[From Jan Talmon (931201:09060]
WARNING
When I returned from my holidays, two weeks ago there were 300+ mail
messages. In the mean time another 300 have been coming in. So what I
say next may have been said by others since I'm now reading the CSG-L
mailings from 18-NOV-1993.
What I will try to do in the following is to describe why people
(wrongly) think that feedforward exists. This may shed perhaps some light
on the discussion.
I will take the three "real-life" examples that appeared on the net
recently as guidance.
SLIPPERY ROADS
Hans Blom provided a desciption of a near accident he had.
Interpreting his writings correctly, He perceived that the car was not
following the road in the proper way (error signals ???) even though
he turned the steering wheel. He then decided that the actual control
was not effective and that some other type of control was necessary
So rather than turning the steering wheel further, he consioudly decided
to keep it as it was, to break a little and to hope for the best.
WALKING IN A DARK ROOM
When you enter a dark room, you might experience some error as "routine"
visual perception
is not possible. To me it seems that you consiously make a mental picture
of the room and plan your way, such that you try to move in a direction wher
you might have some tactile feedback (walls, more or less fixed obstacles).
Although other may do this rather automatically in their own house, I still
need the tactile perceptions to guide me. On the other hand, subjectively
speaking, my performance in, for example hotel rooms is equal to what I do
at home.
PLAYING TENNIS
When you are serving in a tennis match, you are perceiving the tennis court
and the position where your opponent is. You more or less consiously decide
where your serve should hit the ground. Then you take your position and
serves and hopes for the best.
What stikes me in the above descriptions that in all three situations there
seems to be a kind of planning, or at least a consious awareness of that we
are making a choice. How does that fit in HPCT. To me it seems that at
least at some sufficiently high level in the hierarchy there is some control
going on. Getting home safely, Get into bed, Winning the tennis game/winning
the next point.
To me, during the driving on a road that appears to be slippery involves
reorganisation. An big error is pervieved at a rather high level in the
hierarchy, and the hierarchy is reorganised, getting the appropriate control
systems in to operation. It requires some practice to do the appropriate
reorganisation. It should not be random, but could be based on memory
traces of previous experiences.
Of course at the lower levels, control remains in operation. Keeping
the steering wheel in a fixed position is completely under control against
disturbances.
Regarding walking in a dark room, I'm not quite sure to what extent all is
feedback. However, one can easily envision that to achieve the high level
goal to get into bed, a series of subgoals need to be established. At the
higher levels of the hierarchy there should be a place to do that. So rather
then having the hierarchy of control systems take care of reaching our
goal automatically, we seem to become aware of the subgoals that need to be
reached before we finally reach our bed.
Playing tennis is perhaps even more trickier. We become aware that we need to
hit the ball with a certain speed, at a certain level above our head, with
a certain position of the racket to get the ball land at the spot we have
planned. When we are un esxperienced, we might only be able to set a goal for
body position during the service in order to achieve our higher level goal.
This is under complete control of the hierarchy. as we become more experienced,
we learn what the appropriate reference levels are for speed of racket,
position of racket etc to get the ball at the appropriate place. We might even
learn how to take the speed and direction of the wind into account.
Still we are controlling for a certain perception: get the ball at a specific
spot. The control hierarchy sets refernce levels for lower level control
systems, including those that control the muscles that make the racket
reach a certain point.
What is all common in this is that we seem to make a concious decision to
achieve a certain (sub) goal. To me the fact that we reason about how we
might achieve a certain goal causes people to make the (false) assumption
that feedforward exists. However, when keeping the steering wheel steady
does not provide the required result (namely getting home safely), error
signals will remain, or even become bigger and we could try another
reorganisation (if time permits).
Also when we walk in a dark room, and
we stumble across something in our way, we are reorganising our mental
picture of the room and make another plan how to get into bed. But there
is always some perception that tells you whether you reached your subgoal or
not. And also by playing tennis, we might not reach our goals often to place
the ball on the proper spot. But by training and experience, we are able to
learn how to set the proper reference levels for the lower level control
systems that are instrumental for getting our perception to be the winner
ot the match.
So my conclusion is that feedforward seems to occur because we become
aware of the setting of some reference levels. But it can largely be
explained as being part of the control hierarchy or as reorganisation
based on past experiences.
Jan