Emotion etc.

[From Bill Powers (961015.0800 MDT)]

Hans Blom, 961014 --]

[with general reference to posts by others as well}

I don't know whether McCulloch included behaviors that go together
with intense emotions. Intense emotions arise, however, when these
very normal behaviors are disallowed. Just observe that little boy
who isn't able to get to a bathroom ...

McCulloch was speaking about innate behaviors; I think that before we
declare any behavior (or aspect of a behavior) to be innate, we should first
make sure there isn't some simple explanation in terms of acquired behavior.
The little boy probably gets emotional about not being allowed to go to the
bathroom because of his experiences of what happens when he wets his pants.
The innate behavior is to urinate when pressure in the bladder reaches a
detectable level, as an infant or a horse does. Barring any learned systems
which conflict with this build-in process, that is what will happen. The
learned behavior involves controlling where and when one urinates, which
shows that the innate system comes under the control of the learned one.

It's too easy to take a behavior like grooming as innate simply because
animals uniformly come to show it. A more careful consideration can bring up
other possibilities. For example, suppose that rats inherit a dislike for
certain sensations (zero reference level) having to do with matting of the
fur or scaling and itching of the skin beneath. Considering the limited
tools available to rats for doing something about such sensations, would it
be too surprising to find that most rats learn to scratch with a hind foot
or lick at the affected area? And does this automatically make the
scratching or licking in that situation "innate?"

···

--------------------

Anyway, it might be that an emotion "forces" a context switch and is
somewhat like an (unconscious) analog of what we consciously do when
we focus our attention: it focusses our attention FOR us in such an
inescapable way that it may save our lives.

If emotions are of our own making, then the emotion would change _because
of_ the context switch. The idea that I'm trying to put out for
consideration is that emotion is not a cause of behavior, but a concomitant
part of behavior. The traditional view is that we contain mysterious
intelligences which monitor the world for us, and arouse us or alert us to
important events by triggering emotions; the emotions then direct our
behavior (somehow) appropriately. Such ideas may sound plausible as stories,
but when you start looking at them from the modeler's point of view I think
you run into all kinds of difficulties and implausibilities. Just what sort
of perceptual capabilities would such emotion systems have to have? How
would they (the reticular formation for example) know about events in the
world and give them the required high-level intepretations? And how can such
an arrangement be reconciled with the HPCT model, in which there are quite
different explanations for the relationships between high-level and
low-level systems?

The problem we are having with the introduction of all these old traditional
ideas into the discussion is that they are usually brought in in a very
narrow context, as if we are starting over again and ignoring the model that
has been built up so far. Most often the traditional explanations end up
conflicting with the PCT model; if they are true, the PCT model is wrong,
and vice versa. This is something like introducing arbitrary postulates into
a mathematical development, with no regard for the relationship between the
new postulates and the ones on which the development is already being based.
Most traditional explanations have been ad-hoc, intended to deal only with a
very circumcribed set of phenomena. They generally take the least-effort
route: if dopamine is low in a part of the brain that seems associated with
a strange pattern of behavior, then obviously it is a lack of dopamine that
causes the strange behavior and it is that part of the brain that produces
that kind of behavior. Nobody asks why the dopamine is low, or what other
brain systems regulate its concentration. This is what comes from seeing the
brain as a collection of centers of spontaneous causation rather than as a
large interconnected system.

A theory of behavior is a large interconnected system; you can't arbitrarily
change one part of it without seriously affecting the rest. I suspect that
many people on this net who toss out alternative explanations have less
intimate familiarity with the structure and possibilities of the HPCT type
of organization than I do, since I've been at this the longest. So when they
toss in old ideas or casual suggestions, they don't experience what I see as
reverberations throughout the model that would result from accepting them;
neither do they realize what explanatory problems they are creating. Simple
isolated propositions about causality simply don't work in a model where
causation is distributed over many subsystems.

I'm not trying to cut off debate about new (or even old) ideas. I'm just
asking that they be thought through in a broader context, the context of the
model that we are supposedly exploring here. Before any idea is brought up,
I think it's fair to ask the proponent how the PCT or HPCT model might
explain the same phenomenon WITHOUT adding anything new. The model of
emotions that I have proposed fits the structure of the HPCT model without
requiring the addition of an extensive new class of inherited control
systems which, when you think about it, will require adding a whole new
hierarchical model in parallel with the one we now have, complete with
perceptual functions, comparators, and output functions at every level.
Every question we have asked about the existing HPCT model now has to be
asked about the added features, including questions of experimental test. Is
this really necessary or desirable?

Best,

Bill P.

[From Bruce Gregory (961015.1130 EDT)]

Bill Powers (961015.0800 MDT)

I'm not trying to cut off debate about new (or even old) ideas. I'm just
asking that they be thought through in a broader context, the context of the
model that we are supposedly exploring here. Before any idea is brought up,
I think it's fair to ask the proponent how the PCT or HPCT model might
explain the same phenomenon WITHOUT adding anything new.

I tried to suggest something similar a while back, but much less
clearly. Sign me up. If I go off track, please remind me of my
commitment.

Bruce

[From Bruce Abbott (961015.1230 EST)]

Bill Powers (961015.0800 MDT) --

I'm not trying to cut off debate about new (or even old) ideas. I'm just
asking that they be thought through in a broader context, the context of the
model that we are supposedly exploring here. Before any idea is brought up,
I think it's fair to ask the proponent how the PCT or HPCT model might
explain the same phenomenon WITHOUT adding anything new.

Speaking for myself and my proposal for emotion systems, this step was
rendered unnecessary by the fact that you had already presented that
information. I didn't find your account convincing, and tried to formulate
an alternative that in my view would remain consistent both with PCT and
with the spirit of HPCT while requiring a modification to the latter. I
have no desire to propose any alternatives if the current account can
explain certain facts I am aware of in a way that satisfies me; certainly
making the attempt earns me no brownie points in this venue, so there are
actually some rather strong reasons (having to do with other goals) why I
should just shut up.

The model of
emotions that I have proposed fits the structure of the HPCT model without
requiring the addition of an extensive new class of inherited control
systems which, when you think about it, will require adding a whole new
hierarchical model in parallel with the one we now have, complete with
perceptual functions, comparators, and output functions at every level.

In your previous post you stated that you agreed with me that it would not.

Every question we have asked about the existing HPCT model now has to be
asked about the added features, including questions of experimental test. Is
this really necessary or desirable?

It's not desirable _unless_ it's necessary. If it's necessary, then it must
be done, like it or not. So that's the real question: is it necessary?
What I've heard so far has not convinced me that it isn't, and I'm afraid
that would take far more effort on my part than I am willing to expend to
pursuade you that it is.

Regards,

Bruce

Bill Powers (961015.0800 MDT) sez:

McCulloch was speaking about innate behaviors; I think that before we
declare any behavior (or aspect of a behavior) to be innate, we should first
make sure there isn't some simple explanation in terms of acquired behavior.

I haven't read McCulloch's paper in years. But I don't think the innateness
of the behaviors was the point. In any event, my interest in his model
doesn't not require that his mechanism govern innate behaviors.

of perceptual capabilities would such emotion systems have to have? How
would they (the reticular formation for example) know about events in the
world and give them the required high-level intepretations? And how can such

McCulloch doesn't imply that the RF does this. The RF does have afferent
and efferent connections all over the brain. It may not be able (further)
to analyze inputs to the various neurofunctional areas, but it might well
be able to sense the general state of each neurofunctional area and that,
in conjunction with information about the intrinsic variables, might well
be all it has or needs to commit nervous system to a particular mode.

hierarchical model in parallel with the one we now have, complete with
perceptual functions, comparators, and output functions at every level.
Every question we have asked about the existing HPCT model now has to be
asked about the added features, including questions of experimental test. Is
this really necessary or desirable?

At one time Dave Hays and I reviewed quite a bit of the neural literature
and concluded it supported something like this multiplicity and overlap of
functions. In any event, I don't see that the HPCT model is itself derived
in a strong way from neural data. Yes, in places it is. But the upper
levels of the stack are pure invention and the notion that there is only
one stack seems more related to a general and understandable desire for
parsimony than to observations about real brains.

···

********************************************************
William L. Benzon 518.272.4733
161 2nd Street bbenzon@global2000.net
Troy, NY 12180 Account Suspended
USA
********************************************************
What color would you be if you didn't know what you was?
That's what color I am.
********************************************************

[Hans Blom, 961017d]

(Bill Powers (961015.0800 MDT))

McCulloch was speaking about innate behaviors; I think that before
we declare any behavior (or aspect of a behavior) to be innate, we
should first make sure there isn't some simple explanation in terms
of acquired behavior.

I think that the nature-nurture distinction is often ridiculous. In
an adaptive control system, we find that learning is determined by at
least four factors, which interact somewhat multiplicatively rather
than additively, as is often assumed. One factor is resistance to
external disturbances; if some disturbances are not present in an
individual's environment, he will not learn to cope with them (no
convergence; the initial naivete remains). A second factor is which
goals the individual has; lacking a certain goal (that may be normal
for others), how to fulfill that goal will not be learned. The third
factor is what _can_ be learned, i.e. which internal learning
mechanisms are present. The latter is greatly species-dependent. And
factor four is what control is hard-wired, either initially given and
subjected to further adaptation, or not adaptable at all.

The little boy probably gets emotional about not being allowed to go
to the bathroom because of his experiences of what happens when he
wets his pants.

All factors are present to some extent.

The innate behavior is to urinate when pressure in the bladder
reaches a detectable level, as an infant or a horse does.

Horses -- at least many (social and prey) animals -- do not just
urinate because their bladder has reached a certain fulness. Even the
process of urinating is much more complex than just that, both
because of innate drives and capabilities, and because of life-time
learning.

If emotions are of our own making, then the emotion would change
_because of_ the context switch. The idea that I'm trying to put out
for consideration is that emotion is not a cause of behavior, but a
concomitant part of behavior.

What's the difference if causality is circular? Sometimes an emotion
seems to come over me, but I can also nicely turn on some emotions at
will. A context switch causes an emotion: "great looking apple pie,
I'm suddenly hungry!". Fine. An emotion causes a context switch "I'm
hungry; my, does that apple pie look great!". Fine as well.

If we look for causes, PCT says that we ought to distinguish between
external influences, which are not always "disturbances" but
sometimes "opportunities" as well, and internal influences, changes
of reference level. It is sometimes impossible to distinguish their
effects. Unless, maybe, one has a very accurate model.

Greetings,

Hans