[From Bill Powers (2001.10.18.1058 MDT)]
Bruce Abbott (2001.10.16.1635 EST)--
What would you say to the proposition that emotions are the result of error
signals, rather than direct responses to external events?
Emotions such as fear or anger do seem to result when there is error in
certain variables one is attempting to keep under control. My view is that
the brain includes specialized circuitry for various emotions, circuitry
that provides pre-organized solutions to certain general types of problem
commonly encountered by members of the species. For example, an organism
may perceive a threat to its safety/security (is threat a perception, an
error signal, or both??); the degree of threat perceived will depend, among
other things, on the organism's perception of its own ability to
successfully deal with the threat. The perception of threat (error in
perception of security?) alters reference levels for a number of controlled
perceptions, including physiological ones (producing output changes in
these systems that prepare the body for the vigorous action that may be
required for defense). We sense the physiological arousal thus created,
and this becomes a part of the experience of the emotion. We feel strongly
motivated to engage in particular actions that (if adaptive) tend to
counter the threat -- another aspect of the resetting of reference levels
mentioned above.
I think we're in substantial agreement about the basic nature of emotion.
This sets us apart from those who claim that "emotional responses" to
external events can occur without the participation of higher systems. The
old question was, "When you see the lion, do you run first and then feel
afraid, or feel afraid first and then run?" This question suffers from the
usual problem of what you mean by "you." In a hierarchical system, the
lower systems are faster, so if there are disturbances of low-level
variables they react first, but of course they would not react were it not
for the reference levels specified by higher systems. If a higher system
establishes the reference condition "No experience of large images of
objects that look and move like lions," the reaction to the sight of a lion
may well be very rapid, typical of a third-order system, and the experience
of emotion, if it occurs, may occur essentially simultaneously, and before
any higher-level experience can change (when, for example, you find that
somebody has locked you in the cage with the lion and (relatively slowly)
work out the logical implications of that fact).
I'm not sure we need to postulate "special circuits for various emotions."
As I see them, emotions are basically _high-level perceptual categories_,
the components of which are bodily feelings and imagined reference levels
(the latter being experienced as what we want or our goals). The bodily
feelings arise from preparation for action via the hypothalamus, and the
cognitive component comes from the other perceptual control systems. We
perceive different patterns of (goal plus somatic sensation) as different
emotions. Sometimes it is only the goal aspect that makes the difference,
as in "fight or flight," where the goals differ but the somatic sensations
are apparently identical. In other cases, such as the difference between
fear and depression, the somatic component may also differ (I don't have
anything reliable to say about that, however).
I would agree with you that we probably inherit some control systems, which
is almost the same as saying we have "pre-organized solutions to certain
general types of problem." Of course those inherited control systems also
work when there is no special problem (disturbance), so the PCT view would
de-emphasize the "problem" part of the statement. We may experience a spurt
of emotion when something disturbs us enough to almost knock us over a
cliff, but the same built-in control systems that react against such a
large disturbance are always acting anyway, keeping us upright against all
sort of small disturbances that don't create enough change of somatic state
to lead us to interpret the result as an emotion. Whether we call an
experience an emotion seems to depend largely on how big the error is, and
not so much on what kind of error it is (although the kind of error can
influence what we call the emotion when we name it).
The emotions, if they arise as I propose (and you seem to agree) simply
come as part of the package, since _any_ energetic control action will be
accompanied by a change of somatic state large enough to be called an
emotion, whether the control process be learned or inherited. Some people
react with strong emotions to being called certain names ("socialist" comes
to mind), but one can't claim that such reactions are inherited.
I don't think we need a special category called "emotion" in PCT, although
of course it's a useful label for certain kinds of experiences. The model
doesn't really need that category to explain what happens and how we
experience it.
Where I think we differ is that I believe that there are specialized
"emotion" systems in the brain, whereas you see what are experienced as
emotions as simply the sensed byproducts of strong error in basically any
of the organism's perceptual systems, and not as manifestations of
specialized "emotional" circuitry in the brain.
Yes, that's the difference I see, too. But I really do concede there is
"specialized circuitry" in the sense that higher systems not only send
signals into the reference inputs of midbrain perceptual-motor control
systems, but also send accompanying signals into the hypothalamus and
thence via the pituitary to (chemical) reference inputs of all the major
organ systems. That's probably pre-wired, so it leaves room for the sorts
of innate reactions you may be thinking of. But I think PCT puts a new
light on the old concepts of emotion.
Best,
Bill P.