[Martin Taylor 2013.11.21.14.47]
Rick said I should not apologize for sending
[Martin Taylor 2013.11.17.09.13] and actually commended it, so I am
encouraged to continue.
In that message I pointed out some implications of Bill's suggestion
that each reference input value is the output of an associative
memory “cell” addressed by the many output values of relevant
higher-level control units. In this message, I continue with another
implication, relating the perception of emotion to the storage of
perceptual “situations” for later use as reference values. In a
later message I will address control of avoidance, and
“approach-avoidance conflict”, which is another implication of
Bill’s vector-addressed memory (or so I think at the moment).
To recap, Bill proposed that the content of the associative memory
cells would be the values of earlier perceptions, but he did not
suggest why the perceptual values to be stored might be selected
from the continuum of values that would have been experienced for
any particular perception. I suggested that the stored perceptual
values might be those that exist when error values are changing
sharply – in other words at moments of events at the control level
in question.
In this context an "event" is not the same as the "event level" of
the hierarchy. An event at a level occurs when either the perceptual
value or the reference value changes abruptly. It is an edge,
whether across time or across the related perceptions at a level, or
(commonly) at both.
An emotion is a perception. As such, it presumably is part of the
array of perceptions that are stored when perceptions are stored for
later retrieval. But it is a rather special kind of perception,
because, unlike other perceptions, it has a quality of being
desirable or undesirable. Would you or would you not want to repeat
the situation in which you perceived this emotion? Moreover, emotion
is likely to depend more on control (or lack of it) in imagination
rather than on the actual ongoing control through the environment,
though both do happen. If one is in a situation in which one
imagines good things will happen by one’s efforts, one is likely to
feel a “good” emotion. and conversely.
When perceptions are being well controlled, the error value is not
changing very much, and the controlled perception is unlikely to be
in one’s conscious perceptual field. Nor does one perceive any
associated emotion, at least if in imagination good control
continues – unless this good control comes after a period in which
the perception had been out of control. Such gain-of-control moments
are usually associated with good feelings: “I finally managed to
pull it off!”, “This music is just perfect for me right now”. It is
an “event”, as described above. But when a perception is not well
controlled, the error value can fluctuate strongly, a situation not
ordinarily associated with pleasant emotions. Under such
circumstances, the imagined future possibilities might be dire, and
the emotion might be fear.
Some people enjoy putting themselves into a situation they also
fear. In fact, it is rather normal to do so – getting into a formal
competition or game, climbing a mountain, driving fast in a race,
and so forth. The person imagines they might lose, or even die. Why
should we enjoy fear, especially when the fear could be of death
rather than just of losing the game or the race? I suggest it is
because there is a conflict between avoiding the feared situation
and attaining the other emotion, the emotion we perceive when we
achieve control in the fear-inducing situation. We imagine getting
out of the feared situation by our own actions, and maybe we
succeed. We win the game, get to the mountain top, and what had been
uncontrolled is now controlled. We remember the fear and its
elimination. The state of “having achieved” is one we want to be
able to repeat, or at least in our imagination to recover, even if,
or perhaps especially if, the uncontrolled state had included the
emotion of fear.
Bill proposes that emotion is associated with reorganization.
Reorganization involves change in the way we act through the
environment to control perceptions. If the above account makes
sense, reorganization should usefully happen when failure to control
persists or error continues to increase. There is no point in
reorganizing one’s climbing technique when scaling a difficult cliff
just because one is scared of a possibly fatal fall. The random
component of reorganization allows for no second chances if an
e-coli move is in the wrong direction. Once you have fallen off the
cliff, you cannot go back and do it differently, at least not on a
cliff that induces serious fear.
I suggest that emotion has a different function, although prolonged
unpleasant emotion is indeed likely to occur along with
reorganization. Accepting Bill’s association between chemical and
neuro(-electrical) hierarchies of control, I suggest that the main
effect during the emotion is to chemically alter the gain of some
neural control systems and to switch some control systems to
controlling in “avoidance mode” instead of “attraction mode”). A
situation associated with a good emotion is one that should be
repeated if the occasion arises, whereas one associated with bad
feelings is one to be avoided in future. When it comes to storing
perceptions in the associative memory system that produces reference
values, I suggest that the stored perceptual emotion has the effect
of a switch.
Either way, I suggest that the emotion occurs when perceptions are
stored, and, possibly coincidentally, when one is conscious of those
perceptions. The linkage between consciousness and memory may be
purely coincidental and loose, but it could be deeper and tighter.
···
The above brings up an issue that has not been discussed recently on
CSGnet, though it did come up a long time ago: “avoidance control”:
“Don’t go near the bully, but go anywhere else, the further from the
bully the better”.
This is a topic separate from emotion, and has some possibly
unexpected ramifications involving situation (vector) control, so I
will treat it in a separate thread. In particular, we need to
discuss how a control system could control a perception to be “not
near” a reference value. I will argue that this cannot usefully be
done with a scalar control system, because the controlled perception
cannot cross the value to be avoided; it needs vector control in
more than one dimension.
Martin