Pattern control

[From Bill Powers (931101.1010 MST)]

Gavan Lintern (931101.0836) --

You're asking some very important questions about what is and is
not a controlled variable in collective behavior. If we can come
up with some acceptable and sensible answers I think that many of
our views on these problems will tend to converge.

I think that you correctly reject the idea that the "patterns of
organization that emerge from interactions of the elements is of
no consequence". If patterns show up regularly, they must have
some reason for showing up. There are several kinds of reasons we
can propose:

1. The pattern is truly a side-effect dictated by the properties
of the elements and having no other significance.

2. There is some benefit to the individuals in acting in a way
that creates these patterns, a benefit due indirectly to the
pattern itself.

3. There is an evolutionary advantage conferred by controlling so
as to create the pattern (other means of control being possible);
this advantage does not bear directly on any individual's goals
or behavior, but has a species-wide selection effect.

4. The pattern itself is sensed and controlled by each individual
in the group.

In the CROWD program, the first is the correct choice. We know
this because the program is completely known, and the presence of
arcs and rings has in fact no effect on the individuals.

The second choice might be illustrated by the V-patterns in which
geese fly. While no goose has a control system for producing a V,
it is a fact of aerodynamics that the effort of flying is least
for a goose flying just behind and to the side of another goose.
Assuming that geese will fly so as to avoid excessive effort,
each goose could learn that flying in a certain position relative
to another goose will reduce the effort to an acceptable level.
The lead goose gets fatigued first, and so drops back after a
while to the less effortful position. This creates the impression
that the geese are deliberately taking turns at being the leader,
but a sufficient explanation is simply that the lead goose gets

So here we have a pattern that has a significant consequence for
the participating individuals, yet is not itself a controlled
variable for any individual in the group. The pattern itself has
side-effects that are not immediately noticeable, but which are
important in the control processes of the participating control
systems. It happens that the only way the individuals can control
for least effort is to fly as part of a certain pattern;
deviations from that pattern are corrected not because the
pattern itself is being sensed and controlled, but because a
side-effect of that pattern is being sensed and controlled.

The evolutionary explanation might pertain to something like the
behavior of the bower bird in building its nest. I doubt that the
elaborateness of the nest confers any immediate advantage on any
individual bower bird -- considering the labor involved, the
opposite could well be true. But the male bower bird has
inherited an irresistable urge to perceive a certain sort of
configuration in its environment, and the female has inherited a
tendency to get the hots when such a configuration is seen. With
this combination of inherited control systems with inherited
reference levels, the male bower bird is doomed to building fancy
nests throughout eternity, and the female to swooning when this
is done. The pattern makes no sense in itself, but has come to be
an evolutionary selection force.

The fourth possibility is exhibited in any kind of social
behavior where explicit cooperation is required for any
individual to achieve a goal. The parties involved must explictly
perceive the pattern made by the group as a whole, and control
for maintaining that pattern. A marching band is a simple
example. A conversation is another.

When individuals all perceive the same pattern or a functionally
equivalent one, and each adjusts subsidiary goals in the interest
of maintaining the overall pattern intact, we then have the
nearest approach to a social control system. It is still created
by individual control systems, but now the individuals share
equivalent goals at the pattern level, and each person will act
against disturbances of the whole pattern. If the marcher ahead
drops out, the one behind will move up to fill the gap,
specifically because the pattern is not supposed to have any gaps
in it and not just because the marcher behind is faster.

No one would argue, from this perspective, that the global
organization is a controlled variable. In this sense, there is
consistency with HPCT. They do argue that the global
organization is functional and is the central phenomenon to be
explained. At least some of the comments I have received on
the net suggest that many in HPCT would object to that focus.
They appear to believe that the organization at that level is
irrelevant. Correct me if I have misunderstood.

Perhaps you will agree that not ALL global patterns are
functional in the sense you mean. And perhaps you can see that
under HPCT there is room for patterns that are in fact functional
by one route or another, even when they are not explicitly sought
by any individual. Bruce Nevin and Martin Taylor have, I think,
presented a view like this concerning languagedd acquisition, and
most of us on the net have come to concur.

I agree that global organization is an important phenomenon to be
explained, although I think that there are other phenomena of
equal importance.



Bill P.