Implicit reference levels; emotions; no error

[From Bill Powers (930903.0900 MDT)]

Avery Andrews (930903.1635) --

... the sentence `Andrews in the linguistics department has
skipped three classes for no known valid reason' will suffice
to trigger action. This is the sort of thing that logic seems
to [too?] [to be?] good for (unlike beer-getting), and I
couldn't think of any other way that the right kinds of action
might be initiated.

We need a name for this phenomenon: the "taking the reference-
level for granted" phenomenon.

While it's true that one _could_ set up a stimulus-response
system in the way you suggest, the result of a blindly triggered
action is not likely to have much systematic effect on anything.
For example, one could in fact set up a logical system such that
the proposition in question, if true, triggered "Give Avery
Andrews a new car," or "Buy my wife a present." But how would
that bear on anything that is being affected by Andrews' absence?

Actually, the proposition `Andrews in the linguistics department
has skipped three classes for no known valid reason' is simply a
perception at the sequence level (or whatever level you want to
call it). Before any sensible action can be called for, we have
to know the reference-setting for the truth of this proposition.
If the reference-level is TRUE, no action is required. Perhaps
the Dean is thinking, "Whatever they're doing to get rid of
Andrews, it's working."

Unless you specify the reference condition for the above
perception, or any perception that seems to lead to action,
there's no way that "the right kinds of action might be
initiated." By "right kind" you mean "right for a achieving a
purpose relating to the perceived condition."

This problem is present in every stimulus-response description of
behavior. The description assumes a reference condition without
explicitly mentioning it. It appeals to what "everybody knows."
The burnt child avoids the fire, they say. Why should that be
true? Oh, you mean that the child prefers not being burned?

The only way to see how to model behavior under PCT is to make
everything explicit instead of allowing common-sense notions to
creep into the background assumptions (too much). I can't
explain, under PCT, why a man seeing a 3-ton safe falling on him
will dodge to one side unless I make explicit his reference level
for perceiving 3-ton safes falling on him. As soon as you make
the reference level explicit, you can see how to organize the
control system. If you just assume that it's natural to dodge
when you see a safe falling on you, you'll end up with SR theory,
which won't work for the man who has desperately been looking for
a method of suicide that won't void his insurance. One reason
that SR theory works so poorly is that it stereotypes behavior
and particularly reference levels.


Bruce Nevin (Fri 93093 07:53:28 EDT) --

( Avery Andrews 930903.0920 ) --

I think that is an intriguing suggestion, that the fixing of
attention is a `pattern-action' rule with a rather open-loop
flavor that starts up (or increases the gain on) control
systems that subsequently run closed loop.

See above.

( David Goldstein 09/02/93 ) --
HPCT and emotions

Some of our friends in the eastern hemisphere have directed a
lot of systematic investigation to these and related questions,
and, at least in the literature written in Sanskrit (or its
descendant, Pali), can be most exhaustive in identifying and
categorizing what is going on. Greg Williams, do you have any
references to works summarizing these findings specifically
with respect to different kinds of emotions and the error
signals that occasion them?

"Systematic investigation" wasn't the strong suit of our friends
in the ancient Far East. They were naturalists, not theorists and
particularly not experimentalists or modelers. I take my hat off
to them as observers of the human condition, but I wish they had
been more inclined to put their flat statements of fact to some
sort of test.

The PCT notion of emotion, as I understand it, is not that
emotions are "occasioned" by error signals. That seems to leave
the emotion as an independent thing that is triggered off by
error. My concept is that "emotion" is a _name_ that we give to
the combination of a sensed somatic state and a goal-seeking
process, with the name of the emotion being selected mainly in
terms of the goal involved. So "fear" is the name of a somatic
state (essentially the same as the somatic state in "anger" or
"excitement") associated with the goal of getting away from
something. The somatic state results from the error in the same
way that the state of the lowest-level motor systems results from
it, and is simply part of the process of error-correction.
As David suggests, the somatic component can be experienced at
many levels: certain as a configuration of sensations, and
probably as a transition, too ("relief" or "alarm"). And probably
Tom Bourbon (9308xx-9xx) --

Your wonderful posts are to be studied at length, not commented
upon hastily. Remember Dag's dictum: no error, no comment. How
many authors of good books have you written to lately?
Best to all,

Bill P.