[Martin Taylor 2014.11.26.16.45]
I don't know whether this will help clarify "control of behaviour",
but my previous mention of protocols seems to have been forgotten,
or if not forgotten then ignored in the long “Understanding control
of behaviour: why it matters” thread. So here again is an
introduction to the idea of a protocol. I’ll use a person as the
example controller, though it could be any organism, or even a
computer.
Figure 1. Four ways Archie's output can affect Beth's control.
1. Archie might augment or diminish Beth's ability to influence the
CEV (e.g. help carry a heavy load, or sit on what Beth wants to
move).
2. Archie might improve or worsen Beth's ability to perceive the
CEV.
3. Archie's actions might disturb the CEV (this is the usual type of
interpersonal effect considered on CSGnet, and it will be the main
one considered in what follows).
4, Archie might diminish or augment the effect of some other
disturbance on Beth’s CEV.
Of course, although these influences are shown in Figure 1 as
side-effects, they might equally well be intended effects that
influence a perception Archie is controlling. The influences are the
same in either case.
Beth's actions have, of course, the same four possibilities for
affecting Archie’s control of whatever he happens to be controlling.
Figure 2. A side-effect loop
Figure 2 shows one possibility, in which Archie's actions in
controlling some perception influence Beth’s CEV, and Beth’s actions
in countering that disturbance influence Archie’s CEV and thus the
perception Archie is controlling. These influences form a loop of
which neither Archie nor Beth is aware (or rather, if they are aware
of it, they are using perceptions not shown in this diagram).
Although it is most probable that the loop gain is positive, and
would make control by both of them more difficult to an
exponentially increasing degree, it is also possible that the loop
gain is negative, making both better able to control. A side-effect
loop would be more likely to have negative loop gain if the
side-effects involved interactions of type 4, but bear with me for a
while.
Figure 3. Another way of showing the same side-effect loop (Archie
has morphed into Alan).
Figure 3 looks more complicated than Figure 2, but the side-effect
loop is the same. In Figure 3, the main effects of Alan’s and Beths
actions are shown as influencing two quite separate parts of the
external environment (their two CEVs). The corresponding input is
shown in the box for each partner. The side-effects of Figure 2 are
shown dashed in Figure 3, which emphasises the main effects, to
distinguish it from…
Figure 4. A protocol loop (The protagonists have become Ivan and
Cora).
In Figure 4, Ivan's action to correct against a disturbance
deliberately influences a perception Cora controls, and Cora’s
action deliberately influences some perception Ivan controls.
Neither of them acts directly on the environmental variable
corresponding to their disturbed perception, either because they
cannot and the partner can, or because it is more convenient to have
the partner do it.
Figure 4 illustrates a "protocol loop" that has been developed
during what I call co-reorganization in Cora and Ivan. A protocol
requires that each partner controls for displaying something that
indicates a willingness to participate in the execution of the
protocol. Cora’s smile, for example, might indicate to Ivan that
Cora controls for perceiving Ivan to be content, so if Ivan displays
(perhaps by saying “Would you mid closing the door”) that he is
controlling for perceiving the door to be closed, that would disturb
Cora’s controlled perception, which could be brought closer to its
reference value if she behaved by closing the door, and that action
woul simultaneously bring Ivan’s controlled perception of the door
state to its reference value.
The development of a protocol requires a language of some kind, so
that each partner can display to the other something of what
perceptions they are controlling, and what actions of the other
would disturb a controlled perception in a way that would produce a
reasonably consistent behaviour. By using the protocol, Ivan
controls Cora’s behaviour in a way that Cora has signalled to Ivan
that she will allow.
It is, of course, possible for Ivan to control some aspect of Cora's
behaviour without using a protocol loop, if Ivan can guess a
perception Cora is controlling, and what Cora will do if he disturbs
it. This cartoon illustrates that possibility (I have asked for
permission to reproduce it).
Protocols, however, are a good part of the glue holds a society
together. Violation of the expectations of a protocol can lead to
confusion and conflict, as we often see when people of different
cultural backgrounds interact.
Martin