis seems unforgivably crude to me.
Best, Bill P.